Yang Qi
ABSTRACT
The Overhead Contact System (OCS) is vital for railway electrification but prone to defects from environmental and mechanical stresses. Traditional manual inspections are inefficient and error prone. This study develops an OCS defects inspection and training management system based on high-definition imaging, data mining, and K-means clustering to enhance defect detection accuracy and training quality. Following the ISO 25010 standard and rapid prototyping method, the system adopts a three-tier database architecture consisting of data sources, preprocessing, and modeling layers. K-means clustering categorizes students into advanced, moderate, and poor skill groups, enabling targeted training interventions. The five-layer system architecture provides support for administrators, teachers, and students through modules such as defect classification, exam management, and defect analysis. Evaluations confirm compliance with ISO 25010 in functionality, security, and maintainability. Practical implementation shows enhanced student defect diagnosis skills and supported the optimization of teaching decisions.
Keywords: OCS defects inspection, Data mining, K-means clustering algorithm, Decision support
https://doi.org/10.57180/nitp3295