Identification of Fault Causes of 110kV Cable-Overhead-Hybrid Transmission Lines Based on Dual-Channel Synchronous Waveform Identification Network

Accurate identification of fault type is great significance to the site operation and maintenance of 110kV Cable-Overhead-Hybrid transmission lines. Starting from the real-time data collected by field traveling wave fault location device, this paper propose a identification method of fault causes of 110kV Cable-Overhead-Hybrid transmission lines based on Dual-channel synchronous waveform identification network. Firstly, the fault waveform law of 110kV Cable-Overhead-Hybrid transmission lines is fully analyzed with real-time collected data, and the input waveform is converted into time-frequency image with short-time Fourier transform to establish the input amount of the whole system. Secondly, Dual-channel Higher-order feature extraction network formed by the cascade of ConvNeXt network and CBAM attention mechanism is constructed to accurately obtain the detailed features of the corresponding time-frequency map of each channel, and Adaptive feature fusion network is constructed to realize the combination of cross-channel fault features. Thirdly, reconstruct the full connection layer based on MFM unit,and Classify and identify line faults combined with the Softmax classifier. Finally, the overall accuracy of the proposed algorithm is about 94.3% through the data test. And from the accuracy, precision, recall and F1 score for comparison, the proposed algorithm is better than RF, CNN, RNN, LSTM, ConvNeXt five models.