Open AccessA Recognition Method for Time-domain Waveform Images of Electric Traction Network Overvoltage Based on Deep Learning
Author(s) -
Jia Junyi,
Mingli Wang,
Qi Wang
Publication year - 2021
Publication title -
iop conference series. earth and environmental science
Language(s) - English
Resource type - Journals
eISSN - 1755-1307
pISSN - 1755-1315
DOI - 10.1088/1755-1315/645/1/012073
Subject(s) - waveform , time domain , overvoltage , computer science , convolutional neural network , preprocessor , artificial intelligence , domain (mathematical analysis) , frequency domain , pattern recognition (psychology) , test set , artificial neural network , computer vision , engineering , mathematics , voltage , telecommunications , electrical engineering , mathematical analysis , radar
A large amount of time-domain waveform images of overvoltage of electric traction network are collected during the on-board test of EMU. The traditional time-domain or frequency-domain analysis and classification method can’t be directly applied to time-domain waveform images. In order to analyze these images, firstly, image preprocessing is carried out. Then, the data set of time-domain waveform images is established. Finally, the CNN (convolutional neural network) classification model is trained based on data set. The results show that the sensitivity of recognition is above 81%, and this method is suitable for the time-domain waveform image of overvoltage of electric traction network.