Neural Network Faulty Line Detection Method in Small Current Grounding Systems Based on Rough Set Theory

摘要

It is a long-term issue about faulty line detection of single-phase grounding faulty in the small current grounding systems. If only one faulty line detection method is used, faulty information is analyzed and used partially which is not enough for faulty line detection; and there are different conditions for every method. In order to compensate the shortcoming of one method, a fuse method is used to ensure the reliability of the line detection result. First the data set was preprocessed by Rough Set theory, so the redundancy information was thrown off, and the simplified data set was obtained. Second the neural network was designed and trained by the simplified data set. At last, fusing those detection results, a better faulty line detection result was reached. Simulation results by EMTP show that the method of the faulty line detection is valid and the method has some study value and will be used in distribution systems.