A Study on the Localization of Partial Discharge Using Ngccphat CNN

Current methods for detecting partial discharge sound sources in transformers are not very accurate and are susceptible to interference from environmental factors and calculation algorithms. Therefore, this study proposes a deep learning method of convolutional neural network to locate partial discharge in transformers. The generalized cross-correlation positioning algorithm is utilized, along with a deep learning framework, to compute the cross-correlation delay between audio signals. This enables the identification of the source location for partial discharge sounds. Through the audio data set collected by the simulated microphone array of the partial discharge signal, the partial discharge positioning calculation model was trained, and finally the trained calculation model was used for positioning testing, which verified the accuracy of the partial discharge sound source positioning algorithm. Experimental results show that the network used not only completes the localization of partial discharge sound sources, but also improves the accuracy of sound source localization.