A study on time series models and criterion rules based on condition monitoring of the tracking test system

Due to lots of measuring positions, long duration, complex test objects and various data types in the tracking test system, there is a badly need to find a simple, effective and convenient analysis method to process the test data and figure out the potential rules. This paper finds a method of identifying the optimal time-series model automatically, through studying different models and criterion rules. This method doesn´t need to do any artificial hypothesis to the evolution rules of parameters, but estimates the changing rules of parameters from data adhering to the thought of “let the data speak for themselves”, so it achieves the self-explanatory of the data. As a conclusion, the accuracy of the method introduced in this paper is verified through simulating to the gear acceleration data measured from EMU tracking test and the method is effective and convenient.