Analysis of wavelet decomposition properties of wind turbine signal

Abstract

In order to improve the accuracy of judging the different working states of the mechanical vibration signal, the signal is decomposed into several frequency bands through the method of wavelet packet decomposition. Experimental results show that under different working states, mechanical vibration signals may have obvious differences in specific frequency bands, and this difference is beneficial to improve the accuracy of fault diagnosis. The experimental results also show that for some signals which are difficult to distinguish in a specific frequency band, they may still be difficult to distinguish even if these signals are extended to the entire frequency band. Wavelet packet decomposition that can divide a signal into several frequency bands may be more effective for improving the diagnosis rate of engine faults, Through the decomposition of different levels of the wavelet method, the diagnostic accuracy rate has increased by more than 20%.