Phase modifier partial discharge detection method based on TCN network

The invention discloses a phase modifier partial discharge detection method based on a TCN network, and belongs to the technical field of TCN networks. The invention belongs to an acoustic feature detection method for partial discharge of a phase modifier. According to the method, a time convolution network is combined with causal convolution and cavity convolution to carry out feature extraction on acoustic emission signals, correlation between sound signals triggered by partial discharge and parameters such as voltage, current and equipment temperature is fully considered, and abstract features learned by the network are combined with other information to further carry out feature learning. According to the method, feature fusion is carried out on acoustic emission signals, voltage, current, equipment temperature and other parameters, further feature extraction is carried out on fusion information, data information can be fully utilized, information loss caused by manual feature selection is avoided, the TC