Vehicle state and parameter estimation under driving situation based on extended kalman particle filter method

Individual parameters of vehicle dynamic systems were traditionally derived from expensive component indoor laboratory tests as a result of an identification procedure. These parameters were then transferred to vehicle models used at a design stage to simulate the vehicle handling behavior and the cost of measurement was high. At the same time, acquiring the vehicle's driving status and parameters had important significance for the process controlling of the vehicle. Normally, the status and parameter of the test vehicle needed to be estimated together, which were then transferred to vehicle models and used at a design stage to simulate the vehicle handling behavior. A vehicle dynamics system containing constant noise and non-linear model was established, Runge-Kutta method was used to simulate the model. The extended Kalman filter algorithm was used as the importance density function to update particles in particle filter, with which the local state estimated values and parameters can be calculated. The simulation results showed that the proposed algorithm improved the accuracy of standard particle filter.