Fault Line Selection In Small Current Grounding Power System Based On The PSO- SVM

In the power system, china’s low voltage distribution network widely used the way of small current grounding. Due to the increasing complexity of the distribution network structure, when a small current grounding system in single-phase ground fault occurs, the fault current value will decrease as well as on-site electromagnetic interference and other phenomena, it is difficult to quickly and accurately identify the fault line. Now, the accuracy of line selection method is low, in the result of reducing the reliability and automation level of distribution network. According to the above questions, In this paper, on the basis of understanding the line selection methods both at home and abroad,around the support vector machine and particle swarm optimization algorithm for researching fault line selection in the small current grounding.In the first,this paper elaborated the research situation and principle the fault line selection methods in small current grounding at home and abroad, and pointed out the shortcomings of existing methods, as the theoretical background of the new algorithm;By analyzing the transient and steady-state process of neutral point grounding and resonance system when the single-phase grounding fault occurs,To determine that use the method of wavelet packet transform(WPT) and Fourier transform(FFT) to extract the steady-state fundamental component,steady-state fifth harmonic component,and transient energy component from the zero sequence currents when selecting the fault line.In the second, analysised of the support vector machine classification learning algorithm and particle swarm optimization algorithm based on the theory of SVM and PSO,then using PSO to optimize the SVM machine parameter,proposed a solution of fault line selection based on PSO-SVM.Finally, build the small current grounding system model on the platform of matlab7.0,according to the different fault types for the simulation analysis form the sample data. Combined with libsvm toolbox and Matlab,and took some samples for SVM training in order to get line selection model, then extracted the test sample for inspecting line selection model, thus,the method of PSO-SVM has certain superiority in the fault line selection is verifie.