Anomaly Diagnosis Method for Electric Vehicle Charging Equipment based on Semantic and Probabilistic Reasoning

Abstract

Against the backdrop of the large-scale advancement of new energy vehicle charging networks, anomalies such as charging failures and slow charging have become increasingly prevalent. Conventional diagnostic methodologies are constrained by a single information dimension, fragmented textual data, and inadequate intelligence. To address these challenges, this study proposes an anomaly diagnosis method for electric vehicle charging equipment based on semantic and probabilistic reasoning. The method integrates user review texts, equipment operation monitoring data, and wave recording data, excavates implicit fault clues via semantic models, extracts feature variables through the combination of wavelet packet transform and distance correlation analysis, and constructs a hierarchical Bayesian reasoning model encompassing "charging state-feature variable-fault cause" to achieve precise identification of abnormal states and root cause localization. Validation through practical operation and maintenance cases demonstrates that the proposed method effectively fuses multi-source heterogeneous data, achieving a diagnostic accuracy of over 91% for typical charging anomalies. It significantly enhances the reliability and practicality of charging anomaly diagnosis in complex scenarios, thereby providing robust technical support for the intelligent operation and maintenance of charging equipment.