Electrical equipment sound identification method based on characteristic frequency band selection

The invention discloses an electrical equipment sound identification method based on characteristic frequency band selection. The method comprises the following steps of: acquiring and preprocessing sound time domain data, and converting the sound time domain data into a frequency domain by using Fourier transform; the method comprises the following steps of: randomly sampling a set number of samples by adopting a Metropolis-Hastings sampling method to establish a partial least square regression model; filtering the frequency band with the minimum weight of the absolute value of the regression coefficient by using a logarithmic attenuation function; selecting a frequency band with a set retention ratio by using weighted sampling, and selecting a corresponding frequency band with a relatively small root-mean-square error of interactive verification as a characteristic frequency band; and establishing a neural network, and identifying the sound state. According to the method, the space dimension of the training