Research On Intelligent Fault Diagnosis Of Coal Mine Belt Conveyer

The coal mine belt conveyor,as one of the key transportation equipment of the coal mine production system,is known as the "mine artery".Once it fails,may bring immeasurable damage to property and even casualties.How to carry on more accurate analysis to the belt conveyer's breakdown to obtain the effective fault decisionmaking model,then analyze the actual running state with this model to timely find out the fault symptom and make early warning,has become the urgent problem to be solved in the fault diagnosis of coal mine belt conveyor.There are many kinds of faults in coal mine belt conveyor,and the symptoms may be mixed,which seriously affect the timeliness and reliability of fault diagnosis.In this paper,a fault diagnosis technology based on Rough Sets and Neural Networks is used to solve the problem.Firstly,the feasibility and effectiveness of Rough Sets(RS)and Levenberg-Marquardt optimized Back Propagation Neural Networks used in fault diagnosis of coal mine belt conveyor are deeply studied.Then,on this basis,a fault diagnosis model based on APRSNNS(Approximate Precision Rough Sets Neural Networks)is built,which has higher classification accuracy,stronger fault tolerance and lower time complexity.After in-depth study of the coal mine belt conveyor failure mechanism,this issue runs analysis on the essential characteristics of the fault selection.In order to make the rough sets reduction algorithm have better fault tolerance,reduce the inevitable error interference in practical application and get the optimal attribute sets,based on the full understanding of rough sets attribute reduction algorithm,an improved reduction method is proposed,which pull-in an error factor ?.According to the characteristics of the fault diagnosis of conveyor and the rough sets can only deal with discrete data,a dispersing method based on SOM and relative attribute dependency is proposed to disperse attributes and meanwhile optimize the key symptom attributes.Based on the analysis of BP algorithm,the L-M algorithm is introduced to increase the speed of training.And an improved fault diagnosis model of coal mine belt conveyor based on APRS and L-MBP is proposed.The minimal attribute sets and all attribute sets are entered L-MBP neural networks as training samples,the comparison shows that,within minimum target error,the APRS-LMBP model diagnostic is real-time,with high prediction accuracy.Compared with the standard BP,the results also show that this paper's model has more advantage in the diagnosis,can fully remove the redundant information,accelerate the training speed of the networks and reduce the error of diagnosis.That's to say,it's a more efficient fault diagnosis model for large scale automatic production system.