Research On Fault Diagnosis And Prediction Of Hydraulic Support

As one of the key equipment of modern fully mechanized mining face,the main function of hydraulic support is to effectively support the roof of mining area and provide safe working space.Because of the bad working environment,high working intensity,complex equipment structure and other factors,the hydraulic support fault occurs frequently.The long maintenance time seriously affects the production efficiency of coal enterprises,improves the mining cost of coal,and the fault maintenance is not timely or even causes safety accidents.In view of the current situation,this paper studies the fault diagnosis and fault prediction of the hydraulic support,realizes the rapid identification and location after the fault occurs,and predicts the operation status of the equipment one hour later.The main work of this paper is as follows:(1)The fault status of the hydraulic support is analyzed.First of all,this paper analyzes the current research situation of fault diagnosis and fault prediction,and finds that it is less in the application of hydraulic support equipment,which requires more theoretical exploration and practical application.According to the information collected from field research,combined with expert knowledge,it analyzes four types of common faults and their main manifestations of hydraulic support,and probes into the relevant causes to build a liquid Fault tree for common faults of press bracket.(2)The fault diagnosis of hydraulic support is studied.First of all,this paper determines the monitoring indicators of hydraulic support equipment,collects the operation data of four kinds of faults and reduces the dimension,reorganizes them into training sets and test sets,and constructs the classification model of support vector machine based on libsvm-3.23 toolkit in MATLAB.Then,according to the complexity and relevance of the fault causes,the historical records of hydraulic support fault causes are sorted into excel Table,structure learning and parameter learning of Bayesian network,using probability value to express the uncertainty relationship between fault causes.Secondly,based on the classification results of support vector machine as known evidence,input to Bayesian network,Bayesian network reasoning.Finally,the feasibility of the fault diagnosis model is verified by an example.(3)The fault prediction of hydraulic support is studied.The traditional "post maintenance" has not kept up with the pace of productivity change.With the coal mining equipment gradually entering the intelligent stage,it is of great significance to the fault prediction of hydraulic support.In this paper,the design of fault prediction method is carried out.The goal of the method is to predict whether the equipment has fault one hour ago and one hour later.The running state of the hydraulic support after one hour is used as the label to mark the running data at this time.Different samples(fault after one hour and no fault after one hour)are classified correctly by using the support vector machine method.The cross validation grid search method,particle swarm optimization algorithm and artificial fish swarm algorithm are used to optimize the penalty coefficient c and kernel parameter g in SVM,and the combination of parameters obtained by each method is selected to build the prediction model.By comparing the classification results of the test set,according to the classification accuracy,the artificial fish support vector machine is selected for the fault prediction of the hydraulic support.This thesis has 29 figures,23 tables and 89 references.