Content area
In the evolving landscape of industrial manufacturing, the quest to achieve Zero-Defect Manufacturing (ZDM) is paramount, especially in environments characterized by low-volume, high-complexity, and customized production. This praxis research focuses on a cutting-edge prediction model designed to predict manual assembly process yield and reduce defects per unit by leveraging the capabilities of an Automatic Optical Inspection (AOI) system. This system, augmented by the innovative use of Augmented Reality (AR) technology, provides immediate feedback to operators in the circuit card assembly process, marking a pivotal step in addressing the challenges of manual assembly errors and enhancing the overall quality of the manufacturing process.
Drawing upon a foundation of interdisciplinary research, this study synthesizes insights from diagnostic tools for in-line identification of critical assembly steps, mathematical models estimating human error rates, and the impact of assembly complexity on defect prediction. Previous studies have highlighted the complexity of predicting defects due to varying factors such as human reliability, assembly process intricacies, and the evolving nature of learning-forgetting along with phases of fatigue-recovery cycles. Notably, the application of AR in industrial settings has shown promising outcomes in improving operational performance, albeit with identified challenges in adoption and integration into existing workflows.
The proposed prediction model is based on the process yield of each individual component installed in Printed Circuit Board Assemblies (PCBA) as part of the manual assembly process. Having information or data regarding the behavior of the component utilized within a product, the yield of the product can be predicted based on the proportion of components used within the PCBA. Then, knowing the product yield, the number of defects and related rework can be predicted. This model becomes a self-adaptive mechanism that updates in real-time with new data acquisition, incorporating assembly complexity as a predictor for defect occurrence. This approach not only aids in identifying critical workstations that deviate from expected defect levels but also facilitates quality engineers in pinpointing and rectifying non-conformities efficiently. Additionally, integrating AR technology for immediate feedback aims to bridge the gap between manual assembly operations and digital oversight, enhancing human-machine interaction and ensuring a more intuitive and error-resistant assembly process.
This praxis research endeavors to create guidance for process owners, such as engineers and production managers, on how to develop a financial justification to support investment in a manufacturing technology capable of guiding the operator to assemble a product correctly on the first attempt. This is achieved by reducing the cognitive load on operators, enabling more accurate and faster identification of potential assembly errors, and promoting a culture of continuous improvement and learning within the manufacturing domain.
This praxis research stands at the intersection of theoretical innovation and practical application, promising not only to advance our understanding of defect prediction and reduction strategies but also to offer tangible benefits in terms of increased quality, efficiency, and cost-effectiveness in manufacturing operations. Through a meticulous examination of case studies and empirical data, this study seeks to validate the effectiveness of the proposed model and AR system in achieving ZDM, thereby contributing valuable insights to the fields of manufacturing engineering, operational management, and human factors engineering.
