Research On Personalized Diagnosis Methods Driven By Finite Element Method Numerical Simulation For Mechanical Faults

In recent years,a large number of researches have been carried out in the field of mechanical engineering at home and abroad to diagnose the damage faults of the shaft,gear and bearing of the mechanical system,and many high accuracy and reliability diagnosis methods based on intelligent diagnosis have been proposed,such as artificial neural networks(ANNs),extreme learning machine(ELM),clustering analysis,support vector machine(SVM),etc.,which provide the technical basis for the damage identification of complex mechanical system.However,the damage fault feature of actual running equipment is weak and often submerged due to the heavy back ground noise of mechanical system.Moreover,the fault monitoring and diagnosis needs the personalized data of a certain mechanical equipment,while the laboratory data and the actual equipment operation data are not from the same mechanical equipment,which can only be used to study the general principle and development rule of the fault phenomenon.Even for the same type of mechanical equipment,due to the difference in assembly,installation and operation conditions,its working condition is also different.Therefore,it is necessary to obtain a large number of different types of fault samples which can reflect the actual operation conditions based on the operation status of the actual equipment.How to obtain a large number of fault samples reflecting the actual operation state of mechanical system is the key to achieve personalized diagnosis and engineering application by using above excellent mechanical fault diagnosis methods.In this study,numerical simulation analysis method is used to establish high-precision simulation analysis model and high-performance simulation analysis of mechanical system,obtain a large number of different types of fault samples reflecting the actual operating conditions,make up for the defects in the existing high-performance fault diagnosis methods,which are lack of fault samples or difficult to obtain,and explore the new way of mechanical fault personalized diagnosis under specific working conditions.This paper carries out the following contents:Aiming at the problem of bearing fault type identification,a kind of classification method based on finite element method(FEM)simualtion driving SVM is proposed to form the framework of bearing fault personalized diagnosis.Through the FEM simulation,the fault samples are obtained and used as the training samples of SVM,classify the measured signals(test samples)and judge the specific fault types of the test samples,so as to realize the personalized diagnosis of bearing fault.Similar to bearing fault diagnosis,a personalized diagnosis method based on FEM simulation driving ELM,and a personalized diagnosis method based on FEM simulation driving convolution neural network(CNN)are proposed respectively for the fault diagnosis of gear transmission system and rotor system.The experimental results show that the personalized diagnosis method proposed in this study has a good effect in the fault diagnosis of common mechanical parts such as bearings,gears,rotors,etc.,can effectively analyze the service performance of mechanical systems and prevent major accidents,which has important academic significance and application value.