| This thesis is concerned with fault detection problem in running bogie of high-speed railway vehicles. Reliability of the railway system is critical for safe and punctual operation. The safety of high-speed trains is closely related to the components in the running bogie and wheels. Kalman filters as well as nonlinear filters are employed to estimate states and key components’parameters of the high speed train. Simulative results verify the feasibility of Kalman filter based methods under varies situations in the running bogie.Firstly, dynamic model of the high speed train running bogie is established. This model describes the motion of lateral displacement and yaw angle of the wheelsets, bogie and vehicle body, which capture the essential dynamic characteristics related to the problem being considered. A discrete state-space model is established based on the high-speed train running bogie dynamic model. The running training simulation is carried out in the simulation softwares SIMULINK and SIMPACK.Kaman filter is employed to exam and value the states of the running train. Sensors on the vehicle are interfered by Gaussian white noise and other random inaccuracies. The Kalman filter simulation results give some key states accurately.Additionally, to estimate some key components’parameter value, such as anti-yaw damper, lateral damper and wheelsets conicity, the state-space model needs to be written as nonlinear when the key parameters are chosen as a part of state and the discrete dynamic model is rebuilt. Two nonlinear estimation methods are used to estimate the key parameters’ value. The simulation results show the effectiveness of those methods.The last part of this paper adds a more precise description of one key component (the anti-yaw damper), which is nonlinear. The nonlinear character is described by a set of piecewise linear curves. The Extended Kalman filter simulation results show feasibility of EKF even under nonlinear parameter estimation in the running bogie. |
PDF Full Text Request