Fault Diagnosis Of Gear-box Based On Particle Swarm Optimization And Support Vector Machine

Gear-box is an important part of mechanical equipments, and the research of monitoring and fault diagnosis has great guidance value. In this thesis, some research with particle swarm optimization and support vector machine will apply to the fault diagnosis of the gear-box.This paper firstly considers the JZQ250 gear-box as the study object. Based on the fault mechanism and fault style of the gear-box, the experiment flat is established. The experiment project on condition testing and fault diagnosis of the gear-box is designed, and the vibration signals of the gear-box are tested.Secondly, this thesis introduces the basic principle and the development of particle swarm optimization, and analyses how to select the PSO parameters. In order to verify the performance of the standard PSO algorithm, this paper selects four standard functions to do some simulation experiments. The experimental results show that the standard PSO algorithm can find the optimal solution quickly in the condition of satisfying the convergence criteria. Based on the VC dimension and and the Structural Risk Minimization, the basic theory of the support vector machine is discussed, and Least squares support vector machine has great application. The Iris data in the UCI is quoted to do some simulation experiments. The optimization of SVM model parameters by standard PSO algorithm shows that the SVM has great superiority on pattern recognition.Finally, a method to do SVM parameter optimization that uses PSO algorithm is proposed. Binary-fault and multi-fault SVM model is set up. Combined with the gear-box experiment data, the model can find the optimal parameters quickly. And fault diagnosis comes true successfully. The effectiveness of the SVM algorithm on small sample size is verified.