Abstract: Aiming at the problem of discharge mode monitoring of polluted insulators in high voltage transmission lines,a new one-dimensional convolutional neural network structure (1D-CNN) is designed for monitoring the discharge patterns of fouled insulators in high-voltage lines, and a fouled insulator discharge pattern monitoring method based on acoustic emission signals and 1D-CNN is proposed. After preprocessing the acoustic emission signals collected in the laboratory under different discharge states, the convolutional neural network is used to perform adaptive feature extraction and feature dimensionality reduction on the discharge signal samples to reduce the training model parameters and computation, and finally the Softmax function is used to classify the prediction results. The recognition results show that the model can achieve a recognition rate of more than 99.84%, which effectively solves the process of manual preprocessing of data in the traditional insulator fouling degree monitoring method and can be effectively applied to the task of fouling insulator discharge pattern monitoring.

Key words: discharge of polluted insulator, convolutional neural network, acoustic emission signal, fault diagnosis, deep learning