Research On Fault Diagnosis Of Rolling Bearing Based On LCD And ELM

Among all kinds of mechanical equipment,rolling bearings are regarded as one of the most commonly used parts,which are mainly popularized in military,armored vehicles and navigation systems.Once damage occurs in the operation project,it will affect the normal operation of the entire system,and even more security accidents will occur.For failed bearings,timely fault diagnosis and identification and troubleshooting are essential.As a result,the fault diagnosis technology came into being,allowing the equipment to monitor the status of the diagnostic object during operation or without disassembling the equipment,to determine whether it is in an abnormal or faulty state,and to analyze the cause of the fault.In order to improve the operating reliability and life of the equipment,prevent problems before they occur,and avoid the occurrence of failures.This paper proposes a fault diagnosis method of local feature scale decomposition combined with extreme learning machine for rolling bearings,and validates it with actual measured signals from engineering,so that it achieves the expected results.The main research content of this paper is as follows:(1)Since the monitoring signal collected during the state detection of the rolling bearing is weak,non-stationary and easily interfered by the noise of surrounding equipment,the vibration signal of the rolling bearing is processed by wavelet noise reduction and decomposed into several intrinsic scale components intrinsic scale components(ISC)through the local characteristic-scale decomposition(LCD),and select the ISC component that can best represent the significant fault feature,analyze its Hilbert envelope spectrum,and use it as the feature vector for bearing fault diagnosis.Through comparative experiments,it is proved that the features extracted in this research can better highlight the original impact signal of rolling bearings,and the recognition accuracy rate reaches 90.55%,both in terms of recognition rate and recognition speed are higher than traditional diagnosis methods,and the diagnosis results meet the actual engineering needs.(2)In order to overcome the problem of random initialization of the hidden layer parameters of the extreme learning machine,the kernel function is introduced on the basis of the extreme learning machine,and a particle swarm optimization kernel extreme learning machine mode is proposed for the selection of the inherent parameters of the kernel extreme learning machine.The recognition method uses the feature vector extracted from LCD decomposition and Hilbert envelope spectrum as the input of particle swarm optimization-kernel extreme learning machine(PSO-KELM).Experimental results show that the recognition accuracy rate reaches 99.44%.Compared with the particle swarm optimization extreme learning machine and the genetic algorithm optimization kernel extreme learning machine for diagnosis,this method can effectively improve the recognition accuracy and realize the fault identification of rolling bearings.(3)The Matlab user interface module is used to establish the LCD and PSO-KELM rolling bearing graphical user interface(GUI)fault detection system,which simplifies the fault diagnosis process,reduces equipment maintenance time and cost,solves the high professional requirements of the traditional rolling bearing fault diagnosis method for the operator,it improves convenience and human-computer interaction.