Research on the Fault Diagnosis Method of Automotive Charging Pile Based on the Improved MLP with SAE

다국어 초록 (Multilingual Abstract)

In order to improve the situation that the fault data set of electric vehicle charging pile has unbalanced data distribution under each fault and the small amount of data leads to the inconspicuous data features, this paper proposes a method of SAE-MLP model for fault diagnosis of charging pile fault data. This paper frstly utilizes AE to realize the augmentation of the original data, which solves the problem of inconspicuous features, and utilizes the MLP neural network which is good at handling small sample data to classify the faults on the AE processed data. Combining the features of both SAE and MLP two models can efectively improve the accuracy of fault diagnosis of the data in this paper compared with the previous methods of small sample data processing, and minimizes computational costs.