Research On Life Prediction And Maintenance Decision For Inertial Unit Of Train Positioning System Based On Copula Theory

High-speed railway is the backbone of modern public transport.It has become the main means of transportation for the masses with its advantages of high speed and high capacity.Train Operation Control System is an important equipment to ensure the safety and high speed of trains.Among them,the Train Positioning System is the key equipment to provide speed and position information for the Train Operation Control System.At present,with the development of “Next Generation Train Control System” and the rapid growth of satellite navigation and inertial navigation combined positioning technology,combined positioning based on GNSS and inertial systems has gradually become a hot technology in the field of train positioning.However,due to the disadvantages of satellite signals such as being instable and being easily interfered,in order to ensure the accuracy of train positioning information and the reliability of system operation,it is of great significance to carry out Prognostics and Health Management(PHM)based on performance status for inertial equipment.This paper fully considers the application background of high-speed trains and the characteristics of high integration and long life of inertial equipment.Based on the performance degradation monitoring data of the laboratory train positioning inertial unit,the PHM research of inertial unit has carried out from four aspects: construction of performance evaluation index system,modeling of performance degradation,prediction of remaining useful life(RUL)and condition-based maintenance optimization decision.The main research of this paper is as follows:(1)According to the constituent structure and performance evaluation criterion of inertial unit,the random error index of inertial sensor has been used to characterize the performance state of the equipment.The Allan variance method has been used to identify the error from the drift data of inertial unit.And according to the principle of performance index selection and the performance monitoring data of the target equipment,the performance index system including the random walk coefficient of gyroscope and accelerometer has been constructed,which provides the data basis for the later research.(2)One adaptive prediction method of RUL based on Copula theory and nonlinear drift Wiener process has been proposed for the performance degradation modeling and RUL prediction of the train positioning inertial units.The performance degradation model and the marginal cumulative distribution function of RUL of gyroscope and accelerometer have been established by using the Wiener process with nonlinear drift.Based on Copula function establish the joint PDF of RUL of binary performance parameter.The parameters of performance degradation model and Copula function have been estimated step by step by STF & EM joint algorithm and MLE algorithm respectively,which realizes the adaptive prediction of RUL.(3)Based on the replacement-reward theory and the predicted distribution of RUL of inertial unit,an optimal strategy model of condition-based maintenance for inertial unit is constructed.The optimal replacement time has been solved by using the improved adaptive genetic algorithm,which provides a scientific and reasonable predictive maintenance decision for the health management of inertial unit.This paper uses the performance degradation monitoring data of the laboratory train positioning inertial unit for simulation verification.The results show that: RUL prediction results of the binary performance parameters based on Copula function show that the PDF curve of RUL has a smaller variance,a smaller expected value,and the average MRE of multiple monitoring points before failure is 3.27% less than that based on the gyroscope.It has proved that the RUL prediction method proposed in this paper can reduce the uncertainty of prediction,and improve the accuracy.According to the maintenance strategy model of train positioning inertial unit,the optimal replacement decision has obtained before the equipment fails under the condition of binary performance index and gyroscope.However,the gyro-based decision lags behind the binary performance parameter decision by 1.68 hours,which will increase the system operating risk,and prove that the optimization decision based on binary performance parameter is more reliable and effective.This thesis contains 52 figures,20 tables and 93 references.