Metering Fault Diagnosis Method Based on Multi-Granularity Anomalous Environmental Influence Factors

In order to solve the problem of metering fault diagnosis of power data collection terminal equipment, a metering fault diagnosis method based on multi-granularity anomalous environmental influence factors is proposed. Firstly, specific feature extraction rules are formulated based on experts' experience, and a subset of equipment metrology data corresponding to environmental state data is extracted as candidate shapelet features from the time series data according to a specific time window. Then, the KL divergence is used to calculate the correlation between the collected data sequence and the candidate shapelet features to obtain the optimal candidate shapelet features. Finally, the optimal shapelet feature data and the basic temporal feature data of metering acquisition are input into the gating mechanism for training, construct a loss function based on output metering fault diagnosis data and actual anomalous sample labels, iteratively feedback parameter tuning, and output a metering fault diagnosis model. Experimental results show that the algorithm can effectively improve the diagnostic accuracy of metering faults and enhance the reliability and stability of the operation of power collection terminal equipment.