Machine learning algorithm based identification method for leakage current types of multiple earthing system Identification method for leakage current types of multiple earthing system

The purpose of this study is to explore the identification methods of leakage current types of grounding systems based on machine learning algorithm. Through the mechanism of leakage current generation in complex power supply area, data preprocessing, feature extraction and model training are carried out, and nonlinear parameters are solved by machine learning algorithm. According to the parameter results of the solution, whether there is leakage current in various grounding systems is judged, and finally a leakage current type identification method based on machine learning algorithm is obtained. The principle structure of leakage current type identification of various grounding systems is established, the prediction error of wind power is normalized by exchange power, and the prediction error function is selected as the weight update formula of machine learning algorithm to realize leakage current type identification combined with multi-point topology analysis in complex stations. The experimental results show that the proposed method has high and stable power supply reliability, and the difference between the proposed method and the actual value is small, so it has high identification accuracy.