High-speed Train Tracking Test And Researches On Time-series Model Of Conditon Monitoring

With the vigorous development of the high-speed railway, the condition monitoring for the security service becomes more and more important. In order to guarantee the security of the train in operation, a lot of studies have been done to the security service condition at home and abroad. At present, most of the trains in the developed countries are equipped with an on-board real-time monitoring system. And they are continuously doing studies on the real-time monitoring to the high-speed train, but their core technology and the methods which have a guiding significance have not been published to us. The studies on the high-speed train condition monitoring are still in its primary stage in our country, and the most previous studies were focused on the influence of the structure, properties and the relationship between the wheel and rail to the key system. However, the studies on the influence of operating condition to the security service condition were very lack.In order to analyzing systematically the influence of different conditions on the service performance of high-speed trains, a long term monitoring has been done to the parameters of the key system, based on the high-speed train. Due to the large amount of the tracking test data and the sort of types, the traditional determination methods in monitoring system based on threshold value have not been unable to meet the needs of the real-time dynamic analysis. Through studies on the time-series analyzing to the massive data, analyzing on the different criterion function, and comparing the prediction error of different types of time-series models, this paper gives conclusions as follows:Through the analyzing studies on the time-series prediction error to the massive data, this paper figures out the best time-series length is2000, when the prediction error of the prediction model of the time-series model is the minimum.Through comparing the different types of time-series model in different conditions, this paper gives the optical type and order of time-series model in different conditions.Through analyzing studies on the prediction error of optical time-series model and the change trend of the time-series parameters, this paper gives the general time-series model in different conditions.Due to the studies on the time-series model is in the primary stage, this paper suggest that it’s necessary to study deeply on the penalty factor in the time-series automatic modeling theory. In that way, this theory can be applied to the condition monitoring of the high-speed train better, so that it can do real-time monitoring to the high-speed train and forecast the data in real-time and warn.