Research on engine condition recognition based on wavelet time-frequency graph and multi-scale convolutional neural network
Aiming at the problem that the traditional method is difficult to accurately identify the non-stationary automotive engine audio signal,an engine condition recognition method based on wavelet time-frequency graph and multi-scale convolutional neural network is proposed.First,the original signal is transformed into a wavelet time-frequency graph through continuous wavelet.Secondly,the wavelet time-frequency graph is pre-processed uniformly.Finally,the processed image is input into the convolutional neural network to extract multi-scale features and classify them.This method effectively combines the wavelet time-frequency analysis with the image analysis capability of convolutional neural network,which has the advantage of pro-cessing nonlinear stationary signals.The recognition accuracy and robustness are better when the test set data speed is different.
NETL
NSTL
万方数据
