An accurate fault location method for distribution lines based on data fusion of outcomes from multiple algorithms

Abstract

Accurate fault location technology for distribution lines is of great importance as it can shorten power outage time and enhance power supply reliability. Current and voltage transformer errors and line parameter errors will result in location error of the existing algorithms, which cannot meet the requirements of power supply reliability. To solve this problem, a data-fusion model based on multiple algorithms is proposed in this study, which makes full use of the location results of existing algorithms. First, the complementarity among the four selected algorithms is analyzed. Then the feasibility of improving fault location accuracy by constructing a data-fusion model is explained and the overall structure of the data-fusion model is determined. Finally, an artificial neural network with a strong fitting ability is used to complete the construction of the data-fusion model. The performance of the model is tested for different factors, such as fault positions, line parameters, fault resistances, and fault inception angles. The test cases based on PSCAD/EMTDC show that the constructed model has excellent generalization ability. Additionally, performance of the model is not affected by distributed generation and sampling rate. The proposed method does not use high-sampling devices and is cost-effective.