Research On Belt Conveyor Fault Diagnosis Based On Multi-information Fusion

Belt conveyor plays an extremely important role in port coal transport process.Belt conveyor is in a state of continuous transport with high load for a long time,and often suffers such faults as stuck roller,torn belt,damaged drum and so on.At present,a single sensor is mainly used to detect the fault information of the belt conveyor,but the single sensor has the problems of fuzziness,incomprehensiveness and low accuracy in the process of information collection.In view of this problem,this paper selects the multi-sensor information fusion technology to carry out the fault diagnosis research of belt conveyor.The main research contents of this paper are as follows:By analyzing the common faults of the belt conveyor,two types of sensors,sound sensor and infrared thermal imager,are selected to collect the information of the belt conveyor in actual operation.According to the characteristics of acquired sound signals and infrared images,a belt conveyor fault diagnosis system based on multi-information fusion is designed.Lab VIEW software is used to receive and process information,and Lab VIEW is used to call MATLAB programming software to realize multi-source fault information processing.The sound signals and infrared images of belt conveyor in different running states collected on site are taken as the basis of analysis.Due to the noise of different sizes in the collected sound signals,the improved Boll spectral subtraction method with simple operation and ideal effect is adopted to reduce the noise of the sound signals.Then,the wavelet packet decomposition algorithm is selected to analyze the energy spectrum of the denoised sound signal,and the energy of each subband is used as the detection feature vector.Aiming at the problem that infrared image will be polluted by noise in the process of imaging and transmission,an improved mean adaptive filtering algorithm is chosen to de-noise infrared image by analyzing the shortcomings of traditional median filtering and adaptive median filtering.Secondly,target segmentation is carried out on the infrared image after denoising.According to the characteristics of the infrared image of the belt conveyor,a target segmentation algorithm combining the salience region target segmentation method and the edge detection algorithm is proposed,which can effectively segment the contour of the fault part of the belt conveyor.In order to meet the real-time requirements of the fault diagnosis system,Hu invariant moments with less computation were used to extract the shape features of the fault area,and seven invariant moments constructed by Hu moments were used as detection feature vectors.The fault diagnosis model of belt conveyor information fusion was constructed by combining VNWOA-BP neural network with improved D-S evidence ethics.Aiming at the limitations of BP neural network,the VNWOA-BP neural network was optimized by von Neumann whale algorithm,and the VNWOA-BP neural network model was established to diagnose the fault of belt conveyor.By analyzing the experimental results,the validity of VNWOA-BP neural network to the fault diagnosis method was verified.Aiming at the shortcomings of traditional D-S evidence theory synthesis rules,Deng Yong’s synthesis rules were adopted to improve D-S evidence theory,and decision-level fusion experiments were carried out on the output vectors of VNWOA-BP neural network.Obtain the final result of fault diagnosis of belt conveyor.The experimental results show that the multi-information fusion fault diagnosis model can effectively improve the accuracy of fault diagnosis and enhance the reliability of fault diagnosis system.According to the above research,the process of the belt conveyor fault diagnosis system is combed.Through the analysis of the belt conveyor fault diagnosis example,it can be seen that the multi-information fusion belt conveyor fault diagnosis system is effective and feasible.