Research On Fault Diagnosis System Of Belt Conveyor Based On EEMD

With the development of port specialization and automation,port security is the premise of port operation,especially in the booming period of the fifth generation of smart ports.Belt conveyor is a bridge connecting land and sea,and it is an essential part of port operation.The failure of belt conveyor often occurs in the operation,which directly causes great economic loss and even damages the safety of human life.At present,the main way to solve these problems is manual inspection,which has the defects of low efficiency,low fault diagnosis rate and low feedback efficiency.For this reason,based on the actual situation of Rizhao Port,this paper analyzes and classifies the causes of the relevant faults of the belt conveyor,and then develops a fault diagnosis system of the belt conveyor based on the sound signal according to the port operation mode,mainly from the following aspects:(1)For the complex noise environment of the fault signal of the belt conveyor,a denoising algorithm based on the combination of Bayes theorem and wavelet threshold is proposed.The wavelet coefficients of the sound signal collected by the belt conveyor are obtained by wavelet transform,which are analyzed systematically.The likelihood ratio is formed by the prior probability and cost factor,and then the threshold is detected according to the likelihood ratio Threshold estimation determines the best threshold of wavelet threshold.Through experimental verification,the wavelet denoising algorithm based on Bayesian threshold can improve the signal-to-noise ratio of sound signal better.(2)aiming at the feature extraction of the fault signal of the belt conveyor,in order to reduce the modal overlap in the process of signal feature extraction,a method of combining EEMD algorithm with Hilbert transform is proposed.EEMD is used to decompose the denoised belt conveyor signal,and then the feature mode function(IMF)is extracted by Hilbert transform.(3)For the fault diagnosis of the belt conveyor,the signal features of the characteristic mode function(IMF)extracted from the empirical mode decomposition by the Hilbert transformation are taken as the control group,and the comparison group of the fault characteristic frequency domain extracted from the real-time collected signals is compared to realize the fault diagnosis of the belt conveyor according to the difference.In order to realize fault diagnosis better,the eigenvalues of each modal components obtained fromEEMD decomposition of acoustic signal based on Bayesian threshold de-noising are calculated and preliminary eigenvectors are formed.The appropriate eigenvectors are selected according to the clustering effect of fault modes in each mode,and the eigenvectors of different belt conveyor faults are obtained.The training set characteristics are calculated according to the Euclidean distance minimum method Vector and eigenvector of each fault can realize fault state recognition.Finally,the diagnosis results are compared with the standard of manual judgment to verify the accuracy of fault diagnosis and analyze the reasons that affect the accuracy.The feasibility of the system is verified by the experimental simulation analysis of the sound of multiple groups of belt conveyor collected in the field.At the same time,compared with the traditional manual fault diagnosis,the system reduces the processing delay time,greatly improves the fault diagnosis efficiency,directly reduces the continuous damage of the conveyor belt,and has an important contribution to reducing the economic loss of Rizhao Port and ensuring the life safety of workers.