Research And Development Of Fault Diagnosis Expert System Based On Rough Set-Neural Network Of Hydraulic Press

Large CNC hydraulic press is the highly automated equipment,which typicallycombined with the mechanical,electrical,hydraulic and gas.Because of correlationand levels among systems or components,the relationship between the faultsymptoms and the fault reasons is complex and changeable,so it is very hard tojudge.The diagnosis not only waste time and manpower，but also seriously affectthe enterprise benefit,and even cause accidents. It is the current research key anddevelopment trend to put forward a simple and efficient diagnosis method and todevelop fast and accurate fault diagnosis system,which has great practicalsignificance. This paper studies and discusses fault diagnosis technique ofRZU2000HM hydraulic press and develops the fault diagnosis system.1. Combining the working principle and fault mechanism of the hydraulicpress RZU2000HM, collecting its occurrence breakdown case, analyzing its failuremode and fault feature, discussing the reasons.This paper introduces in detailed thehydraulic cushion control system and the Ascension clamping controlsubsystem,and takes the hydraulic cushion control system as the specific diagnosisobject to analysis the fault feature and make their state testing point for diagnosisof information.2. Combining Rough set, neural network, expert system together,this paperputs forward the model structure of fault diagnosis expert system based on roughset neural network. This structure uses rough reduce the neural network faultsamples for removing redundant attributes and optimizes the structure of the neuralnetwork for Comprehensive using the knowledge of the sample,and therefore it canimprove the neural network learning efficiency and generalization ability.Thereasoning machine of fault diagnosis expert system is made up with knowledge andrough set neural network.3. According to the proposed model structure of the diagnosis, using mixedprogramming technique using VC and Matlab based on COM, and the ACCESSdatabase technique to develop fault diagnosis system. Firstly Constructingknowledge base: Through training neural network to construct the knowledge baseof rough set neural network reasoning machine, take the fault diagnosis decisiontable as a knowledge reasoning machine’s knowledge base. And then Fault reasoning for verification: Comparing the inferred conclusion with the actualsituation, it can prove that the diagnostic methods and fault diagnosis platform inthis paper are both valid.