Research Of Power Distribution Network Fault Line Selection Method Based On Wavelet Correlation Analysis

With the rapid development of China’s power distributing system, requirements from clients of different levels have gradually become stronger in terms of quality and reliability of electricity supplying. Due to the fact that single-phase grounding fault is of greatest probability in distributing network, Finding and clearing fault line with speed and precision has been the most critical task of power distributing network.This paper analyses the status quo of fault line selection technique for small current grounding system, to highlight the merits and shortcomings of fault line selection technique, and to outline the difficulties of realization as well. Whereas the features of single-phase grounding fault burying a common law, the fault line could be selected on the ground of much information about high frequency waves contained. Then standard of wavelet base selection and singularity detection of signals are further discussed, and the concept of correlation analysis is also explained. In addition, an MATLAB simulation example is used to study significant application of correlation analysis in fault line selection.When a single-phase grounding fault occurs in a system, transient current signal boasts the superimposition of a fundamental wave, multi-frequency waves and local noise signals. The defining factor in the process of fault line selection is the transient capacity current component. This paper, which aims at deeply analyzing the features and laws of fault line selection, and which combine the theory of wavelet transform and correlation analysis, suggests a modified criterion of fault line selection which is based on wavelet packet and correlation analysis.The criterion is as follows. The first step is to decompose the transient voltage signal by using the wavelet packet transform, in order to get a certain frequency band corresponding to maximal energy, that is, selected frequency band. The following step is to use each line’s zero-sequence current for wavelet packet decomposition and to select the node corresponding to selected frequency band for wavelet packet reconstruction. Next, establish a correlation matrix with each line’s wavelet coefficients calculated according to correlation theory, and acquire each line’s average correlation coefficient with other lines through figuring out the average of elements in the matrix. Thus the fault line could be selected by comparing the different value of maximal and minimal value with the threshold value to pick out the fault line.Besides writing program code, many experiments have been carried out to simulate all sorts of fault modes for small current grounding system on the platform of MATLAB/SIMULINK to verify feasibility and correctness of this algorithm. By comparing the experiment data and the conventional methods’ data, a conclusion has been drawn that the modified proposal of fault line selection could well expand the margin of fault line detection.