Application research on condition monitoring of Power-Shift Steering Transmission with hypersphere support vector machine

Abstract

The hypersphere support vector machine is an efficient method to distinguish the state of mechanism. The state of Power-Shift Steering Transmission(PSST) is studied using hypersphere support vector machine and Principal Component Analysis(PCA) with spectrometric oil analysis data. The fundamental of hypersphere support vector machine and PCA is researched. The influence of model parameters for performance of hypersphere support vector machine is analyzed. On the basis of training samples, the calculation and analysis of test samples are done. It has been proved that hypersphere support vector machine is suitable to judge the state of PSST. Moreover, the performance of hypersphere support vector machine can be improved greatly if abnormal samples are taken into account.