Electric precipitation spark discharge identification method based on BP neural network

The invention discloses an electric precipitation spark discharge identification method based on a BP neural network, and the method comprises the following steps: collecting a sound signal, and obtaining PCM audio data; preprocessing the audio data; extracting three characteristic parameters of MFCC, short-time energy and short-time zero-crossing rate of the audio data; establishing a BP neural network spark discharge audio identification model; taking the three characteristic parameters of the acquired multiple groups of sound signals as sound samples; training and testing the BP neural network spark discharge audio identification model; and extracting three characteristic parameters of the acquired sound signal to be recognized, and inputting the three characteristic parameters into theBP neural network spark discharge audio identification model meeting the requirements to complete identification of the electric precipitation spark discharge sound. According to the method, a soundidentification method is a