Structural Health Monitoring and Failure Analysis of Large-Scale Hydro-Steel Structures, Based on Multi-Sensor Information Fusion

Abstract

Large-scale hydro-steel structures (LS-HSSs) are vital to hydraulic engineering, supporting critical functions such as water resource management, flood control, power generation, and navigation. However, due to prolonged exposure to severe environmental conditions and complex operational loads, these structures progressively degrade, posing increased risks over time. The absence of effective structural health monitoring (SHM) systems exacerbates these risks, as undetected damage and wear can compromise safety. This paper presents an advanced SHM framework designed to enhance the real-time monitoring and safety evaluation of LS-HSSs. The framework integrates the finite element method (FEM), multi-sensor data fusion, and Internet of Things (IoT) technologies into a closed-loop system for real-time perception, analysis, decision-making, and optimization. The system was deployed and validated at the Luhun Reservoir spillway, where it demonstrated stable and reliable performance for real-time anomaly detection and decision-making. Monitoring results over time were consistent, with stress values remaining below allowable thresholds and meeting safety standards. Specifically, stress monitoring during radial gate operations (with a current water level of 1.4 m) indicated that the dynamic stress values induced by flow vibrations at various points increased by approximately 2 MPa, with no significant impact loads. Moreover, the vibration amplitude during gate operation was below 0.03 mm, confirming the absence of critical structural damage and deformation. These results underscore the SHM system’s capacity to enhance operational safety and maintenance efficiency, highlighting its potential for broader application across water conservancy infrastructure.

1. Introduction

Large-scale hydro-steel structures (LS-HSSs), which are integral to the technological units of water reservoirs and power plants, play a pivotal role in hydraulic engineering by effectively regulating the inflow or outflow of water in supply or discharge channels [1]. LS-HSSs primarily encompass a wide range of hydraulic gates (such as floodgates and ship locks), ship lifts, and various types of hoists, along with auxiliary equipment like lifting rods and locking devices [1,2]. They are indispensable in hydraulic engineering, serving crucial functions in water resource management, flood control, power generation, and navigation facilitation [3,4]. These structures are fundamental to ensuring the safety and functionality of dams and are designed to withstand significant environmental stresses throughout their extended service lives that often span several decades [5].

However, continuous exposure to harsh environmental conditions, including fluctuating water levels, sediment loads, fluid-induced vibration, corrosion, and varying pressures, progressively compromises the structural integrity of LS-HSSs over time [6,7]. The gradual degradation of these structures poses substantial risks, especially as they age, making them increasingly susceptible to failure [8,9]. The failure of LS-HSSs can lead to catastrophic outcomes, including uncontrolled water release, severe flooding, damage to downstream infrastructure, and loss of life [10,11]. Historical accidents highlight the severe consequences of such failures. For instance, the destruction of the Wachi Dam gate in 1968 led to severe damage, highlighting critical design issues and operational vulnerabilities [9]. The event underscored the importance of robust structural integrity and maintenance practices for dam gates to prevent catastrophic failures. In addition, the Folsom Dam spillway gate failure in 1995 was a significant event in which a radial gate collapsed, causing uncontrolled water release [12]. This failure was attributed to metal fatigue and inadequate maintenance practices, highlighting the importance of regular inspection and timely repairs. More recently, Hartford et al. [13] reviewed multiple flow control failures in dam safety, focusing on mechanical and operational issues that led to gate malfunctions. Their study identified systemic weaknesses in dam gate operations that can lead to catastrophic outcomes if not addressed in a timely manner. These incidents underscore the pressing need for reliable, continuous monitoring systems that can detect early signs of structural distress and prevent disasters before they occur.

A structural health monitoring system (SHMS) offers the real-time monitoring of structural conditions, enabling the early detection of potential failures and the implementation of preventive measures [14]. By providing continuous data on structural performance, an SHMS facilitates timely maintenance and reduces the likelihood of unexpected failures, thereby protecting human lives and economic assets [15]. Furthermore, integrating the SHMS with comprehensive safety evaluations is particularly crucial, as it allows engineers to assess the integrity of these structures more accurately and ensures that appropriate interventions are made before minor issues escalate into significant hazards [16]. Given the critical role of LS-HSSs in ensuring the safety and efficiency of hydraulic engineering, an SHMS has now become increasingly vital [17]. Han et al. [18] proposed a hydraulic vibration monitoring system for sluice gates in long-distance water conservancy projects, emphasizing the importance of real-time monitoring to detect abnormal gate vibrations early. Their study provides critical insights into vibration control techniques in high-risk sluice gates. Zhang et al. [19,20] developed indirect monitoring methods for ice load assessment on steel gates in cold regions, addressing the uncertainty inherent in environmental factors. Their hybrid reconstruction method effectively evaluates loads that are otherwise difficult to measure directly, showcasing the benefits of indirect structural health monitoring (SHM) in extreme conditions. Zhang et al. [21] presented a novel approach to monitoring sluice gate vibrations using monocular digital photography combined with a robotic measurement system. Their method highlights the potential of visual monitoring techniques in vibration analysis and their contribution to non-invasive SHM methods. Eick et al. [22,23] explored vision-based, non-contact monitoring methods to evaluate the tension in miter gate diagonals. These studies demonstrate the application of advanced imaging technologies for monitoring partially submerged gates, offering new possibilities for SHM in complex environments where traditional sensors may not be practical. The studies mentioned above have contributed to expanding the SHMS field regarding LS-HSSs. Building upon conventional contact-based sensing and monitoring techniques, they highlight the emerging trend of integrating advanced non-contact monitoring technologies with indirect assessment methods to enhance operational safety and efficiency in challenging environments. However, the information thus collected on operation and management exhibits many facets, variations, and intricacies. Consequently, it is impractical for decision-makers to rely solely on a single indicator to formulate precise and efficacious structural health evaluations or maintenance decisions [24,25,26]. These studies may fail to fully comprehend the intricate interplay between the various stressors affecting LS-HSSs when using a single index method. Consequently, this can lead to incomplete or inaccurate evaluations of structural integrity. This limitation underscores the necessity for more comprehensive SHMS approaches that enable a holistic perspective of structural safety for LS-HSSs.

Despite the evident advantages of SHMSs, their application in the water conservancy sector remains underdeveloped, compared to other industries such as civil engineering [15,27,28,29], aerospace [30], marine structures [31], ship structures [32], and energy transportation (oil and gas pipelines, etc.) [33]. In these fields, SHMSs have successfully monitored various structures, providing significant value in terms of extending their service lives and ensuring operational safety. However, in the context of LS-HSSs, several shortcomings have been identified in current SHMS practices. (i) Existing SHMSs for hydraulic engineering projects typically focus on single-sensor technology and its applications (strain, vibration, etc.), enabling the detection of abnormal equipment operation and timely alarm triggering. However, due to the SHMS’s reliance on a single monitoring parameter, it lacks the systematic analysis capability required to address the associated issues caused by this specific indicator. For instance, the vibration monitoring method is widely employed for online monitoring [18], primarily focusing on measuring vibrations during operation. However, it does not encompass the detection of gate deformation, tilting conditions, and changes in opening–closing status caused by these vibrations. (ii) Moreover, with an increased demand for on-site monitoring, the lack of data fusion and integrated analysis methods restricts the ability to capture the intricate interactions between different indicators and their cumulative effect on structural integrity [16]. Although there have been relevant theoretical and methodological studies on safety assessment [34,35] or health evaluation [36,37] for LS-HSSs, the existing research has not considered integration with the SHMS, providing limited insights into making overall structural safety evaluation timely and accurate. Consequently, further research should focus on developing comprehensive analytic strategies that integrate multiple sensor data sources, enabling more precise and holistic assessments of structural health and safety. (iii) In addition, research on SHMSs within the water conservancy sector has predominantly focused on specific types of structures, such as the miter gates [22,23,38,39]. In contrast, other critical structures, mainly radial gates [40], have received less attention. In recent years, with their increasingly widespread use in reservoirs and hydropower stations, radial gates have played a pivotal role in numerous hydraulic engineering projects. Meanwhile, radial gates are subject to complex loading conditions, including dynamic water pressures and fluctuating environmental stresses, making them particularly vulnerable to failure [41]. Furthermore, due to the diverse configurations of gate structures and variations in operational conditions, radial gates exhibit distinct characteristics and disparities in terms of their structural attributes and failure modes [9,26,42]. The failure of a radial gate could have severe consequences, such as the uncontrolled release of water and subsequent flooding, which could endanger lives and cause extensive property damage. Despite their considerable significance, there is a shortage of literature on monitoring radial gates. This gap in the research also highlights the need for dedicated study and the application of SHMSs to radial gates.

Therefore, in intelligent water conservancy construction, it is crucial to research the technology architecture of multi-sensor information fusion for SHMS, based on technologies such as sensing technology and the Internet of Things (IoT). This entails proposing corresponding implementation pathways and methodologies to enhance state perception, anomaly analysis, and maintenance decision-making capabilities during the operation and maintenance processes of LS-HSSs. Such efforts will significantly facilitate the digitalization, networking, and intelligent transformation of conventional equipment in hydraulic engineering. To address these gaps, the main research objectives of this paper are as follows:

Firstly, this paper proposes an enhanced SHM framework that integrates finite element analysis, multi-sensor information fusion, the Internet of Things (IoT), and comprehensive safety evaluation into structural health monitoring and operational management to build a closed-loop system of perception, analysis, decision, and optimization.

Then, we construct an SHMS that integrates real-time monitoring, abnormal early warning, and the dynamic evaluation of structural safety.

Finally, through on-site deployment in the spillway of the Luhun Reservoir, the effectiveness of the system in terms of real-time anomaly detection and decision support is verified. This method aims to improve the safety, reliability, and operation efficiency of LS-HSSs in the water conservancy field. In addition, by extending SHMSs to previously understudied structures such as sluice gates, this study will contribute to the broader field of SHMSs and provide valuable insights for future research and practice. The main contributions of this study are as follows:

(i) Innovative SHM framework. An advanced framework that integrates finite element analysis, multi-sensor information fusion, and IoT, creating a closed-loop system for enhanced decision-making.

(ii) New SHMS. A novel SHMS has been developed, enabling real-time monitoring and dynamic safety evaluation.

(iii) Practical validation. The SHMS was successfully validated at the Luhun Reservoir spillway, showcasing effective real-time anomaly detection and maintenance support.

(iv) Digital transformation. This research advances the digitalization of hydraulic engineering, addressing gaps in SHM for LS-HSSs like floodgates and offering valuable insights for future applications.

The remainder of this paper is structured as follows. Section 2 describes the analytical methods and theories. Section 3 introduces the framework of the SHMS. Modeling, simulation analysis, and sensor deployment are also presented. Section 4 provides the application results through a case study on a radial gate. Meanwhile, the effectiveness and limitations of the proposed approach are discussed. Finally, Section 5 concludes the paper by providing a comprehensive summary of the essential findings and offering insightful suggestions for future research directions.

2. Methodology

Continuous monitoring and observation of the structural health status of LS-HSSs are necessary to ensure their safety and obtain real-time information on their condition. This will help to mitigate the potential hazards associated with complex hydrological and operational conditions. Utilizing finite element analysis technology enables efficient time and cost savings in research while facilitating computer simulations to determine the LS-HSSs’ natural frequency and vibration mode and their primary structural stress and deformation distribution. These simulation results serve as a theoretical foundation for selecting and arranging SHMS sensors and analyzing the operational status of LS-HSSs.

2.1. Fluid-Structure Coupling Theory

There is a typical fluid–structure interaction between the LS-HSS and the water body, where dynamic water pressure is generated near the LS-HSS when water flows. This is achieved through the interaction between the water body and the LS-HSS. When conducting fluid–structure coupling analysis, it is necessary to follow the laws of mass conservation and momentum conservation to describe the behavior of the water body, and we express these laws as conservation equations [43]:

(1)${\displaystyle \frac{\partial {\rho }_f}{\partial t}}+\nabla \times \left({\rho }_f\right)=0$

(2)${\displaystyle \frac{\partial {\rho }_f\mathit{v}}{\partial t}}+\nabla \times \left({\rho }_f\mathit{v}\mathit{v}-{\mathit{\tau }}_f\right)={\mathit{f}}_f$

where ${ \rho }_f$ represents fluid density, $\mathit{v}$ represents fluid velocity, $t$ represents time, ${\mathit{\tau }}_f$ represents shear stress tensor, and ${\mathit{f}}_f$ represents the volume force vector. Without considering the energy equation, the conservation of the solid part can be directly derived from Newton’s second law:

(3)${\rho }_s \ddot{{\mathit{d}}_{\mathit{s}}}=\nabla \times {\mathit{\delta }}_{\mathit{s}}+{\mathit{f}}_{\mathit{s}}$

where ${ \rho }_s$ represents the density of the solid; ${\mathit{\delta }}_{\mathit{s}}$ represents the stress tensor, which describes the distribution of stress within the solid; ${\mathit{f}}_{\mathit{s}}$ represents the volume force vector acting on the solid; $\ddot{{\mathit{d}}_{\mathit{s}}}$ represents the acceleration vector of the solid, reflecting its state of motion. The boundary conditions for fluid motion are expressed as follows [43]:

(4)$\left\{\begin{array}{c}{\displaystyle \frac{\partial f}{\partial x}}{u}_x+{\displaystyle \frac{\partial f}{\partial y}}{u}_y+{\displaystyle \frac{\partial f}{\partial z}}{u}_z=0\\ f\left(x,y,z\right)=0\end{array}\right.$

where $x,y, z$ represent the water flow direction, the width direction of the spillway, and the vertical upward direction, respectively. The finite element equation of the fluid–solid coupled system can be expressed as follows [43]:

(5)$\left[\begin{array}{cc}{\mathbf{A}}_{ff}& {\mathbf{A}}_{fs}\\ {\mathbf{A}}_{sf}& {\mathbf{A}}_{ss}\end{array}\right]\left[\begin{array}{c}∆{\mathit{X}}_{f,k}\\ ∆{\mathit{X}}_{s,k}\end{array}\right]=\left[\begin{array}{c}{\mathit{B}}_f\\ {\mathit{B}}_s\end{array}\right]$

where ${\mathbf{A}}_{ff, }{\mathbf{A}}_{fs,}{\mathbf{A}}_{sf, }{\mathbf{A}}_{ss}$ represent the stiffness matrices of the different parts, corresponding to the coupling relationship between the fluid and solid; $∆{\mathit{X}}_{f,k}$ represents the displacement increment of the fluid part; $∆{\mathit{X}}_{s,k}$ represents the displacement increment of the solid part; ${\mathit{B}}_f$ and ${\mathit{B}}_s$ represent force vectors corresponding to the forces in the fluid and solid parts.

2.2. Principle of Vibration Analysis

Vibration is a prevalent phenomenon in LS-HSSs, resulting in varying degrees of oscillation during operation. Among these characteristics, the vibration frequency and mode hold the utmost importance. Assuming that there are no water constraints and disregarding damping effects and dynamic loads, we derive the equation for free vibration as follows [44]:

(6)$\mathit{M}\ddot{\mathit{\delta }}+\mathit{K}\mathit{\delta }=0$

where $\mathit{M}$ and $\mathit{K}$ represent the mass matrix and stiffness matrix, respectively, while $\ddot{\mathit{\delta }}$ and $\mathit{\delta }$ denote the acceleration and displacement of the node. During this period, the system experiences harmonic motion, described as $\mathit{\delta }=\mathit{\phi } \mathrm{c}\mathrm{o}\mathrm{s}(\mathit{\omega }t+\theta )$, where $\mathit{\phi }$ represents the amplitude. By substituting this expression into Equation (6), we obtain $\left(\mathit{K}-{\mathit{\omega }}^2\mathit{M}\right)\mathit{\phi }=0$. In cases of free vibration, it is important to note that not all nodes within the structure exhibit an amplitude φ equal to zero. Hence:

(7)$\left|\mathit{K}-{\mathit{\omega }}^2\mathit{M}\right|=0$

Equation (7) represents the inherent frequencies of the LS-HSS, and, by determining the degree of freedom as n, we can calculate the nth-order frequency. Organizing them in ascending sequence provides us with a series of natural frequencies for the structure, denoted by $\mathit{\phi }$, to represent their respective structural modes.

In practical scenarios, the LS-HSS undergoes vibration due to the pulsating pressure exerted by water. When water is present, Equation (6) is modified to derive the fluid-structure vibration equation, considering their interconnected interaction:

(8)$\mathit{M}\ddot{\mathit{\delta }}+\mathit{K}\mathit{\delta }=-\mathit{P}$

where $\mathit{P}$ represents the dynamic water pressure at the interface between the fluid and the structure.

Considering the known boundary conditions and the fluid–structure coupling at the contact surface, it can be observed that the normal velocity of water nodes is equivalent to that of the gate leaf nodes, while the disturbance velocity and pressure in the far field water tend toward zero. These conditions allow us to establish a proportional relationship between node pressure and acceleration, expressed as $\mathit{P}={\mathit{M}}_{\mathit{p}}\ddot{\mathit{\delta }}$, where ${\mathit{M}}_{\mathit{p}}$ represents the additional mass matrix. Rearranging this equation leads to the formulation of a fluid–structure coupling vibration equation for LS-HSS.

(9)$\left(\mathit{M}+{\mathit{M}}_{\mathit{p}}\right)\ddot{\mathit{\delta }}+\mathit{K}\mathit{\delta }=0$

However, accurately determining the pulsating pressure $\mathit{P}$ caused by water in real-world engineering scenarios is challenging, due to the intricate hydrodynamic loads exerted on the LS-HSS. This study relies on the “Load Code for Harbour Engineering (JTS 144-1-2010)” to determine the standardized hydraulic force acting upon the structural components:

(10)$F_w=C_w{\displaystyle \frac{\rho }{2}}{v}^{2}A$

where $F_w$ represents the standard value of the water flow force (kN); $C_w$ is the water flow resistance coefficient; $\rho$ is the water density; $v$ is the design flow velocity of water; $A$ is the projected area of the component that is perpendicular to the direction of flow.

2.3. Safety Evaluation Method

2.3.1. Single-Indicator Early Warning Method

• (1). Stress evaluation and early warning

The stress values are extracted from the specific installation positions of on-site sensors. The online monitoring data are then compared with the finite element calculated values, allowable stress values, and material limit values to evaluate each measurement point’s current stress safety status. The evaluation criteria are presented in Equation (11) [45].

(11)$\left\{\begin{array}{c}Safe:\sigma \le {\sigma }_{fin}\\ Relatively safe:{\sigma }_{fin}\le \sigma <\left[\sigma \right]\\ Dangerous:\left[\sigma \right]\le \sigma <{\sigma }_s\\ Destructive:\sigma \ge {\sigma }_s\end{array}\right.$

The equation is defined as follows: $\sigma$ represents the stress value obtained from SHMS, ${\sigma }_{fin}$ represents the stress value calculated through finite element analysis when under full load, $\left[\sigma \right]$ represents the allowable stress intensity, and ${\sigma }_{s }$ represents the yield strength of the material.

• (2). Vibration evaluation and early warning

Following the requirements of NB/T10859-2021 (Technical Requirements for an Online Monitoring System of Metal Structure Equipment in Hydropower Projects) [45], the threshold values for system warnings and alarms are set as indicators of moderate and severe hazards (Table 1). If these thresholds are exceeded, an alert or alarm will be triggered. Specifically, a warning will be issued when the average displacement exceeds 0.25 mm, while an alarm will be activated if it exceeds 0.5 mm.

• (3). Slanting evaluation and early warning

The evaluation index for the gate tilt is determined based on the distance between the outer edge of the gate leaf and the groove [45]. The calculation method is shown in Equation (12). A warning or alarm will be triggered if the tilt exceeds the specified limit. When facing downstream, a left tilt is indicated by “+” while the correct tilt is indicated by “−”. The angle can be converted into an actual deviation amount.

(12)$d=20-\left({\displaystyle \frac{L_1}{2}}\times \cos \left(\left(\theta \right)-\left|\alpha \right|\right)\times {\displaystyle \frac{3.14}{180}}\right)-{\displaystyle \frac{L_2}{2}}$

Equation (12) [45] is defined as follows: 20 is the initial value of the margin (adjustable), L₁ is the diagonal length of the gate, L₂ is the width of the gate, θ is the angle between the diagonal of the gate and the x-axis, and $\left|\alpha \right|$ represents the absolute value of the inclination angle.

2.3.2. Comprehensive Evaluation Method

The SHMS software developed in this study is equipped with a four-level early warning system to enhance the user’s ability to assess the safety status of LS-HSSs. The workflow of the four-level early warning system is shown in Figure 1. The warning level could reflect the safety conditions of the LS-HSS. Based on the real-time monitoring value transmitted from the LS-HSS, the average values of the monitoring points are selected as control indexes. The color of these control indexes determines the warning level. A single monitoring point may contain multiple aspects of LS-HSS operation information, such as structure deformation, stress from the surrounding water, and risk prediction. Users can input the allowable maximum warning value into the software, according to the related risk management specification. Then, the values of the four warning levels can be preset according to the operation’s needs. Thus, the automatic early warning function of the monitoring items can be realized according to either the preset standard values or threshold values of the different measuring points and the risk level.

However, the data obtained from monitoring LS-HSSs during operation exhibit considerable ambiguity and uncertainty, and insufficient correlation processing may lead to false alarms. Data fusion technology can establish connections among multi-source monitoring data [46], forming a cohesive information system that can facilitate evaluative decision-making, which enables deterministic evaluation, thereby significantly reducing the uncertainty present in the safety assessment system for LS-HSSs and effectively improving the accuracy of safety evaluations. Therefore, based on multi-sensor data fusion, the comprehensive processing capability for multi-source uncertain information can be enhanced, thus facilitating the effective monitoring and diagnosis of LS-HSSs.

D-S evidence theory offers a practical approach to handling uncertainty and information conflict, allowing for the combination and reasoning of evidence to enable more reliable decisions and judgments, and is widely applied in fields such as multi-source data fusion, decision support systems, pattern recognition, and diagnostic systems [46,47]. To facilitate the users’ ability to judge the safety level of the LS-HSS, a safety evaluation method based on the D-S theory of evidence is designed inside the platform’s software. The process workflow is shown in Figure 2. The subsequent segments offer fundamental contextual knowledge regarding the D-S theory and application protocols employed in decision-making.

1. Frame of discernment: the discernment framework consists of a set of M hypotheses that are both mutually exhaustive and mutually exclusive, as follows [47]:

(13)$\Theta =\left\{S_1,S_2,\dots ,S_M \right\}$

The uncertainties within the D-S theory are expressed by assigning probabilities to the space $\Theta$, similar to how classical probability theory operates. However, there is a significant distinction: the D-S theory entails allocating a likelihood to every subset within $\Theta$. The definition of the power set $2^{\Theta }$ can be stated as follows [47]:

(14)$2^{\Theta }=\left\{\varphi ,\left\{{\theta }_1\right\},\left\{{\theta }_2\right\},\dots ,\Theta \right\}$

The empty set is denoted by $\varphi$. The power set $2^{\Theta }$ has the following characteristics:

The total count of its elements is equivalent to 2^M. Every possible combination within the set $2^{\Theta }$, except for a possible singleton of values, i.e., their union. For example, $\left\{{\theta }_1,{\theta }_2,{\theta }_3\right\}={\theta }_1\cup {\theta }_2{\cup \theta }_3$. The basic probability assignment (BPA) allocates the entire probability to the power set ${ 2}^{\Theta }$.

2. Mass function:

The mass function $M:2^{\Theta } \to \left[0, 1\right]$ demonstrates the following characteristics:

(15)$m\left(\varphi \right)=0$

and

(16)$\sum _{\forall \theta \in 2^{\mathsf{\Theta }}}m\left(\theta \right)=1$

3. D-S rule of combination: the evidence needed by the D-S theory is represented by information obtained from a network comprising numerous autonomous sensors. Data fusion utilizes normalization to streamline and condense the received data by emphasizing the consensus among various sources while disregarding contradictory evidence.

The combination rule for two pieces of evidence, $m_1$ and $m_2$, which is also referred to as joint mass, is defined by the D-S theory in the following manner:

(17)$m_{1,2}\left(A\right)={\displaystyle \frac{1}{1-K}}\sum _{B\cap C=A\ne \varnothing }{m}_1\left(B\right)·m_2(C)$

Equation (7) adheres to both the commutative and associative properties, while K denotes the level of disagreement between two evidence sources and can be derived using the following method:

(18)$K=\sum _{B\cap C=\varnothing }{m}_1\left(B\right)·m_2(C)$

In Equation (17), the normalization factor ${\displaystyle \frac{1}{1-K}}$ aids in aggregating information by disregarding contradictory evidence.

Based on multi-sensor information fusion, the operational safety status of the LS-HSS site can be effectively reflected. Initially, the real-time monitoring values from the on-site SHMS are chosen as indicators for the safety evaluation. Each sensor can indicate four safety levels of the equipment’s operation, using green, blue, orange, and red to denote very safe, relatively safe, somewhat unsafe, and unsafe (stabilized, basically stable, instability, and danger). An SHMS can encompass various dimensions of safety evaluation information, including vibration, stress, and posture. Users can input different safety-level thresholds (A, B, and C) for each monitoring item into the system, based on relevant risk management regulations, before the equipment is in operation. On the one hand, the system can identify the safety risk level based on the preset thresholds of a single sensor and can issue the corresponding warnings or alarms. On the other hand, based on the monitoring data from multiple sensors, the system calculates the basic probability assignment (BPA) values (such as m1, m2, m3, m4) for the operational states of each sensor using threshold segmentation methods, membership functions (such as Gaussian functions and triangular functions), statistical models, or machine learning models (such as support vector machines, decision trees, and neural networks). For similar types of sensors (e.g., Class A sensors containing multiple stress sensors), Dempster’s combination rule can be used to fuse the BPA values of each sensor, resulting in a comprehensive state BPA for that class of sensors. Subsequently, the comprehensive BPA values of various sensors are further fused, ultimately yielding an overall health status assessment result for the equipment and providing accurate guidance for operational safety, as illustrated in Figure 2.

Although D-S evidence theory has significant advantages when handling uncertainty and multi-sensor information fusion, its application also faces some challenges. As the number of sensors in the system increases, the computational complexity of the combination rules rises significantly, which may affect real-time processing capabilities. Furthermore, the fusion results may exhibit biases when data from the different sensors conflict. To address these issues, future research can introduce artificial intelligence analysis methods (such as support vector machines and isolation forest algorithms) to analyze and identify anomalous data in real time, reducing false alarm problems caused by sensor anomalies and enhancing the stability of SHMS operation. In contrast, based on the stable operational data and evaluation results of an SHMS, algorithms can be optimized by incorporating weighted D-S theory, integrating with prior probabilities, or utilizing adaptive fusion models based on neural networks. This can further improve the accuracy of evaluating the safety status during equipment operation, supporting more reliable maintenance decisions.

3. SHMS Design and Test on Radial Gates

3.1. Framework of the SHMS

The present study integrates finite element analysis, multi-sensor information fusion, the IoT, and automatic control systems into the health monitoring and operational management of an LS-HSS. Its objective is to monitor the operational characteristics in real time continuously, establish a visualized SHMS, and ensure reliable and safe operation throughout the entire lifecycle of the LS-HSS. Figure 3 illustrates the conceptual framework of the system.

The first step involves creating a detailed three-dimensional model and conducting a numerical analysis. The analysis results are then utilized to identify the critical areas of the structure, which serve as a guide for on-site sensor installation. This approach enhances the efficiency of monitoring unit deployment and ensures accurate data collection, providing a solid foundation for the SHMS design.

Secondly, the corresponding sensors are installed after identifying the hazardous structural points to monitor stress, strain, vibration, attitude, environment, and operational parameters. Data transmission is carried out using the Modbus/TCP protocol. The dynamically collected data undergo processing through filtering, noise reduction, averaging, regression analysis, and other post-processing techniques utilizing a dynamic data acquisition and integrated analysis system, before transmitting the processed monitoring data to the database.

Then, the operational status and safety reliability of the LS-HSS will be comprehensively assessed by establishing a structural safety evaluation index database and comprehensive evaluation scheme, based on the historical data and simulation results.

Finally, developing a comprehensive SHMS integrates functions such as real-time condition monitoring, anomaly detection, data analysis, and three-dimensional visualization to achieve seamless integration of models, data, and services, thereby enhancing the operational management capabilities of LS-HSS. The system architecture is illustrated in Figure 4.

3.2. Modeling and Simulation Analysis

This study is centered on the Luhun Reservoir and encompasses system development, deployment, and application testing. The Luhun Reservoir (as shown in Figure 5) is a significant water conservancy project located in the middle reaches of the Yishui River, a second-level tributary of the Yellow River, 67 km from Luoyang.

The target hydro-steel structures are installed in the spillway channel (Figure 5), located 500 m east of the dam. It comprises an inlet channel, a gate chamber with a transitional section, a spillway section, a flow diversion section, and a tailwater channel. This open-type overflow channel project, initiated in December 1959, reached a successful conclusion in August 1965. The gate chamber structure was designed in a “U-shaped” configuration, featuring three openings: east, central, and west. Each opening measures 12 × 10.5 m. A hydraulic lifting mechanism facilitates the operation of each curved steel gate, each weighing 89 tons. The elevation of the inlet weir crest stands at 313 m, while the remarkable maximum flood discharge capacity reaches 3810 m³/s. The detailed technical parameters of the arc-shaped working gate for the spillway are shown in Table 2.

Due to technological limitations and inadequate preparation in the early stages, the Luhun project was constructed simultaneously with exploration, design, and construction. Consequently, specific safety hazards exist during its operation. Therefore, in December 2003, the Luhun Reservoir Operation Center implemented risk prevention and reinforcement measures for critical projects such as spillways. The main initiatives included strengthening the dam body and west dam head, enhancing the lining of flood discharge tunnel walls, demolishing and reinforcing the flood discharge gate piers, fortifying the irrigation tunnel structure and protecting the tailwater channel with masonry, reinforcing the water conveyance tunnel structure while removing the outlet energy dissipation pool, and updating and renovating the mechanical-electrical equipment as well as the hydro-steel structures.

In January 2016, the Luhun Reservoir Operation Center commissioned the Nanjing Hydraulic Research Institute to organize experts for an on-site inspection and safety evaluation of the reservoir’s engineering and hydro-steel structures. According to the inspection results, the manufacturing and installation quality of the hydro-steel structures complies with the design and specification requirements. The equipment operation is reliable, ensuring safe operations. The overall evaluation of the dam’s hydro-steel structures’ safety is rated as Grade A (SL 258-2000: Guidelines on Dam Safety Evaluation). However, due to its long-term operation in complex environments, the target LS-HSS still faces unresolved challenges. Figure 6 illustrates some of the typical issues found during on-site inspections in June 2023.

This study employs the finite element method to conduct a structural and mechanical characteristics analysis of an LS-HSS under various hydrodynamic conditions. Based on this finite element analysis, the overall structural mechanical characteristics (stress, strain, modal, etc.) under different conditions can be investigated. According to the finite element analysis results, the risk value of structural response can also be determined. Under the condition that no limit is specified in the corresponding specification, using the finite element analysis result as the monitoring system’s early warning value can help to locate structural damage and carry out the corresponding structural repair in good time.

This research used the Ansys software to model and conduct numerical simulations of the spillway’s radial gate. A diagram of the spillway structure is shown in Figure 7.

Figure 8 shows the finite element model of LS-HSS, which consists of 32,150 shell elements, 1129 solid elements, and 66,997 nodes. The fluid part adopts the Fluent module to construct a finite element model of the liquid region. The fluid domain includes 16,320 solid elements and 18,970 nodes.

In terms of structural composition, an LS-HSS can be classified as a thin plate structure, due to the smaller dimensions of its panels, main beams, and support arms compared to its other directions. Consequently, shell elements are employed in the finite element model to simulate these structures accurately. The trunnion conforms with solid element characteristics and is modeled using solid elements. The material utilized for LS-HSSs is Q345B, with the following specific material properties: a density of 7850 kg/m³, an elastic modulus of 206 Gpa, and a Poisson’s ratio of 0.3.

When the radial gate is closed, the displacement on both sides is constrained, and the vertical degree of freedom of the gate is fixed at 0. A cylindrical support is located at the hinge to restrict all degrees of freedom except for tangential direction, with a standard Earth gravity of 9806.6 mm/s². When the gate is opened, a tensile force of 1500 KN is applied at both lifting lugs, while other constraints remain consistent with those when the gate is closed. The operation of the radial gate involves two distinct loads, namely, static water and dynamic water. The corresponding calculation conditions are as follows:

(1) Under static water conditions, the water levels in front of the gate are set at 0 m, 1.5 m, 3 m, 4.5 m (the limit of water level in flood season), and 6.5 m (the maximum utilization level in non-flood season). The gates remain closed in all these scenarios. The hydrostatic analysis module allows for directly adjusting the free surface position to accurately simulate the static water pressure exerted on the gate.

(2) Under dynamic water conditions, the radial gate’s water levels are set at 6.5 m (the maximum utilization level in non-flood season), and its opening heights are set at 0 m, 1 m, 2 m, 3 m, 4 m, 5 m, and 6.5 m. Fluent is utilized to simulate the flow field close to the gate and analyzes the gate’s stress and deformation under varying opening heights.

3.2.1. Force Analysis Under Static Water Conditions

(1) Equivalent stress analysis: By altering the position of the free surface, the impact of the water level in front of a radial gate on its equivalent stress during closure can be analyzed. Figure 9 shows the hydrostatic pressure distribution corresponding to different water levels.

Figure 10 depicts the equivalent stress distribution of the radial gate at a water depth of 6.5 m. Figure 11 shows the change in the maximum equivalent stress of the main structure as the water level changes from 0 to 6.5 m.

In static water conditions, different water levels result in the maximum equivalent stress on the radial gate occurring on its main beam, reaching a peak of 133.06 MPa. The panel experiences its highest equivalent stress, which is mainly concentrated in the middle section of the lower part, with a maximum value of 82.23 MPa. Additionally, the hinge support bears a maximum equivalent stress of 49.59 MPa. Therefore, when analyzing potential overload failure and arranging stress sensors for the gates, priority should be given to areas with higher stresses, such as the radial gate’s main beam and panel.

(2) Deformation analysis: Figure 12 shows the total deformation of the radial gate at water depths ranging from 0 to 6.5 m. In static water conditions, the maximum total deformation of the radial gate occurs on the main beam, with a maximum total deformation value of 7.11 mm. The maximum total deformation of the girder is 5.72 mm.

3.2.2. Force Analysis Under Dynamic Water Conditions

(1) Equivalent stress analysis: Considering the radial gate’s high stiffness, the strain induced by the flow field is relatively minimal. Therefore, only the fluid’s impact on the gate is considered when accounting for the dynamic loads caused by water flow. Therefore, this paper employs a single-phase fluid-structure coupling method to simulate the dynamic water load on the gate. Considering the current circumstances, we set the inlet velocity to 3 m/s and configure the outlet as a pressure boundary condition, with the pressure set to standard atmospheric level. The standard k-ω model is utilized as the turbulence model, while the VOF method tracks the free surface accurately. By establishing a coupling interface, we effectively apply the dynamic loads generated by water flow to the panel and subsequently import the resulting pressure into a static structural analysis of the radial gate. This enables us to simulate the stress deformation caused by dynamic water pressure accurately. When the gate is opened, the dynamic water pressure value is determined by measuring the average pressure on the coupling surface corresponding to the different gate openings once the water level in front of the gate stabilizes. Figure 13 illustrates the flow field and load conditions on the coupling surface of the radial gate after it has been opened.

Figure 14 depicts the stress distribution when the gate opening measures 1 m. Under this operational condition, the maximum equivalent stress on the gate is observed along the main beam, exhibiting an increasing trend as it approaches the bottom section.

Figure 15 shows the maximum equivalent stress of the radial gate at different opening heights ranging from 0 to 6.5 m. The maximum equivalent stresses in the gate panel, girder, and radial arm slightly decrease as the gate opening increases from 0 to 6.5 m. Similarly, a noticeable decreasing trend is observed in the maximum equivalent stresses of the main and secondary beams as the opening size increases. Therefore, when analyzing the overload failure and arranging the stress sensors, priority should be given to considering high-stress areas such as the main beam, secondary beam, and girder.

(2) Deformation analysis: Figure 16 shows the maximum total deformation at different opening heights ranging from 0 to 6.5 m. The gate opening height significantly impacts the maximum total deformation of the gate when it generally ranges between 0 and 4 m. However, no apparent trend is observed in the change of maximum total deformation at different parts of the radial gate when the gate opening exceeds 4 m. As the gate opening increases, there is a decrease in the maximum deformation observed in each part of the gate, which aligns with the variation pattern of the equivalent stress within the gate corresponding to its opening.

3.2.3. Modal Analysis Under Fluid Coupling Conditions

Considering the additional mass contribution of water, an analysis is conducted on the vibration characteristics of the radial gate. Table 3 presents the first six natural frequencies, with the gate’s primary natural frequency at approximately 1 Hz and the secondary to the sixth natural frequency at around 20 Hz. Figure 17 illustrates the modal vibration patterns of the radial gate, providing a visual representation of its displacement. The analysis results indicate that water coupling has a minimal impact on the first natural frequency of the gate. Generally, the first three to five modes are classified as low-order modes as they correspond to the fundamental vibrational characteristics of the structure. These modes are crucial for capturing the overall response of the structure to dynamic loads, including water pressure and flow-induced vibrations [9,49]. Modes beyond the first five are typically considered high-order modes. These modes represent more complex vibrations and are important for understanding local structural behavior and interactions under specific conditions [9,49]. In this paper, the low-order modes primarily exhibit overall gate vibrations with low self-resonant frequencies that closely align with the pulsation frequency of water flow. This renders them more susceptible to resonance and they require careful attention during operation. In addition, in the low-order modes, there is significant vibration of the leaf and noticeable vibration of the radial arm. In the high-order modes, the vibration of the leaf gradually diminishes. The vibration of the radial arm becomes less evident and is primarily manifested as vibrations in components such as the girder.

3.3. Sensor Deployment and Data Acquisition

Throughout the entire lifecycle of an LS-HSS, the main factors affecting its safe operation include structural strength, stiffness, vibration resistance, the performance of the hoist, and environmental influences. The central monitoring aspects of this study include the following:

(1) Structural stress. Stress–strain monitoring aims to obtain the stress curves of large structures or of the supporting components during operation. On the one hand, it can reveal the stress distribution pattern and identify stress concentration areas in critical structural elements, thus validating the design rationality. On the other hand, it enables the timely detection of operational abnormalities, facilitating damage diagnosis analysis and lifespan prediction. Working according to the finite element simulation analysis of the structural stress hot-spot distribution obtained from the previous context, considering factors such as sensor installation and data transmission, the layout points and actual installation of the stress–strain sensors are shown in Figure 18 and Figure 21a.

(2) Structural vibration. Valuable information on an LS-HSS is transmitted externally through vibrations, which serve as a crucial parameter in data statistics, damage identification, and condition assessment. Consequently, monitoring and analyzing these vibrations remain the focal point of ongoing research.

The process of vibration monitoring primarily involves collecting and analyzing the vibration data to determine the intensity levels. For LS-HSSs, routine operations such as water blocking, flow, or opening and closing typically exhibit a low vibration intensity. The increase in vibration intensity is often attributed to factors such as equipment damage, wear, cracks, or improper installation. Monitoring changes in vibration intensity makes it possible to analyze and evaluate the safety conditions of the LS-HSS. The sensors should be positioned close to areas with higher vibration amplitudes or mechanical rotating components, ensuring that monitoring is not compromised. Working according to the finite element simulation analysis, the arrangement of vibration measurement points can be observed in Figure 19 and Figure 21b.

(3) Operational posture. Valuable information regarding changes in the LS-HSS’s operating posture during operation can be obtained by measuring the inclination of the tested plane relative to the horizontal plane. During regular water blocking, overflow, or the opening and closing operations of a gate, its inclination should remain within the designated allowable range according to the design specifications. However, if equipment synchronization problems or unexpected situations arise, deviations from the normal working range may occur in its operating posture. Monitoring changes in the equipment’s operating posture enables the analysis and evaluation of a metal structure’s operational status. When installing sensors, affixing them to rigid equipment structures is crucial. The layout points for attitude measurement are illustrated in Figure 20 and Figure 21c.

(4) Environmental operating parameters. Real-time monitoring of temperature, humidity, and wind speed should be conducted at key operational locations, assessing their safety impacts on equipment and personnel (Figure 21d).

(5) Operating parameters of the hoist. Real-time monitoring of parameters should be conducted during the operation, including opening/closing force, motor current, system oil pressure, stroke, etc., to assess the working performance and status of the gate mechanism and conduct safety evaluations (Figure 22).

(6) Hydrodynamics of the gate slot. Monitoring the flow velocity, pressure distribution, and water level in front of the gate should be conducted to assess the scouring degree of the gate slot and gate leaf during the discharge process, along with safety evaluations (by accessing the relevant hydrological information from the dam safety monitoring system [50]).

The central monitoring items and sensors selected for the LS-HSS are shown in Figure 21 and Figure 22, including: DH1205K resistance-based surface strain gauges with a standard measurement range of ±1000 µε, a sensitivity of 700 µε, and operation within a temperature range of −20 to +80 °C; CA-DR-3005 capacitive acceleration sensors, with a standard measurement range of ± 5 g, a sensitivity of 400 mv/g, and operation within a temperature range of −40 to +120 °C; ACA626T high-precision dual-axis digital output inclinometers, with a standard measurement range of ± 30°, a sensitivity of 0.001°, and operation within a temperature range of −40 to +85 °C; RS-FSXCS-N01-1 integrated ultrasonic weather stations, with a standard measurement range as follows: wind speed ranging from 0 to 60 m/s, wind direction spanning from 0 to 359°, humidity varying between 0% RH and 99%RH, and temperature encompassing −40 °C to +12 °C. Furthermore, a multi-channel dynamic signal testing and analysis system (DH5922D) and DHDAS dynamic signal acquisition and analysis software are employed to collect data during the gate’s operation. Additionally, access is gained to the PLC controller of the gate’s hydraulic system to gather information regarding the operational and control states of the hydraulic equipment. The collected data are then integrated and transmitted to the remote monitoring center via I/O modules and intelligent gateways, using the Modbus/TCP protocol for further processing. The network architecture diagram of the SHMS is shown in Figure 23.

4. Application Results and Discussion

4.1. SHMS Functional Integration and Analysis

Field application tests were conducted through the on-site deployment of the SHMS for the LS-HSS. As depicted in Figure 24, the system facilitates the real-time 2D/3D dynamic visual interaction of equipment operating status. It enables the automatic identification and judgment of abnormal operation processes while supporting a simulation analysis based on virtual scenes to enhance the efficiency and accuracy of on-site operation and maintenance work. In the event of abnormal conditions such as stress, vibration, or posture during the opening and closing process, the system will promptly update its real-time status information to alert users of any potential faults that may arise. This ensures smooth and consistent equipment operation during water resource scheduling by adjusting the control parameters and implementing the necessary maintenance measures.

The detailed integration process of the SHMS for LS-HSSs is shown in Figure 25. It seamlessly integrates into the entire service timeline of the equipment, encompassing functions such as data acquisition, exception analysis, safety evaluation, and maintenance decision-making. The data are an essential component required for the smooth functioning of the SHMS, which can be divided into two main categories: the initial data obtained during the product design phase (such as design indexes, process parameters, CAD models, and CAE models) and derived data associated with the service process. The derived data encompass condition-monitoring information (e.g., vibration, stress, deformation, and temperature), maintenance records (including equipment archives, repair schedules, details of any replaced parts, and repair types), service-related data (such as crack identification records, wear diagnosis findings, leakage maintenance information, and corrosion prediction results), and knowledge-based data (consisting of rules, experience standards, and mechanisms) as well as fusion data that include information-analytical model insights, machine learning datasets, etc.

4.2. Experiment Results and Analysis

Taking the reservoir spillway radial gate as an example, this chapter aims to demonstrate the operating process and validate the feasibility and effectiveness of the SHMS under different operating conditions. Simultaneously, through an analysis of the system monitoring data, a comprehensive evaluation of the gate’s safety status is conducted, and recommendations for operational maintenance are provided.

Firstly, based on the system’s historical operational data, we analyze the application status of the system functions and confirm any abnormal conditions during the device’s historical operation. Secondly, we conduct opening and closing tests (with a current water level of 1.4 m and the gate opening set at 1 m), analyze the monitoring data, and evaluate the overall safety status of the radial gate. The test conditions on-site are depicted in Figure 26, while the operational status of the hoist and control system is illustrated in Figure 27.

4.2.1. Static Online Monitoring: Water-Blocking Condition

The monitoring results of static stress under normal water-blocking conditions are presented in Figure 28a. The static stress results demonstrate relatively consistent values over time, and the static stress at each measurement point corresponds with the simulated analysis results across various operating conditions. All values remain below the material’s allowable stress, satisfying the strength requirements. No alarms or anomalies were detected during system operation. Furthermore, the system can conduct a multidimensional comparative analysis of historical data from each monitoring point. The mean stress values and their corresponding statistical analysis results are illustrated in Figure 28b. However, the reservoir has not reached its average water storage capacity, due to the arid climate. As a result, it was not feasible to conduct static stress tests under typical water storage conditions during the experiment. Subsequently, based on system operations, further enhancements will be made to gather and analyze multi-operational data to ensure equipment safety.

4.2.2. Dynamic Online Monitoring: Opening and Closing Conditions

To ensure the safety of the downstream river channel, long-term, large-scale flood discharge could not be conducted during the experiment, leading to the choice of small-scale, short-duration tests. The vibration monitoring data and analysis during the equipment operation are illustrated in Figure 29, indicating that the equipment functioned normally without alarms or anomalies. The study of the standard vibration data examined whether there were abnormal vibrations in the gate, caused by factors such as the form of the bottom edge, erosion in the gate groove, an improper support structure, and operational parameters during the small-scale discharge. The results indicated that the maximum amplitude during the test was around 0.03 mm; it was located at the edge beam’s lifting lug and spanned the width of the faceplate, corresponding to a gate opening of 1.0 m. According to the vibration displacement standards of the Arkansas River in the U.S. and NB/T10859-2021 (Technical Requirements for an Online Monitoring System of Metal Structure Equipment in Hydropower Projects) [45], this vibration level is essentially negligible (less than 0.05 mm) and can be disregarded without any harmful effects.

Figure 30 shows the stress monitoring results under dynamic water pressure when the radial gate is opened to 1.0 m. During the discharge process, the dynamic stress values at different measurement points that are induced by flow vibrations are approximately 2 MPa, showing no significant impact load. Compared to the static stress monitoring results, during the opening and closing process of the gate, as the gate opens, the water level corresponding to the gate panel decreases, and the equivalent stress decreases with increasing opening degree. Meanwhile, the equivalent stress in other parts of the gate increases with the opening degree, mainly due to the backward movement of the center of gravity during the opening process, which leads to a gradual concentration of stress at the support arm and in other locations. Overall, the dynamic stress results exhibit relatively consistent values over time, and the static stress at each measurement point aligns with the corresponding simulated analysis results under various operating conditions. All values remain below the material’s allowable stress, satisfying the strength requirements. No alarms or anomalies were observed during system operation.

4.2.3. Comprehensive Safety Evaluation and Discussion

The system ran smoothly and accumulated operational data for the LS-HSS through systematic on-site deployment, functional debugging, and application. It played a crucial role in ensuring regular, stable, and reliable operation while enhancing its digital operation and control capability. Based on the issues identified during the LS-HSS monitoring process, the monitoring results exhibit relatively consistent values over time and remain below the allowable value, satisfying the system requirements, as shown in Table 4.

Based on the monitoring results obtained during the experimental period, the following recommendations are proposed to ensure the continued safe, stable, and reliable operation of the LS-HSS:

(i) Regular inspections and non-destructive testing should be conducted to promptly identify and mitigate local corrosion, as demonstrated in Figure 6a. Additionally, periodic strength assessments based on the SHMS are essential to guarantee the structural integrity of the affected components. By implementing these measures, potential weaknesses can be addressed early on, preventing further degradation.

(ii) Regular inspections of the water seal system are paramount to swiftly detect and replace leaking seals, as demonstrated in Figure 6c. Following replacement, continued monitoring is necessary to ensure the water seal’s ongoing effectiveness. In cases where water ingress has compromised the structural integrity of the surrounding elements, additional strength assessments based on the SHMS may be warranted. Swift intervention is crucial to prevent further damage.

(iii) To maintain operational stability, the control system should incorporate a tilt correction feature, as demonstrated in Figure 6b. During gate operations, manual inspections should verify the feature’s functionality. If the SHMS detects tilting, immediate corrective action must be taken to prevent potential damage and ensure smooth operation.

(iv) Operating procedures should discourage stopping the gate at small opening angles to minimize resonance and stress on the gate. Engineers and operators should undergo training emphasizing the importance of avoiding these conditions. Regular safety analyses based on the SHMS should inform any necessary adjustments to operating procedures, ensuring the gate’s long-term reliability and safety.

In summary, the value of this study lies in its comprehensive approach to addressing the challenges of SHM for LS-HSS. Previous studies have primarily focused on numerical simulation, prototype testing, or the online tracking of single variables, as indicated in Table 5. This has allowed researchers to explore the operational mechanisms and identify failure modes and abnormal behavior in the structures during operation. These studies have contributed significantly to understanding the failure mechanisms and identifying anomalies in LS-HSSs, which are critical for ensuring their safe and reliable operation. Building on this foundation, the present study introduces a novel framework and implementation method for SHMS that integrates IoT technology, modeling and simulation analysis, and safety evaluation methods. By combining real-time data collection with advanced simulation models and safety evaluation techniques, this research offers a more comprehensive and integrated solution for the safety and reliability assessment of LS-HSSs. The proposed framework supports the continuous monitoring of structural health. It enhances the user’s ability to make informed decisions regarding maintenance and risk management, complementing and supporting the existing body of research. However, several limitations to this study must be acknowledged:

(i) This study did not achieve full-scale data collection across all operational conditions, due to safety management requirements at the field sites. Further efforts should be made to align data collection with the safety management requirements of reservoir operations, particularly during high-water-level storage in the rainy season and flood discharge scenarios during the flood season. This will involve accumulating monitoring data on equipment operation under these conditions, considering factors such as maintenance and degradation during the equipment’s lifecycle. Such data will be crucial for the dynamic optimization of early warning rules and for advancing research in fault prediction and risk forecasting.

(ii) The issue of false alarms caused by sensor anomalies remains challenging [51,52]. To address this issue, future research should focus on developing specific algorithms that can identify and mitigate false positives in the sensor data. This would include refining the existing SHM algorithms to better distinguish between actual structural issues and sensor errors, ensuring that false alarms do not compromise the system’s reliability.

(iii) A significant limitation is the scarcity of publicly available fault data and case studies, which restricts the ability to validate and improve the SHMS under various failure scenarios [53]. Future research should explore the construction of coupled multi-physics simulation models and develop degradation models to overcome this issue. This includes building surrogate models that act as virtual sensors, enabling the simulation of failure scenarios without real-world data. These virtual sensors can generate synthetic data that can be used for multi-source data fusion, uncertainty information reasoning, and decision-making in safety evaluations. The effectiveness of safety assessment methods based on virtual sensor data will also need to be evaluated, providing a pathway for advancing the SHMS without comprehensive real-world failure data.

5. Conclusions

This study presents a comprehensive SHM approach that is tailored explicitly for LS-HSSs. The proposed framework seeks to address several identified gaps in the current SHMS by integrating multiple monitoring indicators, offering a more holistic assessment of structural health. Therefore, the SHMS provided in this paper has demonstrated outstanding performance in field operations and comprehensive utilization, and it is pivotal in ensuring the regular, stable, and reliable operation of equipment. In addition, it significantly enhances the digital operation and control capabilities of LS-HSSs in reservoirs, marking a substantial advancement in the digital transformation of the water conservancy sector. The main advantages are summarized below.

(i) The SHMS provided in this paper enhances the monitoring of LS-HSSs through the fusion of IoT technology and multi-sensor data, enabling improved anomaly detection, early warning, and maintenance decision-making processes. This integration supports a closed-loop management system that significantly contributes to the safety, reliability, and operational efficiency of LS-HSSs in the water conservancy sector.

(ii) Additionally, by extending the application of SHM to previously under-researched structures like radial gates, this study contributes to the broader field of SHMSs. It provides valuable insights for future research and practical applications in similar environments.

(iii) Moreover, the SHMS’s development, on-site deployment, and testing were successfully conducted on the radial gate of the Luhun Reservoir spillway. The results indicated that static stress levels remained below allowable limits, and dynamic monitoring revealed negligible vibration behavior, underscoring the gate’s operational safety. Comprehensive analysis of the monitoring data facilitated a robust safety evaluation, leading to targeted maintenance recommendations. Meanwhile, the system enables real-time 2D/3D dynamic visual interaction with the operating status of an LS-HSS. It automatically detects and evaluates abnormal operational processes while supporting simulation analysis through virtual scenes. This capability enhances the efficiency and accuracy of on-site operation and maintenance tasks.

Building on the findings and advancements of this study, several areas require further research and development to enhance the efficacy and reliability of SHMSs for LS-HSSs:

(i) Optimization of sensor deployment. To improve data collection efficiency and reduce costs, future research should focus on developing advanced sensor placement strategies. This includes multi-objective optimization algorithms and AI-driven techniques that cater to complex scenarios, ensuring high-precision monitoring with optimized resource use.

(ii) Addressing false alarms. The challenge of false alarms due to sensor anomalies remains significant. Future work should refine the SHM algorithms to better distinguish between structural issues and sensor errors. Developing robust methods to mitigate false positives will be critical to maintaining the system’s reliability and effectiveness.

(iii) Integration of digital twin technology. The limited availability of fault data and case studies restricts SHMS validation. The development of digital twins for LS-HSSs represents a promising direction for improving dynamic safety assessments and maintenance decision-making. Research should focus on virtual-real fusion simulations, complex scenario modeling, and real-time model corrections to enhance the accuracy and utility of these digital twins. For instance, by building surrogate models as virtual sensors, researchers can simulate failure scenarios and generate synthetic data. This facilitates multi-source data fusion and reliable safety evaluations without requiring comprehensive real-world data.

Author Contributions

Conceptualization, H.Z. and R.Z.; methodology, H.L.; software, H.L. and Y.S.; validation, H.L., Y.S. and S.Z.; formal analysis, H.L. and R.Z.; investigation, H.L.; resources, H.L. and R.Z.; data curation, Y.S.; writing—original draft preparation, H.L.; writing—review and editing, H.L. and R.Z.; visualization, H.L. and S.Z.; supervision, H.Z. and R.Z.; project administration, R.Z.; funding acquisition, H.Z. and R.Z. All authors have read and agreed to the published version of the manuscript.

Data Availability Statement

Data are contained within the article.

Acknowledgments

The authors would like to express their sincere appreciation to the experts for their provision of insightful information and constructive comments throughout our research.

Conflicts of Interest

The authors declare no conflicts of interest.

Footnotes

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Figures and Tables

Figure 1. Four-level early warning method.

Figure 2. Safety evaluation method.

Figure 3. A conceptual framework of the structural health monitoring system.

Figure 4. The architecture of an IoT-enabled structural health monitoring system.

Figure 5. The overall layout of the project.

Figure 6. Some existing defects of the target LS-HSS: (a) localized corrosion; (b) gate opening deviation and tilt; (c) water seal leakage.

Figure 7. Diagrams of the spillway structure [48].

Figure 8. The finite element model of the LS-HSS.

Figure 9. Hydrostatic pressure distribution corresponding to different water levels.

Figure 10. The equivalent stresses at a water depth of 6.5 m.

Figure 11. Maximum equivalent stresses at water depths from 0 to 6.5 m.

Figure 12. Maximum total deformation at water depths from 0 to 6.5 m.

Figure 13. The flow field and load on the coupling surface in dynamic water conditions.

Figure 14. Equivalent stress distribution (opening height: 1 m).

Figure 15. Maximum equivalent stress at opening heights from 0 to 6.5 m.

Figure 16. Maximum total deformation at opening heights from 0 to 6.5 m.

Figure 17. Mode of vibration (partial).

Figure 18. Arrangements of the strain-measuring points of the LS-HSS: (a) radial arm (left) arrangement of measuring points; (b) panel, beam, and girder arrangement of measuring points; (c) radial arm (right) arrangement of measuring points.

Figure 18. Arrangements of the strain-measuring points of the LS-HSS: (a) radial arm (left) arrangement of measuring points; (b) panel, beam, and girder arrangement of measuring points; (c) radial arm (right) arrangement of measuring points.

Figure 19. Arrangements of vibration measuring points of LS-HSS: (a) radial arm (left) arrangement of measuring points; (b) panel, beam, and girder arrangement of measuring points; (c) radial arm (right) arrangement of the measuring points.

Figure 19. Arrangements of vibration measuring points of LS-HSS: (a) radial arm (left) arrangement of measuring points; (b) panel, beam, and girder arrangement of measuring points; (c) radial arm (right) arrangement of the measuring points.

Figure 20. Arrangements of the tilt-measuring points of the LS-HSS.

Figure 21. Sensors selected and installed on the LS-HSS: (a) dual-axis digital inclinometer; (b) capacitive accelerometer; (c) resistive strain gauges; (d) temperature and humidity anemometer.

Figure 22. Sensors selected and installed on the hoist: (a) hydraulic system; (b) hydraulic cylinder; (c) automatic control system.

Figure 22. Sensors selected and installed on the hoist: (a) hydraulic system; (b) hydraulic cylinder; (c) automatic control system.

Figure 23. The network architecture of the structural health monitoring system.

Figure 24. Integrated interface for system functions.

Figure 25. Data flow diagram of the integrated system functions.

Figure 26. On-site testing conditions of the LS-HSS.

Figure 27. The operational status of the hoist and control system.

Figure 28. Analysis of online monitoring data under water-blocking conditions. (a) real-time monitoring data; (b) statistical analysis data.

Figure 29. Analysis of online monitoring data under opening and closing conditions.

Figure 30. Dynamic stress monitoring results.

Table 1

Discrimination criteria for dividing vibration hazards, based on the average displacement of vibrating components.

Average Displacement (mm)	Degree of Vibration Hazard
0–0.05	Negligible
0.05–0.25	Minor
0.25–0.5	Moderate
>0.5	Severe

Table 2

Parameters of the radial gates of the spillway shown in Figure 5.

Technical Indicators	Parameters
Dimensions of the orifice (width × height)	12 × 10.50 m
Dimensions of the radial gate (width × height)	12 × 10.50 m (vertical height)
Weight of the radial gate	89,012 kg
Structural form of the radial gate	Exterior overfall
Maximum water level (static water condition)	323 m
Maximum water level (dynamic water condition)	323 m
Material type	Q345B (Equivalent: S355JR)
Lifting distance	10.9 m
Number of radial gates	3
Hinge form	Self-lubricating spherical plain bearings
Gate operation conditions	Gate operation in dynamic water
Maximum flood discharge	3810 m3/s
The total length of the spillway	435 m
Maximum opening and closing force of the hoist	2 × 1250 KN
Opening and closing stroke	6.5 m

Table 3

The first six-order natural frequencies of the radial gate.

Modal Order	Natural Frequencies
	0 m	3 m	6.5 m
1	1.12	1.14	1.14
2	15.22	9.73	9.23
3	15.74	13.65	13.37
4	20.69	14.92	15.21
5	21.17	15.78	15.32
6	22.61	16.17	15.42

Table 4

Safety evaluation results.

Items	Itemized Evaluation
Stress condition	Normal	Blue warning	Yellow warning	Red alert
Vibration condition	Normal	Blue warning	Yellow warning	Red alert
Tilt condition	Normal	Blue warning	Yellow warning	Red alert
Hoist condition	Normal	Blue warning	Yellow warning	Red alert

Table 5

Research into and applications of SHMS technology for LS-HSS.

Research Methods		Key Technologies	Application Objects
Contact measurement method	[18]	Triaxial accelerometer use, based on MEMS technology and the vibration monitoring method	Sluice gate
Modeling and simulation fusion method	[40]	Numerical simulation, based on building information modeling	Radial gate
IoT and Numerical simulation fusion method	[20]	Strain monitoring method and load distribution analysis	Radial gate
	[5]	Combination of numerical modeling and experimental measurements	Plain gate
Non-contact measurement method	[22]	Vision-based monitoring system of post-tensioned diagonals	Miter gate
	[23]	System for the non-contact monitoring of tension	Miter gate
Condition assessment method based on SHM	[39]	Structural health monitoring for damage prognosis	Miter gate
	[38]	Risk-based inspection strategies based on structural health monitoring	Miter gate

References

1. Li, H.; Zhang, R.; Zheng, S.; Shen, Y.; Fu, C.; Zhao, H. Digital twin-driven intelligent operation and maintenance platform for large-scale hydro-steel structures. Adv. Eng. Inform.; 2024; 62, 102661. [DOI: https://dx.doi.org/10.1016/j.aei.2024.102661]

2. Daniel, R.; Paulus, T. Chapter 3—Structural Types of Hydraulic Gates. Lock Gates and Other Closures in Hydraulic Projects; Daniel, R.; Paulus, T. Butterworth-Heinemann: Oxford, UK, 2019; pp. 35-282.

3. Liu, Y.; Zhang, X.; Yu, H.; Sun, Y.; Sun, C.; Li, Z.; Li, X. Hydraulic model of partial dam break based on sluice gate flow. Ocean Eng.; 2024; 295, 116974. [DOI: https://dx.doi.org/10.1016/j.oceaneng.2024.116974]

4. Yilmaz, A.; Kocaman, S.; Demirci, M. Numerical modeling of the dam-break wave impact on elastic sluice gate: A new benchmark case for hydroelasticity problems. Ocean Eng.; 2021; 231, 108870. [DOI: https://dx.doi.org/10.1016/j.oceaneng.2021.108870]

5. Pastor, M.; Bocko, J.; Lengvarsky, P.; Sivak, P.; Sarga, P. Experimental and Numerical Analysis of 60-Year-Old Sluice Gate Affected by Long-Term Operation. Materials; 2020; 13, 5201. [DOI: https://dx.doi.org/10.3390/ma13225201]

6. Ma, C.; Sheng, C.; Lian, J.; Ma, B.; Liu, F. Failure analysis of a leaf gate jammed in closing process. Eng. Fail. Anal.; 2020; 110, 104391. [DOI: https://dx.doi.org/10.1016/j.engfailanal.2020.104391]

7. Niu, L.; Yan, J.; Gao, D. Research on Corrosion and Defects of Hydraulic Metal Structures. Proceedings of the 6th International Conference on Environmental Science and Civil Engineering (ESCE); Nanchang, China, 4–5 January 2020.

8. Liu, J.; Wang, Z.; Fang, X.; Fang, H. Dynamic Instability Mechanism and Vibration Control of Radial Gate Arms. Proceedings of the International Conference on Intelligent Structure and Vibration Control (ISVC 2011); Chongqing, China, 14–16 January 2011; pp. 309-313.

9. Xu, C.; Liu, J.; Zhao, C.; Liu, F.; Wang, Z. Dynamic failures of water controlling radial gates of hydro-power plants: Advancements and future perspectives. Eng. Fail. Anal.; 2023; 148, 107168. [DOI: https://dx.doi.org/10.1016/j.engfailanal.2023.107168]

10. Xu, C.; Wang, Z.Z. New insights into dynamic instability regions of spillway radial gate owing to fluid-induced parametric oscillation. Nonlinear Dyn.; 2023; 111, pp. 4053-4070. [DOI: https://dx.doi.org/10.1007/s11071-022-08040-y]

11. Zhang, J.; Huang, Y.; Tang, J.; Zhou, F.; Kang, Y.; Feng, B. Inspection of Cracks in the Piston Rod of a Hydraulic Cylinder Using Injected Alternating Current-Field Measurement. Water; 2022; 14, 2736. [DOI: https://dx.doi.org/10.3390/w14172736]

12. Todd, R.V. Spillway Tainter Gate Failure at Folsom Dam, California. Waterpower ’99; American Society of Civil Engineers: Reston, VA, USA, 2012; pp. 1-10.

13. Hartford, D.N.; Baecher, G.B.; Zielinski, P.A.; Patev, R.C.; Ascila, R.; Rytters, K. Operational Safety of Dams and Reservoirs; Institution of Civil Engineering Publishing: London, UK, 2016.

14. Farrar, C.R.; Worden, K. An introduction to structural health monitoring. Philos. Trans. R. Soc. A—Math. Phys. Eng. Sci.; 2007; 365, pp. 303-315. [DOI: https://dx.doi.org/10.1098/rsta.2006.1928]

15. Bertulessi, M.; Bignami, D.F.; Boschini, I.; Brunero, M.; Ferrario, M.; Menduni, G.; Morosi, J.; Paganone, E.J.; Zambrini, F. Monitoring Strategic Hydraulic Infrastructures by Brillouin Distributed Fiber Optic Sensors. Water; 2022; 14, 188. [DOI: https://dx.doi.org/10.3390/w14020188]

16. Ye, Z.; Ye, Y.; Zhang, C.; Zhang, Z.; Li, W.; Wang, X.; Wang, L.; Wang, L. A digital twin approach for tunnel construction safety early warning and management. Comput. Ind.; 2023; 144, 103783. [DOI: https://dx.doi.org/10.1016/j.compind.2022.103783]

17. Negi, P.; Kromanis, R.; Doree, A.G.; Wijnberg, K.M. Structural health monitoring of inland navigation structures and ports: A review on developments and challenges. Struct. Health Monit.—Int. J.; 2024; 23, pp. 605-645. [DOI: https://dx.doi.org/10.1177/14759217231170742]

18. Han, P.; Xue, H.; He, Q. Monitoring Technique and System of Hydraulic Vibration of Sluice Gate in Long Distance Water Conservancy Project. Proceedings of the International Conference on Advanced in Control Engineering and Information Science (CEIS); Dali, China, 18–19 August 2011.

19. Zhang, M.; Qiu, B.; Kalhori, H.; Qu, X. Hybrid reconstruction method for indirect monitoring of an ice load of a steel gate in a cold region. Cold Reg. Sci. Technol.; 2019; 162, pp. 19-34. [DOI: https://dx.doi.org/10.1016/j.coldregions.2019.03.020]

20. Zhang, M.; Qiu, B.; Wei, Q.; Qu, X.; Shi, D. Indirect ice load monitoring and strength analysis of a steel gate considering uncertainties. Measurement; 2019; 148, 106919. [DOI: https://dx.doi.org/10.1016/j.measurement.2019.106919]

21. Zhang, G.; Yu, C.; Guo, G.; Li, L.; Zhao, Y.; Li, H.; Gong, Y. Monitoring Sluice Health in Vibration by Monocular Digital Photography and a Measurement Robot. Ksce J. Civ. Eng.; 2019; 23, pp. 2666-2678. [DOI: https://dx.doi.org/10.1007/s12205-019-0869-5]

22. Eick, B.A.; Narazaki, Y.; Smith, M.D.; Spencer, B.F., Jr. Vision-Based Monitoring of Post-Tensioned Diagonals on Miter Lock Gate. J. Struct. Eng.; 2020; 146, 04020209. [DOI: https://dx.doi.org/10.1061/(ASCE)ST.1943-541X.0002777]

23. Eick, B.A.; Smith, M.D.; Spencer, B.F. Non-contact monitoring of the tension in partially submerged, miter-gate diagonals. Smart Struct. Syst.; 2021; 27, pp. 193-207.

24. Zhang, R.; Tian, Y.; Li, H. Application of improved radar chart in the health evaluation model of hydraulic gate. Mech. Ind.; 2022; 23, 24. [DOI: https://dx.doi.org/10.1051/meca/2022014]

25. Faridmehr, I.; Farokhi Nejad, A.; Baghban, M.H.; Ghorbani, R. Numerical and Physical Analysis on the Response of a Dam’s Radial Gate to Extreme Loading Performance. Water; 2020; 12, 2425. [DOI: https://dx.doi.org/10.3390/w12092425]

26. Lin, J.; Wang, R.; Zhao, Y.; Sun, Y. Research on the Safety Assessment of Corbel Beams of Reservoir Radial Gates Based on Real States. Appl. Sci.; 2023; 13, 7578. [DOI: https://dx.doi.org/10.3390/app13137578]

27. Torres, B.; Poveda, P.; Ivorra, S.; Estevan, L. Long-term static and dynamic monitoring to failure scenarios assessment in steel truss railway bridges: A case study. Eng. Fail. Anal.; 2023; 152, 107435. [DOI: https://dx.doi.org/10.1016/j.engfailanal.2023.107435]

28. Cao, W.; Wen, Z.; Feng, Y.; Zhang, S.; Su, H. A Multi-Point Joint Prediction Model for High-Arch Dam Deformation Considering Spatial and Temporal Correlation. Water; 2024; 16, 1388. [DOI: https://dx.doi.org/10.3390/w16101388]

29. Basu, B.; Bursi, O.S.; Casciati, F.; Casciati, S.; Del Grosso, A.E.; Domaneschi, M.; Faravelli, L.; Holnicki-Szulc, J.; Irschik, H.; Krommer, M. et al. A European Association for the Control of Structures joint perspective. Recent studies in civil structural control across Europe. Struct. Control Health Monit.; 2014; 21, pp. 1414-1436. [DOI: https://dx.doi.org/10.1002/stc.1652]

30. Bergmayr, T.; Winklberger, M.; Kralovec, C.; Schagerl, M. Structural health monitoring of aerospace sandwich structures via strain measurements along zero-strain trajectories. Eng. Fail. Anal.; 2021; 126, 105454. [DOI: https://dx.doi.org/10.1016/j.engfailanal.2021.105454]

31. Katsoudas, A.S.; Silionis, N.E.; Anyfantis, K.N. Structural health monitoring for corrosion induced thickness loss in marine plates subjected to random loads. Ocean Eng.; 2023; 273, 114037. [DOI: https://dx.doi.org/10.1016/j.oceaneng.2023.114037]

32. Sabatino, S.; Frangopol, D.M. Decision making framework for optimal SHM planning of ship structures considering availability and utility. Ocean Eng.; 2017; 135, pp. 194-206. [DOI: https://dx.doi.org/10.1016/j.oceaneng.2017.02.030]

33. Wang, Y.; Zhu, S.; Wang, B.; Qin, J.; Qin, G. Structural health monitoring of oil and gas pipelines: Developments, applications and future directions. Ocean Eng.; 2024; 308, 118293. [DOI: https://dx.doi.org/10.1016/j.oceaneng.2024.118293]

34. Shi, Y.J.; Shao, Y.Y.; Zhang, Z.M.; Jiang, Y.; Qi, L. Safety Evaluation of Hydraulic Steel Gate Based on TOPSIS Model. J. Phys. Conf. Ser.; 2021; 2029, 012100. [DOI: https://dx.doi.org/10.1088/1742-6596/2029/1/012100]

35. Guo, J.; Wen, Y.; Xiao, J. Safety evaluation research of hydraulic steel gate based on BP-neural network. Proceedings of the International Conference on Sustainable Power Generation and Supply; Nanjing, China, 6–7 April 2009; pp. 1-4.

36. Yang, G.; Gu, C.; Huang, Y.; Yang, K. BP neural network integration model research for hydraulic metal structure health diagnosing. Int. J. Comput. Intell. Syst.; 2014; 7, pp. 1148-1158. [DOI: https://dx.doi.org/10.1080/18756891.2014.966999]

37. Yang, G.; Jia, W. Research to health diagnose model of gate and hoist machinery based on AHP. Proceedings of the International Conference on Advanced Engineering Materials and Technology; Sanya, China, 29–31 July 2011; pp. 3036-3042.

38. Van dang, T.; Rigo, P. Risk-Based Inspection Strategies of Miter Gates Based on Structural Health Monitoring. Fatigue and Fracture of Materials and Structures; Springer: Cham, Switzerland, 2022; pp. 135-140.

39. Vega, M.A.; Hu, Z.; Fillmore, T.B.; Smith, M.D.; Todd, M.D. A Novel Framework for Integration of Abstracted Inspection Data and Structural Health Monitoring for Damage Prognosis of Miter Gates. Reliab. Eng. Syst. Saf.; 2021; 211, 107561. [DOI: https://dx.doi.org/10.1016/j.ress.2021.107561]

40. Sun, S.; Zhang, R.; Liu, X.; Liu, C.; Wang, A. Numerical Simulation and Application of Radial Steel Gate Structure Based on Building Information Modeling under Different Opening Degrees. Water; 2024; 16, 636. [DOI: https://dx.doi.org/10.3390/w16050636]

41. Xu, C.; Wang, Z.; Zhang, H.; Li, H.; Li, D. Investigation on mode-coupling parametric vibrations and instability of spillway radial gates under hydrodynamic excitation. Appl. Math. Model.; 2022; 106, pp. 715-741. [DOI: https://dx.doi.org/10.1016/j.apm.2022.02.013]

42. Jiliang, L.; Yunhe, L.; Zhengzhong, W. Calculation models and analysis methods of the parametric vibration of radial gates. IOP Conf. Ser. Earth Environ. Sci.; 2019; 304, 022078. [DOI: https://dx.doi.org/10.1088/1755-1315/304/2/022078]

43. Chapter 1 Introduction to fluid-structure coupling. Modelling of Mechanical Systems; Axisa, F.; Antunes, J. Butterworth-Heinemann: Oxford, UK, 2007; Volume 3, pp. 1-44.

44. Singh, D.K.; Pal, P.; Duggal, S.K. Free Vibration Analysis of Lock Gate Structure. J. Mech.; 2020; 36, pp. 507-520. [DOI: https://dx.doi.org/10.1017/jmech.2020.10]

45. NB/T 10859-2021 Technical Conditions of Online Monitoring System for Metal Structural Equipment in Hydropower Projects; China Energy Press: Beijing, China, 2021.

46. Hassani, S.; Dackermann, U.; Mousavi, M.; Li, J. A systematic review of data fusion techniques for optimized structural health monitoring. Inf. Fusion; 2024; 103, 102136. [DOI: https://dx.doi.org/10.1016/j.inffus.2023.102136]

47. Zhao, L.; Sun, H. Motor Fault Diagnosis Based on D-S Evidence Theory Information Fusion Algorithm. Proceedings of the International Conference on Integrated Circuits and Communication Systems; Raichur, India, 23–24 February 2024; pp. 1-7.

48. Zhang, C.; Zhang, Y.; Zhao, H.; Wang, M.; Wang, T. Numerical Investigation of Erosion Wear Characteristics of Hydraulic Spillway. Appl. Sci.; 2021; 11, 8118. [DOI: https://dx.doi.org/10.3390/app11178118]

49. Géradin, M.; Rixen, D.J. Mechanical Vibrations: Theory and Application to Structural Dynamics; John Wiley & Sons: Hoboken, NJ, USA, 2015.

50. Han, Z.; Li, Y.; Zhao, Z.; Zhang, B. An Online safety monitoring system of hydropower station based on expert system. Energy Rep.; 2022; 8, pp. 1552-1567. [DOI: https://dx.doi.org/10.1016/j.egyr.2022.02.040]

51. Giordano, P.F.; Quqa, S.; Limongelli, M.P. The value of monitoring a structural health monitoring system. Struct. Saf.; 2023; 100, 102280. [DOI: https://dx.doi.org/10.1016/j.strusafe.2022.102280]

52. Zhao, H.; Ding, Y.; Meng, L.; Qin, Z.; Yang, F.; Li, A. Bayesian Multiple Linear Regression and New Modeling Paradigm for Structural Deflection Robust to Data Time Lag and Abnormal Signal. IEEE Sens. J.; 2023; 23, pp. 19635-19647. [DOI: https://dx.doi.org/10.1109/JSEN.2023.3294912]

53. Li, C.; Li, S.; Feng, Y.; Gryllias, K.; Gu, F.; Pecht, M. Small data challenges for intelligent prognostics and health management: A review. Artif. Intell. Rev.; 2024; 57, 214. [DOI: https://dx.doi.org/10.1007/s10462-024-10820-4]