Fault Diagnosis of Diesel Valve Train Based on Genetic Optimization One-Class Support Vector Machine

Aiming at the problem of model selection criteria of standard OC-SVM algorithm and synchronous optimization of multiple model parameters, an automatic model selection method based on the adaptive genetic algorithm is proposed to realize automatic and fast optimization of model parameters. The proposed method is applied to fault detection of diesel valve train. Firstly, numerical simulation results verify that the model parameters (v,s) influence on OC-SVM is integrated. Secondly, the largest AUC value of ROC curve analysis is selected as the evaluation criteria, and Gaussian bandwidth coefficient s and control parameter v are optimized by the use of adaptive genetic algorithm synchronously. Finally, genetic optimization OC-SVM model is applied to fault diagnosis of diesel valve train. The results show that OC-SVM model with genetic optimization is easier to obtain good performance than OC-SVM model with artificial specified parameters.