GA-SVM method for single-phase grounding fault line selection in distribution network based on feature fusion

Aiming at the low accuracy of line selection method when the data amount of single-phase grounding fault in distribution network is small, a genetic algorithm optimized support vector machine (GA-SVM) method for single-phase grounding fault line selection in distribution network based on feature fusion is proposed, which adopts Fourier transform, the active power method and wavelet packet transform decompose the transient zero-sequence current of each line under different fault conditions, extracts four features, including fundamental wave amplitude, fifth harmonic amplitude, average active power component and wavelet energy value. The four features are fused by principal component analysis method, the principal component is extracted, and the feature database is established. 80% of the feature database is used as the training set, 20% as the test set, the samples in the feature database are trained by GA-SVM to realize fault line selection. The simulation model of distribution network with five feeders is built by MATLAB/Simulink. The results show that the proposed algorithm can realize fault line selection through small sample datawith high accuracy and strong applicability.