Research On Fault Discovery Method Of Remote Belt Conveyor Driven By Patrol Information

As the key conveying equipment in the coal industry,the frame,idler and other parts of the remote belt conveyor are far away from the machine head.With the increase of running time,it is often accompanied by structural loosening,deformation and other faults,which is the main object of remote inspection.The traditional manual regular inspection is limited by physical strength and inspection distance,so it is often inefficient and it is difficult to avoid missed inspection.Although sensor node monitoring can record tens of thousands of node information,it has the disadvantages of many monitoring nodes and high layout cost.Therefore,this study chooses the non-contact detection method based on sound to collect the state information of the remote belt conveyor,and through the signal analysis technology to find out whether it is in an abnormal state in time.However,due to the influence of external environmental noise and vibration-noise coupling,the characteristics of the fault signal obtained are weak,so it is challenging to realize intelligent fault detection.In order to solve the above problems,an intelligent fault detection method of remote belt conveyor based on improved maximum correlation kurtosis deconvolution(IMCKD)and multi-channel convolution neural network(MCCNN)is proposed.Firstly,the firefly algorithm is used to search the two influence parameters of the maximum correlation kurtosis deconvolution in parallel,and the original audio signal is adaptively filtered to get the data source for discovery;then it is input to the MCCNN for feature learning and constantly update the network parameters;finally,the features are applied to the classifier recognition to realize the intelligent fault detection of the remote belt conveyor.In order to verify the feasibility and effectiveness of the method,firstly,the data collected by the mechanical fault comprehensive simulation test-bed are used to verify the algorithm.The results show that the IMCKD+MCCNN method can adaptively process the original audio signal,and the constructed MCCNN further improves the ability of the network to extract state features.Secondly,the long-distance belt conveyor system is used to verify the migration ability of the algorithm.The test results show that the IMCKD+MCCNN method can classify the fault types of remote belt conveyor rollers accurately and effectively.even under the strong background noise of SNR=-4dB,it still has an average fault identification rate of more than 90%,and can reach more than 96%in variable working conditions,and has good stability.Finally,with the help of MATLAB GUI,the proposed IMCKD+MCCNN is integrated and encapsulated,and a set of fault detection system software for detecting the running state of remote belt conveyor is developed,which provides end-to-end remote belt conveyor fault detection tools.The system test shows that the software can quickly and effectively distinguish the running state of the remote belt conveyor.