A new method of real-time reliability assessment based on SVR and Bayes algorithms

Real-time reliability assessment methods make it possible to asset system reliability on-line, which promote the development of the reliability research. A new method of real-time reliability assessment based on support vector regression (SVR) and Bayes is put forward in this paper. Firstly, principles of SVR and Bayes methods are particularly introduced, and then a model of real-time reliability assessment is set up. Finally, based on a selected degradation parameter-Small flow at high temperature (SFHT) of a hydraulic pump under experimental testing, process of real-time reliability assessment are successfully achieved, result shows that the model of real-time reliability assessment is effective. This method combines the real-time sample information with historical information, which is a good method for the real-time reliability assessment.