Fault Line Selection Method for Small Current Grounding System Based on Multi-Group GA-BP Neural Networks Model

In China's distribution network, which works with voltages from 6kV to 35kV, the low current grounding device is commonly adopted. The idea is to use a multi-group GA-BP neural networks model to choose single-phase grounding fault lines in low current grounding systems. With this method, you can use multi-fault features to get high-accuracy line selection in settings with complex distribution networks. The first group of GA-BP neural networks model is made with different factors, including the zero sequence current amplitude, the zero sequence current phase angle, the fifth harmonic amplitude, the fifth harmonic phase angle, and the transient current amplitude. After that, the result of the first model is used to found the second GA-BP neural network model that selects fault lines. In the end, MATLAB/Simulink is used to make sure that simulations are correct. The results show that this method works even when there are complex and variable failure environments and can get around the problem of having to use just one fault characteristic to choose a line. Because of this, it greatly improves the accuracy of choosing the fault line.