Estimation of the vehicle speed in the driving mode for a hybrid electric car based on an unscented Kalman filter

Preconditions for the effective implementation of a control strategy in an automobile's active safety system rely on timely and accurate information on the vehicle's running status, especially the vehicle speed. However, the vehicle speed cannot be measured directly without using advanced onboard sensors or specialized test equipment owing to their high mass production cost. In this study, a model-based method for estimating the vehicle speed in different driving modes is conducted in real time, which makes full use of information on the driving wheel's torque for a four-wheel-drive hybrid car. First, a simulation platform that integrates the models of the powertrain system, the non-linear seven-degree-of-freedom vehicle dynamics system and the dynamic UniTire model is established. Next, an unscented Kalman filter algorithm is adopted to estimate the vehicle speed, and the estimated results and the simulated results are compared under different driving modes. Finally, real-vehicle tests at medium and low speeds are performed using a prototype car. The simulations and the test results confirm that the proposed unscented Kalman filter estimation algorithm without a linearizing truncation process can estimate the vehicle speed with high precision and strong adaptability.