Research On Belt Conveyor Fault Diagnosis System Based On Sound Processing Technolog

As an important transportation equipment in the port transportation industry,the operation status of the belt conveyor directly affects the transportation efficiency of the port.Due to longterm high-load operation,belt conveyor is prone to various failures,causing huge economic losses and even endangering the life safety of workers,so it is necessary to monitor the operating status of the belt conveyor.The traditional manual inspection can no longer meet the current production needs due to inefficiency,unreliability and other shortcomings,and the sound signal of the belt conveyor contains a large amount of operating status information.Therefore,this paper develops a belt conveyor fault diagnosis system based on sound processing technology,which mainly includes the following:(1)Aiming at the problem of noise in the sound signal of belt conveyor,an adaptive wavelet threshold function denoising algorithm is proposed.Through theoretical analysis,the continuity and convergence of the function are proved,the problems of discontinuity and deviation of the traditional wavelet threshold function are solved,and the segmented threshold is adaptively selected according to the correlation coefficient to denoise the sound signal of the belt conveyor.Experiments show that this method can effectively remove noise and improve the signal-to-noise ratio of the signal.(2)In order to extract the characteristic parameters that can characterize the sound signal of the belt conveyor and improve the fault diagnosis efficiency,a simple energy normalized cepstral coefficient(WP-SPNCC)based on wavelet packet transformation and CNNs feature extraction based on deep learning were carried out on the denoising signal.The wavelet packet transform is used to replace the Fourier transform in the traditional SPNCC feature extraction,and the WPSPNCC features of the belt conveyor are obtained by Gammatone filter filtering and power-law nonlinear processing.The sound signal of the belt conveyor is converted into a spectrogram and input into the built convolutional neural network for deep learning feature extraction.Experimental verification shows that the extracted WP-SPNCC features and deep learning features can effectively characterize the original signal.(3)The support vector machine(SVM)model is built based on WP-SPNCC features,and the bidirectional gated recurrent unit(Bi-GRU-AM)network model is built based on the attention mechanism based on deep learning features.The extracted WP-SPNCC features are input to SVM for fault diagnosis,and the network parameters are optimally designed to obtain SVM models with high recognition rate.The Bi-GRU-AM model is built to make the state value of the current moment have complete time information,and the attention mechanism is used to assign weights to different features,and the extracted deep learning features are input into the model for training and the network model parameters are optimized,and the Bi-GRU-AM network model with high recognition rate is obtained.The WP-SPNCC features and deep learning features are spliced and fused,and the dimensionality reduction and normalization processing by principal component analysis method are input into the Bi-GRU-AM model to obtain the fault diagnosis results after feature parameter fusion.Experimental results show that the recognition rate of Bi-GRU-AM fault diagnosis model based on feature fusion algorithm is as high as 97.5%.