Research On Intelligent Warning Technology For Belt Conveyor Unit Based On Microphone Array

The belt conveyor unit is the core of the coal mining enterprise,and once its key components fail,it directly affects the safe operation of the whole system;and the belt conveyor unit is a very typical type of equipment,so it is important to study the fault warning of belt conveyor unit.The vibration sensor mainly is used by intelligent warning of mechanical equipment,the method is contact measurement,in some high temperature,high corrosion and other scenarios are not applicable.The acoustic signal has the advantages of noncontact measurement and wide monitoring range,which has good application prospects.The belt conveyor unit is monitored by the microphone array,the analysis of the collected acoustic signal to achieve unit fault warning and fault source location,to avoid major accidents and economic losses in enterprises.Based on this,the research work on intelligent warning technology for belt conveyor unit based on microphone array is carried out,and the specific research content is as follows:(1)In order to improve the warning accuracy of the belt conveyor unit,a detailed analysis of the structural composition,operation principle,operating conditions,sound field characteristics and typical fault mechanism of each part of the belt conveyor unit was carried out to determine that the transmission system adopts a variable operating condition warning technology based on a one-dimensional microphone array and the power system adopts a multichannel fusion warning technology based on a two-dimensional microphone array.(2)In order to solve the problem of difficult warning of variable working condition equipment due to insufficient utilization of working condition change characteristic information and correlation of multi-channel signals,an intelligent warning method for variable working condition by Long Short-Term Memory(LSTM)is derived and proposed by combining the composition and transmission path of acoustic signals of belt conveyor transmission system.Firstly,the relationship between the signals collected by each microphone,the relationship between the load variation and the signals collected by the microphones is analysed,and the fault warning principles for the roller are proposed;then,the signal characteristics of the first microphone are used to predict the signal characteristics of the other microphones,and the LSTM model and self-learning threshold are trained using normal data;finally,the warning model is tested by the fault data,and if an alarm is raised,and the KL scatter of each microphone is determined according to the fault source location.The validation of the proposed method is completed by simulating the signals and by simulating fault experiments using the roller failure test bench.(3)In order to solve the problems of low warning accuracy due to poor dereverberation effect and insufficient utilization of multi-channel signals in complex sound field,the composition and transmission path of belt conveyor gearbox and motor sound signals are analyzed,an intelligent multi-channel fusion warning method is proposed based on Weighted Prediction Error(WPE),Linearly Constrained Minimum Variance(LCMV)and Gaussian Mixture Model(GMM).Firstly,the multi-channel signals are de-reverberated by WPE with adaptive parameter selection;then LCMV beamforming is used for noise reduction and multi-channel fusion;and finally,the features of the fused signals are extracted to achieve intelligent warning of mechanical equipment based on GMM model.The validation of the proposed method is completed by simulating fault experiments using the rolling bearing test bench.(4)The parameters of the microphone array are determined according to the structure of the power system and transmission system of the belt conveyor unit of a coal washing plant,and the deployment of the monitoring system is completed;the acoustic signals of the belt conveyor unit at the site are collected and analysed to verify the effectiveness and reliability of the warning scheme proposed in this paper.