Study on Fault Identification Method of Transmission Lines Based on an Improved CNN

When a transmission line fault occurs, the fault traveling wave data contains the occurrence location information of the fault point, which implies the characteristics of different fault reasons rise from specify location. Fast and accurate positioning and identification of the variable and random faults is difficult and complicated. According to this requirement developing fault identification system just as objective of this study. Firstly, applying wavelet transform analysis for the fault in current data, the data preprocessing filters most of the additive interference burr signals, and afford clean data to realizes fast fault positioning. Then, on the basis of data cleaning and filtering processing, this paper studies the characteristics and main reasons of ten typical faults related to thunder lightning and common sources of high voltage transmission lines. The improved. The test results show that the improved CNN based on batch normalization and hierarchical standardization can effectively handle the nonlinear fault waveforms, and achieve higher recognition speed and identification accuracy in common cases.