A fault state detection method for DC charging pile charging module based on minimum fourth-order moments adaptive filtering algorithm

It is necessary to accurately judge the fault state of the charging module of DC charging pile in order to ensure the safe and reliable operation of DC charging pile. However, the fault signal processing of the fault detection method is poor, resulting in low fault detection accuracy. Therefore, a fault state detection method of DC charging pile based on the least fourth moment adaptive filtering algorithm is proposed. This method is based on the electrical structure of DC charging pile. The minimum fourth moment adaptive filtering algorithm is selected to filter the operating signal of the charging module after analyzing the fault signal characteristics of the charging module. The Lagrange algorithm is optimized to obtain the minimum fourth moment adaptive filtering of the operating signal of the charging module in the process. The shape and distribution characteristics of the running signal are analyzed accurately. The filtered operating signal is input into the improved one-dimensional convolutional neural network and bidirectional gated cyclic unit detection model, fault features of the charging module are extracted, and then support vector machine is introduced to further optimize the fault detection method. The fault state detection results of the charging module are output by using support vector machine as a model classifier and combining fault features. The test results show that the proposed method can effectively process different fault signals of charging modules of DC charging pile, determine the characteristic value distribution of each feature, and reliably detect the fault state of 10 charging modules. Its signal amplitude is only between [-0.1,1.0]m·s-1 after the fault signal is processed by this method, the number of error detection is only 2 times, the detection time is only 0.95s, and the maximum CPU ratio is only 26.7%. Moreover, the number of detected faults is consistent with the actual number under multiple concurrent faults, and the fault signal after processing by this method is better. Moreover, the accuracy of fault detection is effectively improved, and the stable operation of charging module of DC charging pile can be guaranteed to a certain extent.