Identification Method of Dehydration Stage of Washing Machine based on the Fusion of Sound and Vibration signals

The national standard GB/T 4288-2018 requires the comprehensive detection of the washing machine state from the vibration performance, noise and other parameters, and proposes a state identification method in dehydration stage based on the feature fusion of vibration signal empirical mode decomposition (EMD) and Sound signal Mel frequency inversion coefficient matrix (MFCC). Firstly, the signal to noise ratio differential threshold method was used to extract the stationary segment of the highest speed in the dehydration stage, and EMD was performed on the vibration signal, the information entropy of the power spectrum of each order intrinsic mode function (IMF) was calculated, and the Characteristic coefficient matrix of The Mel frequency cepstral coefficient (MFCC) was extracted from the sound signal. Secondly, kernel principal component analysis (KPCA) was used to obtain some principal components. Finally, the fusion features are recognized by short - and long-term memory network (LSTM). Experimental results show that EMD energy entropy and MFCC feature fusion classification model is superior to single feature and can significantly improve the identification accuracy.