Research On Rotating Machinery Fault Diagnosis Based On Vibration Signal Analysis

Vibration signal analysis is widely used in fault diagnosis of rotating machinery. Faultdiagnosis system is consisted of three parts of vibration signal acquisition, the fault featureextraction, and operating status recognition. In this paper，the gearbox is the research object.The fault features include time-domain characteristics, frequency-domain features andnon-stationary time-frequency domain features. They are extracted by signal processingmethod based on vibration signal analysis. The operating status is classified and identified bysupport vector machine.The key step of fault diagnosis of rotating machinery is the fault feature extraction. Inthis paper, the vibration signal and the fault signal characteristics of the gear are analyzed. Thefault diagnosis method is briefly introduced from time domain, frequency domain andtime-frequency domain. This paper focuses on Hilbert-Huang transformation algorithm.Based on this, a new algorithm is proposed to inhibit the end effects which used in theempirical mode decomposition (EMD) to reduce the error. The instantaneous frequencyspectrum parameters are obtained by the normalized Hilbert transform. From the simulationresults, the original signal components is decomposed more effective by the improved EMD.And the instantaneous frequency spectrum parameters can be used as the fault feature of gearstate.In order to improve the recognition correct rate and solve the small sample learningproblem, the immune cat swarm optimization improved algorithm (ICSO) is proposed tooptimize support vector machine in this paper. The population is divided into searching groupand tracking group. The clonal expansion operator is used for local search in the searchinggroup. The number of mutation individual is adjusted according to the fitness value. Thedynamic vaccine extraction and vaccination operator are used for global search in the trackinggroup. The descendant population is updated through the balance of concentration operatorand selection operator. The penalty parameter and kernel parameter of the Support VectorMachine (SVM) is optimized by the ICSO. ICSO-SVM is used for fault classification andidentification. The simulation results show that, the ICSO algorithm has higher convergence.Compared with other swarm intelligence algorithm, the ICSO-SVM algorithm has higherclassification accuracy and generalization ability.The gear fault diagnosis system is implemented on the MATLAB platform. The runningstate of the gear box can be accurately diagnosed in the system.