Statistical Distribution Analysis and Prediction of Time Between Failures in On-board Train Control Equipment

On-board train control equipment is the core equipment to ensure the safety and smooth operation of the train, the prognostics and health management for on-board equipment(OBE) has great significance. In this paper, aiming at the uncertainty of OBE failure and under the circumstances of less failure samples, puts forward that regard the whole OBE as a black box, the analysis and prediction of time between failures is conducted. Firstly, through the statistical analysis of the train operation fault log files, the sample series of time between failures(TBF) is obtained. The statistical distribution analysis is performed for the sample series, and the Weibull distribution model is established and tested, it shows that it has certain rules and can be predicted. Finally, TBF is used for short time prediction based on grey support vector machine, the accuracy of general and serious faults are 86.87% and 90.04% respectively by example verification, it shows that this method has good predictive effect, the prediction results have certain guiding significance for CBM.