Estimation of the State Variables and Unknown Input of a Two-Speed Electric Vehicle Driveline Using Fading-Memory Kalman Filter

This paper studies the stochastic estimation of unavailable state variables and the unknown input of an electric vehicle (EV) driveline equipped with a novel seamless clutchless two-speed transmission. The proposed transmission is explained and the kinematics and dynamics of the driveline, which constitute the basis for the observer design, are presented. For identical inputs, the outputs of the dynamical model are compared to those of the experimental test rig and the simulation model created in the MATLAB/Simulink. The method of modeling the unknown input as a fictitious state variable is combined with the fading-memory Kalman filter (FMKF) in order to provide a robust concurrent estimation of unavailable states and the unknown input. The observer estimates angular velocities of the off-going and on-coming gears and consequently the gear ratio, the input and output torques of the transmission, and the unknown torque exerted on the vehicle based on the speed measurements of the electric motor and wheels. The observability of the states and unknown input of the augmented system is analyzed and the performance of the proposed observer is experimentally assessed for upshift and downshift scenarios. The estimation results are compared with the conventional KF and the deterministic Luenberger observer (DLO).