Residual-Aided Adaptive State Estimation for Distributed Electric-Drive Vehicles
Abstract
Accurate vehicle state estimation is essential for stability control and the active safety of distributed electric-drive vehicles, whereas lateral velocity and sideslip angle are difficult to measure directly using production-level sensors. This article proposes a residual-aided adaptive extended Kalman filter for multisensor vehicle state estimation under varying-speed maneuvers and time-varying measurement conditions. A three-degree-of-freedom vehicle dynamics model is established by considering longitudinal, lateral, and yaw motions. Four-wheel driving torques are converted into longitudinal tire forces, and the Dugoff tire model is used to describe nonlinear lateral tire characteristics. The longitudinal velocity, lateral velocity, and yaw rate are selected as system states, while the sideslip angle is calculated from the estimated velocities. To improve adaptability, the process noise covariance is adjusted using velocity-related, lateral-acceleration, and yaw-rate residual indicators. Meanwhile, the measurement noise covariance is updated using channelwise residual indicators from acceleration, yaw rate, and wheel-speed measurements. CarSim/Simulink cosimulation under a varying-speed double-lane-change maneuver and real-vehicle experiments are conducted to verify the proposed estimator. Compared with the conventional extended Kalman filter, the proposed method reduces the mean absolute errors (MAEs) of yaw rate, sideslip angle, and lateral velocity by 50.10%, 29.58%, and 31.91%, respectively, demonstrating improved tracking accuracy and robustness for distributed electric-drive vehicle state estimation.