Content area
Kenya’s increasingly evolving power grid requires advanced data stream mining techniques for more effective real-time monitoring and control. This paper examines how these techniques are currently being used in Kenya’s smart power networks, highlighting both their advantages and disadvantages. Although the current systems support basic monitoring and fault detection, they are unable to effectively handle high data velocity, variability, and the integration of renewable energy. High prices, weak regulatory support, limited technical skills, poor data quality, and inadequate computational infrastructure are some of the main obstacles. This paper identifies machine learning, adaptive clustering, and edge computing as possible solutions for improving fault detection, dynamic load balancing, real-time monitoring, and the integration of renewable energy. These strategies optimize grid performance while addressing latency and scalability challenges. The study recommends implementing advanced data stream mining techniques, making significant infrastructural investments, boosting technical capacity, and creating supportive legislative frameworks in order to improve Kenya’s smart power grid. Implementing these changes may help Kenya to enhance its grid resilience and reliability, boost energy efficiency, and promote long-term sustainability.
Clinical trial registration: Not applicable.
Introduction
Smart power grids have evolved as a game-changing option for modern energy management, providing real-time monitoring and control capabilities that boost efficiency, reliability, and sustainability. These grids use the Internet of Things (IoT) and advanced analytics to improve electricity generation, transmission, and consumption [1]. The application of data stream mining techniques is crucial to the grids’ functionality, allowing for the continuous processing and analysis of enormous and dynamic data streams created by grid sensors and devices [2]. However, there may be challenges that need to be progressively addressed.
The adoption of smart power grids in Kenya is gaining traction, informed by the country’s rising energy consumption, the need to reduce wastage, and integrate renewable energy sources for improved system efficiencies [3]. Yet, the ability to rapidly process and analyze real-time data streams is critical to the successful adoption of these systems. Despite considerable progress, there are still hurdles in realizing the full potential of data stream mining techniques within the context of Kenya’s unique socioeconomic and infrastructural development [4].
The current data stream mining techniques used in Kenya’s smart power grid systems are primarily focused on basic monitoring and fault detection, thus often failing to effectively account for the huge volume, velocity, and fluctuation of data streams [5]. These shortcomings can limit the data mining techniques’ potential to give meaningful insights for action in real time. Furthermore, Kenya’s diverse energy needs, sporadic renewable energy sources, and infrastructural advancement levels can pose significant challenges to the application of these strategies. Besides the technical limitations, other key challenges entail deficient technological infrastructure, inadequate expertise in advanced data analytics, as well as high implementation costs [5].
Furthermore, regulatory, institutional, and logistical constraints impede the integration of new techniques into existing grid systems, aggravating the challenges of real-time monitoring and control [6]. Hua et al. [7] further noted that despite emerging technologies, such as artificial intelligence, machine learning, and edge computing which provide great potentials for improving accuracy, scalability, and speed of real-world data processing, the challenges are far from over. There is therefore need to further explore how to improve the efficiency and effectiveness of data stream mining techniques in Kenya’s smart power networks. Kenya’s smart power grid systems could significantly increase their performance and sustainability if the existing challenges are continuously addressed within the context of new and emerging technologies.
This study sought to analyze the current status of data stream mining techniques, identify challenges hindering their application, and explore opportunities for improving them. The findings may give practical insights for policymakers, utility companies, and technology providers, thus enabling the transition to more efficient and robust smart power grid systems in Kenya. Secondary data was used in the study, accessed through various reputable sources. These included Kenya’s governmental and regulatory bodies, such as Energy and Petroleum Regulatory Authority (EPRA), Kenya, Ministry of Energy and Petroleum, Kenya, and Kenya National Bureau of Statistics (KNBS), among others.
Other data sources included utility companies, such as Kenya Power and Lighting Company (KPLC), and KenGen; as well as international organizations like the World Bank, and International Energy Agency (IEA). Secondary data search websites were also used, including Google scholar, IEEE Xplore, ScienceDirect, ACM Digital Library, and Web of Science. For effective searching, a combination of relevant keywords was applied, which included “data stream mining”, “smart-grid”, “power grid systems”, and “real-time monitoring”. Filters such as publication date, type of publication, and authenticity of journal were some of the criteria used to identify high quality references for this study. Generally, relevance and quality of the referenced sources were paramount, and were properly adhered to.
Literature review
Smart power grid data stream mining techniques and their roles in Kenya
Data stream mining techniques are critical for obtaining useful insights from the continuous flow of data provided by smart power grids. These techniques enable the real-time monitoring, control, and optimization of grid activities as they facilitate generation of meaningful insights from enormous, continuous data streams produced by sensors, smart meters, and IoT devices [8]. However, their implementation has considerable strengths, shortcomings, and obstacles that require a critical analysis [9]. Figure 1 presents a Smart Grid Architectural Model (SGAM) with interconnected systems and devices for delivering power to the consumers.
Fig. 1 [Images not available. See PDF.]
Smart grid architecture model domains (blue lines: Secure bi-directional communication flow. red dashed lines: Electric power flow).
Source: Gaspar et al. [10])
Real-time insights
Data stream mining techniques enable the near-instant detection of grid anomalies, such as power outages or voltage changes, allowing for quick responses. According to Ahmad et al. [11], real-time insights are critical for preserving grid stability and avoiding cascading failures. Further advantages revolve around continuous data processing, windowing techniques, online algorithms, and incremental learning. Unlike traditional data mining, which operates on static datasets, data stream mining processes data as it flows in. This enables real-time pattern analysis and identification. To deal with the infinite nature of data streams, techniques such as sliding or tumbling windows are used, narrowing the analysis to a specific portion of the stream and delivering up-to-date information (fig. 2). Tantalaki et al. [12] observed that online algorithms are designed to process data in real time using limited memory and computational resources. Algorithms are configured to incrementally update their models with each new data point, adjusting to data variations and presenting the most up-to-date view.
Fig. 2 [Images not available. See PDF.]
Outline of energy management system.
Source: Vinothine et al. [13]
Trend identification, real-time anomaly detection, real-time monitoring, and dynamic decision-making are some of the key types of real-time insights provided by data stream mining techniques. They detect anomalous patterns or outliers as they occur, allowing for fraud detection, network security issues, and the identification of equipment malfunctions. Identifying developing trends and patterns in real time allows for prompt responses and proactive decision-making. Jiang et al. [14] further established that regularly monitoring important metrics and performance indicators ensures smooth operations and the detection of possible concerns. Data mining techniques give up-to-date information to help real-time decision-making in a variety of applications.
Real-time insights are useful in many different ways, including enhanced efficiency, proactive risk management, timely intervention, personalized experiences, and improved decision-making processes [4]. They allow for quick action to prevent problems, optimize processes, and capitalize on opportunities. According to Al-Mashhadani and Kurnaz [15], leveraging power line communication (PLC) techniques facilitates real-time communication between grid components and consumers. By detecting grid faults, smart grids are able to optimize energy transmission, and effectively manage demand responses. However, Wicaksono et al. [16] noted that the end-to-end implementation of demand response systems aimed at manufacturing power users remains a challenge due to multiple stakeholders and subsystems that generate a diverse and substantial amount of data. The reinforcement-learning-based approach can therefore offer general and scalable solutions for dynamic and intricate production environments.
Capabilities for prediction
Data mining techniques provide tremendous predictive capabilities for smart power grids, allowing for proactive management, higher efficiency, and increased reliability [11]. Machine learning-based data stream mining techniques enable grid operators to forecast future occurrences such as demand surges or equipment failures. These projections allow for better resource allocation and maintenance planning, enhancing grid reliability.
Load forecasting: through load forecasting, future electricity demands can be predicted accurately, enhancing grid stability and effective resource allocation. Data mining approaches can forecast short- and long-term electricity consumption by analyzing historical load data, weather patterns, economic indicators, and other factors. Techniques such as time regression models, support vector machines, time series analysis, and neural networks can be applied for this purpose.
Grid stability prediction: Data mining can examine real-time grid data to detect patterns that could lead to instability, such as voltage swings or cascading failures. This enables proactive interventions to ensure grid stability. Anomaly detection, pattern recognition, and time series analysis are frequently utilized in this analysis.
Renewable energy forecasting: the intermittent nature of solar and wind power can pose challenges for grid management. However, data mining can examine weather data, historical generation data, and other parameters to estimate the output of renewable energy sources, allowing for improved system integration [11, 17, 18]. This prediction can be done using weather forecasting, and machine learning models, as well as time series analysis.
Fault prediction and analysis: data mining can be applicable in predicting equipment failures through synthesis of sensor data. By collecting data from transformers, transmission lines, and other grid components, data mining can detect trends that suggest possible breakdowns. This enables proactive maintenance and prevents costly outages. Furthermore, when a defect occurs, data mining can evaluate event logs and sensor data to quickly identify the underlying cause, allowing for timely and faster service restoration [17]. This can be accomplished by classification algorithms, anomaly detection, and rule-based systems.
Customer behaviour predictions: Predicting consumer power usage trends can greatly aid in resolving any potential challenges in the energy consumption ecosystem Data mining can identify patterns in consumer energy consumption by examining smart meter data, and allowing for more targeted energy efficiency initiatives and personalized services. The analysis can be done using clustering algorithms, classification algorithms, and association rule mining approaches [18].
Latest advancements in Kenya include deep learning models, such as long short-term memory (LSTM) which is particularly useful in location-based services, where this can be used to improve forecasting accuracy for demand and fault prediction. Additionally, there is reinforcement learning which optimizes energy distribution by learning from past grid behaviour; and digital twins meant for stimulating grid conditions to predict failures and recommend preventive measures [19] (Fig. 3).
Fig. 3 [Images not available. See PDF.]
Long short-term memory (LSTM) neural network.
Source: Asif et al. [20]
Nonetheless, according to Asif et al. [20], hybrid strategies often yield good results at all lead times. Trends like deconstruction, input data selection, hybridization, and model selection increase model accuracy. Additionally, successful standalone machine learning models like ANN, SVM, RF, genetic algorithms, KNN, and LSTM yield better outputs when hybridized with other ML models.
Overall, predictive capabilities have several advantages, including enhanced grid reliability, efficiency, renewable sources integration, reduced operational costs, and increased customer satisfaction. Predicting and preventing faults and interruptions increases the likelihood of optimizing energy distribution and demand management [11, 19]. Furthermore, better integration of renewable energy sources can be achieved by forecasting and managing their fluctuation. Enabling proactive maintenance and avoiding costly failures can significantly cut operational costs in the energy distribution and consumption matrix. The appropriate predictions can also provide individualized services and a stable electricity supply, thus improving customer satisfaction. Smart power networks are likely to become more resilient, efficient, and sustainable by harnessing data mining predictive skills [2, 3]. Eventually, this paves the way for a more stable and steady future in the sphere of energy consumption.
Scalability
Scalability is critical since smart power grids generate large amounts of data at rapid speeds. However, many modern data stream mining algorithms are intended to deal with the enormous volume and velocity of power grid data. Techniques like clustering, classification, and frequent pattern mining have been developed for real-time settings, maintaining their effectiveness as grid complexity increases [20].
There are nonetheless several scalability challenges that must be continuously addressed for effective and efficient real-time monitoring and control of smart power grids. Data volume, velocity, and variety, coupled with computational complexities and real-time constraints, can present huge challenges in scalability of the techniques [21].
Smart grids generate large amounts of data from a variety of sources, including smart meters, sensors, and control devices. To handle this large amount of data, efficient storage and processing capabilities are required (Figs. 4, 5). Data is created on a constant basis and must frequently be analyzed in real or near-real-time to allow for prompt decision-making [22]. This high velocity necessitates the use of fast and efficient algorithms. Data comes in a variety of formats (structured, semi-structured, and unstructured) from numerous sources, making integration and analysis difficult. Some data mining algorithms, particularly large machine learning models, can be computationally demanding, necessitating significant processing power. Many smart grid applications necessitate real-time or near-real-time analysis, imposing stringent time limits on data processing [22, 23].
Fig. 4 [Images not available. See PDF.]
Data mining techniques categorization.
Source: Chaudhry et al. [23]
Fig. 5 [Images not available. See PDF.]
Clustering techniques categorization for data mining.
Source: Chaudhry et al. [23]
Despite the aforementioned drawbacks, there are various strategies that can be adopted to enhance scalability of data stream mining techniques. Hadoop and MapReduce frameworks can enable distributed storage and processing of large datasets across a cluster of computers. Furthermore, Spark is a general-purpose and accelerated cluster computing system that can be used to handle both batch and stream processing in good time [24]. Through cloud computing, platforms such as Google Cloud, azure, and AWS can offer scalable systems and services for data storage, processing, analysis, and uptake. This implies that cloud-based solutions can provide on-demand resources and flexibility to handle fluctuating data volumes.
Stream processing frameworks, such as Apache Kafka and Apache Flink, can also play a very significant role in scalability. For example, Apache Kafka is a distributed streaming platform designed to handle large amounts of real-time data, whereas Apache Flink is a stream processing framework capable of complicated event processing and real-time analysis [25]. Another essential scalability enhancing strategy is edge computing, which allows processing of data closer to the source, such as smart meters or substations, and in the process reducing the amount of data transmissible to external servers. Edge computing can enhance response times and lower latency in real-time applications [26].
Furthermore, algorithm optimization is enabled through developing of more efficient algorithms with bigger memory space and faster processing capability (Fig. 6). Approximation algorithms are applied to trim down computational burden. Data reduction techniques, such as data aggregation and feature selection, can be used to summarize data at various degrees of granularity, reducing the amount of data that must be processed. Feature selection offers the advantage of choosing the most relevant features for analysis, reducing the dimensionality of the dataset. Specific to smart grids, data mining techniques can be used for scalable anomaly detection, scalable load forecasting, and scalable grid monitoring [25, 26] (Fig. 7).
Fig. 6 [Images not available. See PDF.]
Components involved in real-time stream processing.
Source: Mehmood and Anees [24]
Fig. 7 [Images not available. See PDF.]
Energy management system optimization approaches.
Source: Balta-Ozkan et al. [27]
Regarding anomaly detection, distributed algorithms and stream processing frameworks can be used to discover faults in real-time across a wide area network. Moreover, cloud-based machine learning models are used in the prediction of power demand for large regions. In grid monitoring, edge computing is used for processing of sensor data locally while only transmitting relevant information to central servers [25]. By adopting these strategies, data mining techniques can be successfully scaled to manage the huge data volumes and real-time requirements of modern smart grids, enabling more efficient and reliable grid operations.
Integration with IoT
Advanced data stream mining techniques work seamlessly with internet of things (IoT) devices, allowing the grid to improve energy distribution and incorporate renewable energy sources [28]. This combination unlocks the potential for advanced grid control and optimization. The role of IoT in smart grids include enhancing of data acquisition, communication systems, and distributed intelligence [8, 29]. Smart meters, sensors, and actuators are some of the IoT devices used across the grid to capture real-time data on voltage, current, power flow, temperature, and equipment status. In addition, IoT enables communication networks that allow for seamless data transmission between grid devices and control centers. IoT devices can also undertake local processing and control tasks, allowing for distributed decision-making and quicker response times [29].
Data mining enhances IoT-enabled smart grids in various ways, including real-time analysis, turning data into insights, grid organization, predictive maintenance, and demand response organization [8, 30]. IoT devices generate vast volumes of data, which can be overwhelming. Data mining techniques extract useful insights from this data by showing patterns, trends, and anomalies that are impossible to detect manually.
Data stream mining algorithms process the continuous flow of IoT data in real time, allowing for the detection of defects, anomalies, and other key events (Fig. 8). Data mining, which analyzes sensor data from IoT devices, can forecast equipment faults and enable proactive maintenance, lowering downtime and costs [31]. By analyzing real-time data from IoT devices, data mining can improve energy distribution, reduce transmission losses, and boost overall grid efficiency. Data from smart meters (IoT devices) can be used to build demand response systems, which encourage consumers to minimize their energy consumption during peak periods.
Fig. 8 [Images not available. See PDF.]
Holistic view of a real-time analytics.
Source: Bhattacharya et al. [32]
Key integration aspects of IoT-enabled smart grids entail data management and security, scalability, and interoperability. Efficient storage, management, and processing of enormous amounts of data created by IoT devices is critical. Protecting the sensitive data collected by IoT devices from illegal access and cyberattacks is important. Moreover, the combined IoT and data mining system must be scalable to accommodate the increasing number of linked devices and data volumes. It is critical to ensure that different IoT devices and systems communicate and exchange data in a seamless manner [33] (Fig. 9).
Fig. 9 [Images not available. See PDF.]
Fault monitoring on distribution transformers using embedded sensors.
Source: Odongo et al. [34]
Fault detection and diagnosis, smart meter analysis, and predictive maintenance of transformers are the ways in which the integration can be helpful. Using sensor data from IoT devices to detect and diagnose grid failures in real time allows for faster restoration of services. Analyzing smart meter data is crucial for understanding consumer energy use patterns, detecting power theft, and implementing targeted energy efficiency initiatives [35]. Analyzing sensor data from transformers, or IoT devices, to predict potential failures and schedule maintenance in advance can also be beneficial to smart power grids. Smart power grids can also gain increased intelligence, automation, and efficiency by combining data mining with IoT, resulting in a more reliable, sustainable, and resilient energy infrastructure [36].
Technical implementation of machine learning and edge computing in power grid in Kenya and other countries
In order to improve real-time decision-making and predictive analytics, Kenya’s electricity grid is implementing edge computing and machine learning (ML) by deploying distributed intelligence at several grid nodes. To predict electricity consumption, identify anomalies, and improve load balancing, ML models, such as convolutional neural networks (CNNs) and long short-term memory (LSTM) networks, are trained on historical and current grid data [5]. These models analyze data from SCADA systems, smart meters, and Internet of Things sensors to find trends that point to malfunctions, power theft, or inefficiencies in the system. However, latency problems arise due to the dependence on centralized cloud processing, which is why edge computing is crucial.
Although Kenya, like many other countries, is incorporating edge computing and ML into its electricity systems, the scale, architecture, and level of sophistication of these implementations vary [36]. For instance, with its more advanced electrical industry, South Africa has more significantly integrated ML for defect detection, load forecasting, and predictive maintenance, especially in Eskom’s grid management [37]. Furthermore, South Africa and India have used advanced grid analytics and predictive maintenance to lower their transmission losses to less than 10%, but Kenya has losses of about 18% [38]. In a similar vein, Kenya’s smart meter adoption rate is still below 30%, while countries like China and Germany have adopted over 80% of smart meters, improving demand-side management in real time [39]. Additionally, machine learning algorithms optimize energy distribution in sophisticated grids like in the US and Germany, lowering downtime and increasing forecasting accuracy. Infrastructure deficiencies, ineffective regulations, and lack of edge computing usage cause Kenya to lag behind, underscoring the necessity of policy-driven investments in data-driven grid modernization [38].
Rapid fault identification and response are ensured by lightweight, real-time machine learning inference at the grid edge made possible by technologies like TensorFlow Lite and Apache Kafka. Integrating renewable energy sources, such as solar, wind, and geothermal, is especially important in Kenya, where edge-based predictive analytics help reduce supply variations [40]. Furthermore, by decentralizing grid data management, blockchain-based edge computing frameworks improve security by preventing cyberattacks [41]. Despite obstacles including sparse computational infrastructure and expensive installation costs, Kenya’s grid resilience, efficiency, and sustainability might be greatly increased by utilizing 5G and LoRaWAN networks for smooth edge-device communication.
Unlike South Africa, Kenya is still in the early stages of integrating machine learning (ML) into the power grid, with a primary focus on demand-side management and renewable energy integration [37]. Kenya Power has tested ML-based smart meters and anomaly detection systems, but their deployment is hampered by lack of technical know-how and processing capacity. In Kenya, edge computing is mostly used in remote renewable mini-grids to enhance real-time reaction and lessen reliance on centralized systems. Scalability and regulatory backing, however, continue to be major obstacles. Though both nations face difficulties in optimizing these technologies for large-scale smart grid transformation, Kenya is making progress in decentralized grid management, particularly in renewables, while South Africa leads in advanced machine learning applications and edge computing integration because of its developed power infrastructure.
Challenges of smart power grid data stream mining techniques in Kenya
While data stream mining has tremendous potential for smart grids in Kenya, there are distinct challenges that must be addressed. These can be broadly seen in terms of infrastructural limitations, data-related issues, technical and expertise shortcomings, and socioeconomic and regulatory challenges. Mishra and Manda [42] noted that consistent and high-bandwidth communication networks are required for real-time data transmission from IoT devices to data processing centers. However, Tuitoek [39] noted that in some parts of Kenya, network coverage may be limited, inconsistent, or prone to outages, limiting data collection and analysis in real-time. Integrating new smart grid technology into existing legacy infrastructure can be difficult, whereas older technology may be incompatible with current communication protocols and data formats, necessitating costly updates or retrofitting.
Furthermore, establishing a large-scale IoT infrastructure, such as smart meters, sensors, and communication networks, can be costly, particularly in a developing nation such as Kenya [39, 43]. Traditional techniques frequently fail to capture nonlinear and dynamic interactions within power networks, especially when dealing with renewable energy sources with highly fluctuating outputs [44]. This constraint has an impact on the precision with which insights can be obtained. There are data-related challenges, such as missing values and data quality, data privacy and security, and data nonuniformity. Data generated from sensors and other sources can be noisy, incomplete, or incorrect for various reasons, including sensor breakdowns and communication errors. Data arrives in multiple formats from many sources, making integration and analysis difficult [42].
While some algorithms are scalable, others experience difficulties as the grid expands or the volume of data grows, especially when incorporating distributed energy resources (DERs). Processing and interpreting high-frequency, high-dimensional data in real time necessitates substantial processing resources. However, many existing algorithms are resource-intensive and struggle to deliver results in limited settings [42, 44, 45]. The real-time nature of data stream mining in smart grids makes it vulnerable to cyberattacks. Techniques must handle security concerns while maintaining processing efficiency. A lack of context-specific adaptability is another potential issue. Many algorithms are designed for generalized use cases and may fail to account for regional differences, such as infrastructure limitations, cultural nuances, or policy contexts [46] (Fig. 10).
Fig. 10 [Images not available. See PDF.]
Taxonomy of anomaly detection challenges in high-dimensional big data.
Source: Thudumu et al. [47]
Regarding technical and expertise challenges, Asiedu et al. [48] found out that deploying sophisticated data stream mining algorithms in resource-constrained situations, such as edge devices and remote substations, can be difficult. Efficient algorithms and light-weight implementations are required. According to Kang and Reiner [49]. Power consumption and grid behaviour can vary over time due to different circumstances, including weather, seasons, and economic activity. Adaptive algorithms that can identify and respond to these changes are required. Furthermore, according to Owino [50], developing and sustaining complicated data stream mining systems necessitates specialized knowledge, which may be scarce in Kenya. Investing in training and educational programs is therefore critical. Standardized data formats and protocols are required to enable data sharing and interoperability. Dealing with these data quality challenges is critical to accurate analysis. Protecting critical grid data from illegal access and hacking should be a top priority [48]. Strong security measures are required to maintain data confidentiality, integrity, and availability.
Socioeconomic and regulatory challenges revolve around digital divide, public awareness and acceptance, and existing regulatory frameworks. Balta-Ozkan [49] noted that inequitable access to technology and digital literacy can obstruct the uptake and effective application of smart grid technologies. This calls for the smart grid technology to be widely understood and accepted in order to be implemented successfully. It is also critical to address issues of data privacy and security. However, clear and supportive regulatory frameworks are required to stimulate investment in smart grid infrastructure and data-driven solutions.
Specific examples/case studies of smart grid challenges in Kenya
Implementing of data stream mining techniques has presented a number of challenges for Kenya’s transition to a smart power system. The Decarbonizing the Energy Mix Initiative, Nairobi’s Distribution Network Reliability, and Mini-Grid Implementations in Rural Areas are notable case studies that highlight these difficulties [51]. A study by Goga and Chang [51] evaluating Nairobi’s grid resilience brought to light problems with failure rates and voltage stability. The intricacy of combining several energy sources, including solar, wind, hydro, and geothermal, makes predictive maintenance and real-time data processing more difficult. The inability to interpret high-velocity data due to lack of sophisticated data stream mining techniques causes delays in fault identification and response.
A study by Kahlen et al. [40] on Kenya Power’s plan to incorporate renewable energy sources and reduce reliance on fossil fuels revealed that data management has proven very difficult. To properly balance supply and demand, variable renewable energy integration calls for advanced data analytics. However, current data stream mining capabilities are inadequate, which results in greater operating costs and inefficient grid management. Another research by Hanbashi et al. [52] on modelling and validation of typical PV mini-grids in Kenya, the findings showed that data collection and analysis have presented difficulties for the mini-grid implementation in Kenya’s rural areas. The deployment of real-time data stream mining techniques is hindered by limited infrastructure and high expenses, which impacts the monitoring and optimization of these isolated grids. Consequently, it is challenging to integrate these mini-grids into the national grid and provide a consistent power supply. These case studies highlight the necessity of improved computational resources, skilled personnel, and supportive legal frameworks in order to successfully apply data stream mining techniques in Kenya’s smart power grid.
Opportunities for improving real-time monitoring and control in Kenya’s smart power grids using advanced data stream mining techniques
Kenya’s energy sector is at a crossroads, with population growth, urbanization, and economic development all driving up the need for reliable power supply. According to Odongo et al. [34], as Kenya moves toward smart power grids, real-time monitoring and control become critical for efficient energy distribution, grid stability, and renewable energy integration. Advanced data stream mining techniques provide transformative potential in meeting these objectives by enabling real-time analytics and decision-making. Furthermore, Kimani [53] revealed that there are several emerging opportunities for improving real-time monitoring and control of the smart power grids, by using AI-driven technology, edge computing, adaptive algorithms, and hybrid models, among others.
While data stream mining techniques have significantly transformed power grid management, they are yet to reach their full potential due to technical, computational, and contextual limitations. The use of AI, edge computing, and hybrid techniques provides substantial opportunity to address these restrictions. However, obtaining optimal performance will require continual innovation, targeted investments, and coordination among researchers, policymakers, and industry players. According to Tchao et al. [54], the starting point could entail improved infrastructural investment, quality data management, human resource capacity building, public engagement, cybersecurity measures, and policy and regulatory support. Investing in reliable communication networks and upgrading current infrastructure is very critical. This also includes using data cleaning and preprocessing techniques to deal with noisy and incomplete data [39]. Implementing robust security processes and safeguards to protect grid data should be given priority. Equally important is investing in training and education initiatives to foster local skills in data stream mining and smart grid technology.
The integration of artificial intelligence (AI) and deep learning is greatly improving data stream mining capabilities in terms of accuracy, adaptability, and scalability. AI models can also self-learn from changing data patterns, making them extremely useful in smart grid applications. Using data stream mining techniques at the edge, or closer to data sources, can minimize latency and processing costs on central systems [24]. This strategy is very promising for rural areas and resource-constrained power grids. Furthermore, integrating various techniques, like clustering and anomaly detection, can overcome the limitations of single methods, resulting in more robust insights [52].
Research on adaptive data stream mining algorithms that can react to changing grid circumstances is ongoing, presenting answers to the rigidity of existing strategies [27]. Meanwhile, it is important to raise public awareness and address concerns regarding smart grid technology by creating clear and supportive policies and regulations.
Using advanced techniques for real-time monitoring can be another promising solution for smart power grid management [44]. Advanced data stream mining techniques, such as deep learning, reinforcement learning, and adaptive clustering, can greatly improve real-time monitoring of Kenya’s electricity infrastructure. These can be effective in renewable energy integration, fault detection and isolation, and dynamic loading balancing. Predictive models and pattern recognition algorithms help to regulate the fluctuation of renewable energy sources like wind and solar, ensuring their seamless integration into the grid [34] (Fig. 11).
Fig. 11 [Images not available. See PDF.]
Data vs. performance comparison.
Source: Owino [50]
Advanced anomaly detection systems can detect defects, such as transformer breakdowns or line disruptions, almost immediately. These strategies use real-time sensor data to identify and pinpoint errors, reducing downtime. As noted by Rafati et al. [56], real-time classification algorithms can forecast and adapt power distribution based on demand patterns, resulting in more efficient energy allocation and fewer overloads.
Predictive analysis can enhance control, with control mechanisms in smart grids relying primarily on data-driven predictions and responses [53]. This process is enhanced through advanced stream mining techniques by enabling demand response management, voltage and frequency stability, and proactive maintenance of smart power grid systems. Improving control through predictive analysis enables machine learning algorithms to forecast demand peaks and automate load adjustments by analyzing historical and real-time data. Furthermore, adaptive mining techniques monitor voltage and frequency aberrations and provide real-time remedial steps to help stabilize the grid. Proactive maintenance involves detecting equipment that is likely to fail, allowing for preventative repairs, and reducing unplanned outages [54].
Overcoming infrastructure challenges through edge computing, and federated learning can also be instrumental in improving real-time monitoring and control in Kenya’s smart power grids using advanced data stream mining techniques. Kenya presents unique infrastructural constraints, such as inadequate grid coverage, inconsistent data from rural places, and outdated equipment [47]. However, the situation can be improved using edge computing, and federated learning. Applying stream mining algorithms on local nodes lowers latency, improves data processing in remote places, and reduces the demand for centralized computing resources. Furthermore, collaborative models enable several grid nodes to learn from shared data without transmitting raw data to a central server, which alleviates bandwidth limitations [13].
Addressing security concerns is crucial in improving real-time monitoring and control of smart power grids. With their growing reliance on data and communication technology, Smart power grids are vulnerable to a variety of cyberattacks. Advanced mining techniques can be integrated with blockchain and encryption technology to provide secure data transmission. Anomaly detection systems can uncover potential cyber threats in real time, hence improving grid security. Identifying vulnerabilities can guarantee data confidentiality, integrity, availability, authentication and authorization, as well as network and device security [13, 47, 54].
It is crucial to protect sensitive data, such as consumer energy usage trends and grid operational data, from unwanted access. This ensures that data is not tampered with or altered while in transmission or storage. Keeping vital data and systems available when needed is critical for preventing denial-of-service attacks. Furthermore, detecting vulnerabilities will allow us to validate the identities of people and devices while also regulating their access to grid systems and data [8, 55]. Similarly, network security ensures the protection of communication networks that connect grid equipment and control centers from breaches and attacks. Individual IoT devices, such as smart meters and sensors, must also be protected against malware outbreaks and tampering.
Encryption, Access Control, Intrusion Detection and Prevention Systems (IDPS), firewalls, authentication and authorization mechanisms, security audits and penetration testing, and security information and event management (SIEM) are some of the critical security measures that can be instituted to forestall cyberattacks to smart power grids [56, 57]. Strong encryption algorithms to protect data during transmission and storage, strict access control policies and mechanisms to limit access to sensitive data and systems, and the deployment of IDPS to monitor network traffic and detect malicious activity can all help to ensure the safety of smart power grids [58].
Additionally, firewalls can regulate network access and prevent unwanted data access. Strong authentication measures, such as multi-factor authentication, can also be implemented to authenticate user and device identities. Security audits and penetration testing should be conducted on a regular basis to detect system vulnerabilities and flaws [59]. SIEM systems can be useful for gathering and analyzing security logs from multiple sources in order to detect and respond to security problems. Additionally, investigating the application of blockchain for secure data sharing and transaction management in the grid is critical.
Specific considerations for data stream mining would include anomaly detection for cyberattacks, secure data aggregation and sharing, and privacy-preserving data mining. To safeguard data, data stream mining techniques are required to detect irregularities in smart power grids that could suggest cyberattacks [60]. It is vital to implement secure mechanisms for data gathering and exchange across grid stakeholders. Equally crucial is the use of techniques like differential privacy and homomorphic encryption, which allow for data analysis without jeopardizing individual privacy. Wang et al. [61] identified security awareness training and incident response planning as two human elements that should be addressed. This can be accomplished by providing grid personnel and customers with regular security awareness training to prevent social engineering attacks and promote security best practices, as well as developing and testing incident response plans to ensure a coordinated and effective response to security incidents.
Kenya’s expanding renewable energy sector, particularly geothermal, solar, and wind, makes advanced data stream mining approaches both timely and significant [10]. These strategies will maximize renewable energy consumption, eliminate technical losses, and help to achieve sustainable energy goals. However, according to Takase et al. [62], these achievements may not be possible without proper and effective policy frameworks. Therefore, there should be policy support and skills development. Regulatory frameworks should promote investment in advanced data mining technologies and establish rules for their implementation. Furthermore, training programs for grid operators and engineers in advanced data analytics and mining techniques are required for successful implementation [63, 64]. Aside from creating security standards and guidelines, it is necessary to encourage information sharing and collaboration among grid operators, security experts, and government entities [65].
Conclusions
This study explored the potentials of data stream mining techniques for real-time monitoring and control of smart power grids in Kenya. Based on the critical analysis, the study concluded that while Kenya has made strides in implementing data stream mining techniques for real-time monitoring and control in its smart power grid systems, these techniques are essentially basic and limited. Current systems are focused on fault detection and basic load balancing, but they lack the sophistication needed to handle the enormous volume, velocity, and fluctuation of modern smart grid data. Notwithstanding limited data on the specific data stream mining techniques currently used in smart power grids in Kenya, given the global trends in smart grid evolution, techniques such as clustering, classification, frequent pattern mining, anomaly detection, and regression are important in the Kenyan situation.
The effective application of data stream mining techniques in Kenya’s smart power grid systems is hindered by several challenges. Some of these challenges include inadequate computational capacity to analyze large-scale real-time data, poor data quality such as noise and missing values, which influence analytics accuracy, and grid operators’ lack of expertise and technical abilities in advanced data mining techniques. Other considerations include the high cost of implementation, particularly for sophisticated algorithms and edge computing technologies, as well as legal and policy gaps that do not prioritize or incentivize advanced analytics adoption.
Advanced data stream mining techniques, such as machine learning, deep learning, and adaptive clustering, offer considerable improvements in real-time monitoring and control. These technologies provide dynamic load balancing, improved fault detection, and greater integration of renewable energy sources. Furthermore, edge computing and federated learning provide practical solutions to infrastructure and connectivity issues, especially in rural and remote areas. Additionally, scalable algorithms and privacy-preserving data mining are very important techniques for handling the growing volume and velocity of data, and addressing user privacy and security concerns, respectively.
In conclusion, this study emphasizes the importance of sophisticated data stream mining techniques to improving Kenya’s smart power grid’s resilience, efficiency, and reliability. The findings emphasize the critical need for workforce capacity building, infrastructure upgrading, and regulatory alignment to enable real-time grid intelligence for industry stakeholders, including electricity utilities, technology providers, and investors. To guarantee data-driven energy management decision-making, policymakers must give top priority to encouraging legislative frameworks, smart grid adoption incentives, and strong cybersecurity safeguards. Beyond Kenya, these observations provide insightful guidance for other developing countries dealing with comparable grid modernization challenges.
Future studies should focus on creating localized machine learning models that are tailored to Kenya’s grid conditions, investigating decentralized energy trading enabled by blockchain technology for increased efficiency, and examining hybrid AI-edge computing architectures for low-latency, real-time grid optimization. Longitudinal research evaluating the socioeconomic effects of smart grid analytics on energy affordability and accessibility will also support sustainable reforms in the power sector.
Recommendations
The study recommended adoption of advanced data stream mining techniques, where power utility companies are encouraged to invest in machine learning and AI-enabled data stream mining tools to enhance grid operations adaptability and predictive capabilities. Within the Kenyan context, a national smart grid strategy and legal framework should be implemented, focusing on smart grid policy which should outline data governance, cybersecurity standards, and interoperability requirements. There is also need for Kenya’s smart power grid operate within regulatory incentives for technological investment. This can allow tax incentives and subsidies for power utilities and private sector players investing in ML, edge computing, and advanced data analytics.
Improved data management practices will ensure high-quality, clean, and reliable data input for stream mining algorithms. Other policy frameworks can include national data management and cybersecurity framework for facilitating smooth integration between cloud systems, IoT devices, and grid components, through data standardization rules. A renewable energy and grid decentralization policy can also be created to support data-driven decision-making on integrating wind, solar, and mini-grids.
Building technical capacity in advanced data mining and analytics will be essential through specialized training programs, especially for grid operators and engineers. This can be done through partnerships with research and academic institutions which can effectively identify local innovation needs in data stream mining. Developing supportive policy frameworks will also be critical in establishing clear regulations and incentives for adopting advanced analytics in smart power grids. Promoting public–private partnerships will help to create an environment for encouraging cost-sharing in advanced mining techniques between government and private entities. Power utility companies should also be encouraged to initiate pilot projects for testing and refining advanced data stream mining techniques in specific regions or use cases, such as renewable energy management or urban load balancing.
Author contributions
Cornelius Mutuku Mutuku wrote the main manuscript paper whereas George Onyango Okeyo and Joseph Muliaro Wafula reviewed the manuscript.
Funding
The authors did not receive support from any organization for the submitted work. No funding was received to assist with the preparation of this manuscript, conducting this study or any other support during submission or preparation of this work.
Data availability
The data supporting the findings of this study are available from the corresponding author upon request.
Declarations
Ethics approval and consent to participate
All authors certify that this part is not applicable.
Consent for publication
All authors certify that this part is not applicable.
Competing interests
The authors declare no competing interests.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
1. Ahmad, T; Zhang, D. Using the internet of things in smart energy systems and networks. Sustain Cities Soc; 2021; 68, [DOI: https://dx.doi.org/10.1016/j.scs.2021.102783] 102783.
2. Hossein Motlagh, N; Mohammadrezaei, M; Hunt, J; Zakeri, B. Internet of Things (IoT) and the energy sector. Energies; 2020; 13,
3. As, KN; Desai, K; Kautish, S. Optimizing green power and green energy through digital technologies. Digital technologies to implement the UN sustainable development goals; 2024; Cham, Springer: pp. 275-304.
4. Tapo AA, Traoré A, Danioko S, Tembine H. Machine Intelligence in Africa: a survey. arXiv:2402.02218. 2024. https://arxiv.org/abs/2402.02218.
5. Ochola, DO. Static synchronous compensator (STATCOM) related factors and performance of voltage regulation project, a case of lake Turkana wind power plant; 2023; Nairobi, University of Nairobi:
6. Zang, T; Wang, S; Wang, Z; Li, C; Liu, Y; Xiao, Y; Zhou, B. Integrated planning and operation dispatching of source–grid–load–storage in a new power system: a coupled socio–cyber–physical perspective. Energies; 2024; 17,
7. Hua, H; Li, Y; Wang, T; Dong, N; Li, W; Cao, J. Edge computing with artificial intelligence: a machine learning perspective. ACM Comput Surv; 2023; 55,
8. Abir, SA; Anwar, A; Choi, J; Kayes, AS. IoT-enabled smart energy grid: applications and challenges. IEEE Access; 2021; 9, pp. 50961-50981. [DOI: https://dx.doi.org/10.1109/ACCESS.2021.3067331]
9. Munywoki, MN. Integrating a web-based GIS in the optimization of the customer connection process for utility company: a case of Kenya power; 2021; Nairobi, Lighting Company:
10. Gaspar, J; Cruz, T; Lam, CT; Simões, P. Smart substation communications and cybersecurity: a comprehensive survey. IEEE Commun Surv Tutor; 2023; 25,
11. Ahmad, T; Madonski, R; Zhang, D; Huang, C; Mujeeb, A. Data-driven probabilistic machine learning in sustainable smart energy/smart energy systems: key developments, challenges, and future research opportunities in the context of smart grid paradigm. Renew Sustain Energy Rev; 2022; 160, [DOI: https://dx.doi.org/10.1016/j.rser.2022.112128] 112128.
12. Tantalaki, N; Souravlas, S; Roumeliotis, M. A review on big data real-time stream processing and its scheduling techniques. Int J Parallel Emergent Distrib Syst; 2020; 35,
13. Vinothine, S; Widanagama Arachchige, LN; Rajapakse, AD; Kaluthanthrige, R. Microgrid energy management and methods for managing forecast uncertainties. Energies; 2022; 15,
14. Jiang, Y; Yin, S; Kaynak, O. Performance supervised plant-wide process monitoring in industry 4.0: a roadmap. IEEE Open J Ind Electron Soc; 2020; 2, pp. 21-35. [DOI: https://dx.doi.org/10.1109/OJIES.2020.3046044]
15. Al-Mashhadani, SWT; Kurnaz, S. A novel function of a research process based on a power internet of things architecture intended for smart grid demand schemes. Appl Sci; 2024; 14,
16. Wicaksono, H; Trat, M; Bashyal, A et al. Artificial-intelligence-enabled dynamic demand response system for maximizing the use of renewable electricity in production processes. Int J Adv Manuf Technol; 2024; [DOI: https://dx.doi.org/10.1007/s00170-024-13372-7]
17. Gawusu, S; Mensah, RA; Das, O. Exploring distributed energy generation for sustainable development: a data mining approach. J Energy Storage; 2022; 48, [DOI: https://dx.doi.org/10.1016/j.est.2022.104018] 104018.
18. Alabi, TM; Aghimien, EI; Agbajor, FD; Yang, Z; Lu, L; Adeoye, AR; Gopaluni, B. A review on the integrated optimization techniques and machine learning approaches for modeling, prediction, and decision making on integrated energy systems. Renewable Energy; 2022; 194, pp. 822-849. [DOI: https://dx.doi.org/10.1016/j.renene.2022.05.123]
19. Marin-Castro, HM; Tello-Leal, E. Event log preprocessing for process mining: a review. Appl Sci; 2021; 11,
20. Asif, M; Kuglitsch, MM; Pelivan, I et al. Review and intercomparison of machine learning applications for short-term flood forecasting. Water Resour Manag; 2025; 39, pp. 1971-1991. [DOI: https://dx.doi.org/10.1007/s11269-025-04093-x]
21. Bhanage, DA; Pawar, AV; Kotecha, K. It infrastructure anomaly detection and failure handling: a systematic literature review focusing on datasets, log preprocessing, machine & deep learning approaches and automated tool. IEEE Access; 2021; 9, pp. 156392-156421. [DOI: https://dx.doi.org/10.1109/ACCESS.2021.3128283]
22. Duchesne, L; Karangelos, E; Wehenkel, L. Recent developments in machine learning for energy systems reliability management. Proc IEEE; 2020; 108,
23. Chaudhry, M; Shafi, I; Mahnoor, M; Vargas, DL; Thompson, EB; Ashraf, I. A systematic literature review on identifying patterns using unsupervised clustering algorithms: a data mining perspective. Symmetry; 2023; 15,
24. Mehmood, E; Anees, T. Challenges and solutions for processing real-time big data stream: a systematic literature review. IEEE Access; 2020; 8, pp. 119123-119143. [DOI: https://dx.doi.org/10.1109/ACCESS.2020.3005268]
25. Hassan, A; Mhmood, AH. Optimizing network performance, automation, and intelligent decision-making through real-time big data analytics. Int J Responsible Artif Intell; 2021; 11,
26. Dileep, GJ. A survey on smart grid technologies and applications. Renewable Energy; 2020; 146, pp. 2589-2625. [DOI: https://dx.doi.org/10.1016/j.renene.2019.08.092]
27. Balta-Ozkan N, Aboli Z, Netto G, Hasan S. Energy digitalisation: towards more equitable outcomes for minority ethnic communities. 2024. https://researchportal.hw.ac.uk/en/publications/energy-digitalisation-towards-more-equitable-outcomes-for-minorit.
28. Mishra, S; Komandla, V; Bandi, S. Leveraging in-memory computing for speeding up Apache Spark and Hadoop distributed data processing. J AI Assist Sci Discov; 2022; 2,
29. Alam, MA; Nabil, AR; Mintoo, AA; Islam, A. Real-time analytics in streaming big data: techniques and applications. J Sci Eng Res; 2024; 1,
30. Nama, P; Bhoyar, M; Chinta, S. AI-powered edge computing in cloud ecosystems: enhancing latency reduction and real-time decision-making in distributed networks. Well Test J; 2024; 33,
31. Majhi, AA; Mohanty, S. A comprehensive review on internet of things applications in power systems. IEEE Internet Things J; 2024; 11,
32. Bhattacharya, SN; Mitra, S. Suresh G (2013) The shear wave velocity of the upper mantle beneath the Bay of Bengal, northeast Indian Ocean from interstation phase velocities of surface waves. Geophys J Int; 2013; 193,
33. Kumar, NM; Chand, AA; Malvoni, M; Prasad, KA; Mamun, KA; Islam, FR; Chopra, SS. Distributed energy resources and the application of AI, IoT, and blockchain in smart grids. Energies; 2020; 13,
34. Odongo, GY; Musabe, R; Hanyurwimfura, D; Bakari, AD. An efficient LoRa-enabled smart fault detection and monitoring platform for the power distribution system using self-powered IoT devices. IEEE Access; 2022; 10, pp. 73403-73420. [DOI: https://dx.doi.org/10.1109/ACCESS.2022.3189002]
35. Mazhar T, Irfan HM, Haq I, Ullah I, Ashraf M, Shloul TA, Ghadi YY, Imran, Elkamchouchi DH. Analysis of challenges and solutions of IoT in smart grids using AI and machine learning techniques: a review. Electronics. 2023;12(1):242.
36. Martins, P; Soofastaei, A. Predictive maintenance of mining machines applying advanced data analysis. Data analytics applied to the mining industry; 2020; Boca Raton, CRC Press: pp. 149-168. [DOI: https://dx.doi.org/10.1201/9780429433368-7]
37. Mlambo, F; Mhlanga, D. Artificial intelligence and machine learning for energy in South Africa. AfricaGrowth Agenda; 2022; 19,
38. Chituc CM. Interoperability standards in the IoT-enabled future learning environments: an analysis of the challenges for seamless communication. In: 2020 13th International conference on communications (COMM), Bucharest, Romania. 2020. p. 417–22. https://www.sciencedirect.com/science/article/abs/pii/S2542660520301529.
39. Tuitoek A. A framework for adoption of software defined wide area networks (Sd-wan) within the enterprises in Kenya. Doctoral dissertation, University of Nairobi. 2022. https://erepository.uonbi.ac.ke/handle/11295/161526.
40. Kahlen J, Gustavo V, Hecke J. Kenya power’s decarbonising the energy mix initiative. 2021. Available at https://newclimate.org/sites/default/files/2022-03/a2aii-kplc_scopingstudy.pdf?utm_source=chatgpt.com.
41. Atinda TE. Blockchain technology and operational efficiency in Kenya’s power sector. Doctoral dissertation, University of Nairobi. https://erepository.uonbi.ac.ke/handle/11295/162821.
42. Mishra, S; Manda, J. Improving real-time analytics through the internet of things and data processing at the network edge. J AI Assist Sci Discov; 2024; 4,
43. Adams, JN; Bélafi, ZD; Horváth, M; Kocsis, JB; Csoknyai, T. How smart meter data analysis can support understanding the impact of occupant behavior on building energy performance: a comprehensive review. Energies; 2021; 14,
44. Völker, B; Reinhardt, A; Faustine, A; Pereira, L. Watt’s up at home? Smart meter data analytics from a consumer-centric perspective. Energies; 2021; 14,
45. Ogello, VA; Rono, BK; Ngure, K; Sedah, E; Thuo, NB; Musinguzi, N; Baeten, JM; Bukusi, EA; Mugo, NR; Haberer, JE. Acceptability and feasibility of long-term, real-time electronic adherence monitoring of HIV pre-exposure prophylaxis (PrEP) use among young women in Kenya: a mixed methods study. PLoS ONE; 2024; 19,
46. Panda, DK; Das, S. Smart grid architecture model for control, optimization and data analytics of future power networks with more renewable energy. J Clean Prod; 2021; 301, [DOI: https://dx.doi.org/10.1016/j.jclepro.2021.126877] 126877.
47. Thudumu, S; Branch, P; Jin, J; Singh, J. A comprehensive survey of anomaly detection techniques for high dimensional big data. J Big Data; 2020; 7, pp. 1-30. [DOI: https://dx.doi.org/10.1186/s40537-020-00320-x]
48. Asiedu MN, Dieng A, Haykel I, Rostamzadeh N, Pfohl S, Nagpal C, Nagawa M, Oppong A, Koyejo S, Heller K. The case for globalizing fairness: a mixed methods study on colonialism, AI, and health in Africa. In: Proceedings of the 4th ACM conference on equity and access in algorithms, mechanisms, and optimization. 2024. p. 1–24. https://doi.org/10.1145/3689904.3694708.
49. Kang, J; Reiner, DM. What is the effect of weather on household electricity consumption? Empirical evidence from Ireland. Energy Econ; 2022; 111, [DOI: https://dx.doi.org/10.1016/j.eneco.2022.106023] 106023.
50. Owino, A. Challenges of computer vision adoption in the Kenyan agricultural sector and how to solve them: a general perspective. Adv Agric; 2023; 2023,
51. Goga, OK; Chang, C-K. Assessing grid reliability for optimal integration of system-integrated modular advanced reactor technology (SMART) for Kenya: a case of the Nairobi distribution network. Energies; 2024; 17,
52. Murthy, V; Ramakrishna, S. A review on global E-waste management: urban mining towards a sustainable future and circular economy. Sustainability; 2022; 14,
53. Kimani KK. Design and simulation of a smart power grid communication network over TV white space using cognitive machine-to-machine networking. Doctoral dissertation, JKUAT-COETEC. 2020. http://ir.jkuat.ac.ke/handle/123456789/5303.
54. Tchao, ET; Quansah, DA; Klogo, GS; Boafo-Effah, F; Kotei, S; Nartey, C; Ofosu, WK. On cloud-based systems and distributed platforms for smart grid integration: challenges and prospects for Ghana’s Grid Network. Sci Afr; 2021; 12, e00796.
55. Mohammadi, M; Mohammadi, A. Empowering distributed solutions in renewable energy systems and grid optimization. Distributed machine learning and computing: theory and applications; 2024; Cham, Springer: pp. 141-155.
56. Rafati, A; Mirshekali, H; Shaker, HR. Overload alarm prediction in power distribution transformers. Smart Grids Sustain Energy; 2024; 9,
57. Yazdi, M. Maintenance strategies and optimization techniques. Advances in computational mathematics for industrial system reliability and maintainability; 2024; Cham, Springer: pp. 43-58. [DOI: https://dx.doi.org/10.1007/978-3-031-53514-7_3]
58. Rotich, IK; Chepkirui, H; Musyimi, PK. Renewable energy status and uptake in Kenya. Energ Strat Rev; 2024; 54, [DOI: https://dx.doi.org/10.1016/j.esr.2024.101453] 101453.
59. Wang, Y; Yang, C; Lan, S; Zhu, L; Zhang, Y. End-edge-cloud collaborative computing for deep learning: a comprehensive survey. IEEE Commun Surv Tutor; 2024; 26,
60. Gunduz, MZ; Das, R. Cyber-security on smart grid: threats and potential solutions. Comput Netw; 2020; 169, [DOI: https://dx.doi.org/10.1016/j.comnet.2019.107094] 107094.
61. Wang, K; Guo, X; Yang, D. Research on the effectiveness of cyber security awareness in ICS risk assessment frameworks. Electronics; 2022; 11,
62. Takase, M; Kipkoech, R; Essandoh, PK. A comprehensive review of energy scenario and sustainable energy in Kenya. Fuel Commun; 2021; 7, [DOI: https://dx.doi.org/10.1016/j.jfueco.2021.100015] 100015.
63. Tauqeer, OB; Jan, S; Khadidos, AO; Khadidos, AO; Khan, FQ; Khattak, S. Analysis of security testing techniques. Intell Autom Soft Comput; 2021; 29,
64. Cui, L; Qu, Y; Gao, L; Xie, G; Yu, S. Detecting false data attacks using machine learning techniques in smart grid: a survey. J Netw Comput Appl; 2020; 170, [DOI: https://dx.doi.org/10.1016/j.jnca.2020.102808] 102808.
65. Xavier, Musonye S. Optimization-based modeling of Kenya’s energy system for pathways towards access to secure, affordable, and sustainable energy services. 2022. https://opinvisindi.is/handle/20.500.11815/3862.
Copyright Springer Nature B.V. Dec 2025