Research On Fault Diagnosis Algorithm Of Urban Rail Vehicle Suspension System Based On Parameter Estimation

Suspension system is one of the main components of urban rail vehicle, which plays a vital role for the vehicle riding comfort and safety, and the change of its parameters usually means the change of vehicle riding comfort, in some severe cases, could result in serious economic losses and accidents. Thus, real-time fault diagnosis for suspension system is extremely important. Since it’s easy to build a mathematical model for vehicle suspension system, the fault diagnosis method based on parameter estimation is widely used, and many theoretical research results have been proposed at home and abroad. However, considering the cost of the test equipment and the affect on the operation of the vehicle, almost all the parameter estimation algorithms are still in the stage of theoretical research. Therefore, it is very necessary and meaningful to develop an effective test platform for validation of the parameter algorithms. In view of this, the test platform of the suspension system has been established in this article. The main work of this paper is:(1) The theoretical features and application of the extended Kalman filter and marginal particle filtering algorithm are introduced, the six degree of freedom model of suspension system has been chosen as a parameter estimation object, and the estimation results of marginal particle filter and extended Kalman filter are compared, which show that the estimation accuracy based on marginal particle filter is relatively higher and the selection of the initial value has no influence its estimation result.(2) A four degree of freedom test platform has been established, and its physical structure is introduced in detail. According to Lagrange equation, the dynamic equation of the test platform has been built and converted into state space form. Since any linear system described by the state space form or any nonlinear and non Gauss system described by it can be estimated by Kalman filter or particle filter respectively. Therefore, this test platform has no essential difference between the vehicle suspension system which usually has a higher degree of freedom.(3) A parameter estimation method based on marginal particle filter is used to estimate the primary and secondary suspension stiffness of the test platform. Then the measurement data based on simulation is replaced by the data from test platform, and the results prove the validity of marginal particle filter.