Research on Noise Reduction Algorithm of Belt Conveyor Sound Signal Based on Improved Empirical Wavelet Transform
This paper proposes an improved empirical wavelet transform algorithm to extract effective sound of belt conveyor in complex working environment.The improved method reconstructs the Fourier spectrum based on the spline interpolation and calculates the local power of each frequency band,the adaptive spectrum segmen-tation is optimized by threshold value.Then,the sound signal is reconstructed according to the decomposed AM-FM signal and from which the characteristic parameters are extracted and the support vector machine is input-ted to verify the feasibility of the fault classification algorithm.Through the simulation experiment of the belt con-veyor sound signal collected in the port is conducted,the results show that the improved algorithm can not only overcome original defects,but also improve the denoising ability of sound signal by about 25％,and the average accuracy of fault diagnosis reaches 96％.
belt conveyorimproved empirical wavelet transformfeature extractionfault diagnosis
NETL
NSTL
万方数据
