Research On Health Status Evaluation Method Of High-speed Train Running Gears Based On Semi Quantitative Information

The running gear system is the core equipment of the high-speed train,which plays a vital role in ensuring the safe and stable operation of high-speed train.With the rapid development of modern industrial technology,the health management of the running gear system has become a research hotspot.Health status assessment and health status prediction are important contents of the health management,and play a great role in the health management of the running gear system.Due to the internal complexity,strong correlation and nonlinearity of the running gear system,there is no sound and scientific theoretical scheme for health state evaluation and prediction.It is of great significance to study this urgent engineering problem.For the sake of solving the problem that it is difficult to model due to the complex structure and failure mechanism of the running gear,so as to achieve the goal of truly reflecting the health status of the running gear system at the current time and precisely predicting the health status of the system at the future time.This paper studies the health status evaluation and health status prediction modeling method of running gear system based on semi quantitative information(combined with expert knowledge and a small amount of monitoring data).It effectively solves the problem of lack of data in the running gear system and improves the accuracy of health status evaluation and prediction of the running gear system.The specific contents include the following two points:(1)Aiming at the problem that it is difficult to establish an accurate evaluation model of the health state of the running gear system,this paper constructs a health state evaluation model based on belief rule base(BRB)through semi quantitative information method.The rules in the model are fused by evidential reasoning(ER)algorithm to realize the reasoning of knowledge and make the expression of knowledge closer to reality.In the process of modeling,too many attributes will lead to rule combination explosion.Therefore,canonical correlation analysis(CCA)algorithm is used to select features to reduce the complexity of the model.In the evaluation process,in order to improve the evaluation ability of the model,the project covariance matrix adaptation evolution strategy(P-CMA-ES)is introduced to optimize the initial parameters given by the experts.The experimental results show that this method can accurately evaluate the health state of the running gear system.(2)Aiming at the problem that it is difficult to establish an accurate prediction model of the health state of the running gear system,taking into account the dynamic characteristics and real-time requirements of the running gear system,this paper constructs a real-time health state prediction framework based on multi-layer BRB model.Firstly,on the basis of sampling data and expert knowledge,a multi module time series prediction model is constructed from a complete feature set.Secondly,according to the time-varying characteristics of the matching degree between the feature and the system in the actual working conditions,the gray correlation analysis(GRA)algorithm is used to realize the priority scheduling of the feature,that is,the feature with high priority is used as the input of the next BRB.In order to improve the accuracy of model evaluation,recursive expectation maximization algorithm(EM)is used to optimize the input parameters.The experiment shows that the framework can effectively predict the health state of the running gear system.