Adaptive Fault Detection in EV Charging Cyber-Physical Systems

Abstract

EV charging cyber-physical systems are increasingly exposed to mixed hardware faults, communication anomalies, controller inconsistencies, and cyber-induced disturbances that often produce overlapping operational symptoms. This article presents an adaptive fault-detection and isolation framework that combines multi-source sensing, residual-based diagnosis, adaptive thresholding, fault classification, and node-level isolation logic for more reliable cyber-physical diagnosis. The results show higher fault-detection accuracy, shorter isolation time, and stronger diagnostic stability than fixedthreshold and non-adaptive baseline methods under varying disturbance conditions. Overall, the study demonstrates that adaptive diagnosis offers a practical and effective foundation for resilient EV charging cyber-physical systems.