Quantitative recognition of rolling element bearing fault through an intelligent model based on support vector regression

The research in bearing fault diagnosis has been attracting great interest in the past decades for its important role in the rotating machinery. Development of effective monitoring and fault diagnosis methods to prevent failures that could cause huge economic loss timely is necessary. Bearing faults with different degrees differ from each other in the aspects of signal amplitudes, impact intervals, etc. Besides, the whole life of the bearing is also a developing process for some sensitive features related to the fault trend. Most of the work in machine fault diagnosis focus on the decision of fault existence and thus ignore the developing process of the faults. Hence, to investigate the fault degree is of great meaning for timely maintenance action. In this paper, a novel intelligent method based on Support Vector Regression (SVR) to conduct rolling element bearing fault size quantitative recognition is proposed. This analysis constructed an intelligent nonlinear model with input feature vectors extracted from raw signals and defect sizes as outputs. Through validation of experimental data, the results indicated that the bearing fault size could be effectively and precisely recognized.