Research On Fault Diagnosis Of Shearer Hydraulic Heightening System Based On Rough Set And Neural Network

Shearer is the key equipment of coal mining.Hydraulic heightening system is the main part of shearer prone to failure,its role is to achieve the height of the rocker arm and cut coal,once the failure,seriously affect the safe operation and work efficiency of shearer.Therefore,the rapid and effective fault diagnosis of the hydraulic heightening system can not only improve the economic benefits of enterprises,but also prevent the occurrence of major accidents.There is a complex nonlinear mapping relationship between fault symptoms and fault types in the hydraulic heightening system,which has many causes and uncertainties.In this paper,a fault diagnosis method combining rough set and RBF neural network is proposed,which reduces the input characteristic dimension,removes redundant information,simplifies the neural network structure,reduces the network training time and computation,and greatly improves the diagnosis efficiency.The main research contents of this paper are:First,on the basis of in-depth analysis of the shearer’s structure and working principle,the model of the shearer’s hydraulic heightening system was established by using simics software in Matlab environment with the MG750-1940 WD shearer as the research object.Through the analysis of the working principle and common faults of the hydraulic leveling system,failure simulation is carried out for typical faults such as leakage fault in hydraulic pump,stuck fault of electro-hydraulic directional valve and leakage fault outside the hydraulic cylinder,and the influence of different faults on the whole hydraulic system is studied and analyzed.Secondly,the rough set-RBF neural network fault diagnosis methods to select coal mining machine hydraulic system in high flow rate,pressure,velocity,displacement,such as feature vectors as the fault symptoms,in a typical fault set up the original fault decision table for fault type,the rough set theory system is applied to carry out isometric dispersal treatment on the original fault data set ofthe hydraulic heightening system,and the attribute reduction is carried out with the reduction algorithm of Genetic algorithm in ROSETTA software.remove the redundancy in the input information,the connotative knowledge and potential rule,get the minimum condition attribute sets and rules.To deal with the data set as a RBF neural network training samples,according to minimum condition attribute set initial RBF neural network topology structure,using gaussian function as the radial basis function,with gradient descent algorithm for network learning algorithm,through the network training system established hydraulic fault symptoms and fault types of mapping relation of fault diagnosis using the Python programming language.Finally,the Python programming language is used to analyze and compare the two methods.The first is the fault diagnosis method of RBF neural network.The normalized original fault data set is imported into RBF neural network to realize the fault diagnosis of the hydraulic heightening system.The second is the fault diagnosis method of rough set-RBF neural network.The fault data set after rough set is discretized and reduced is used as the input of RBF neural network to realize the fault diagnosis of hydraulic heightening system.Simulation results show that the performance of rough set-RBF neural network is better than that of simple RBF neural network under the same target error,its structure is simpler,the convergence speed is faster,the network learning efficiency and fault diagnosis accuracy is higher,and it has a good practical application effect in the shearer hydraulic heightening system.