Model Predictive Control for Energy-Efficient Yaw-Stabilizing Torque Vectoring in Electric Vehicles With Four In-Wheel Motors

This paper considers the problem of stabilizing and energy-efficient torque vectoring for electric vehicles with four independent in-wheel motors. In electric vehicles with four in-wheel motors, four electric motors are separately attached to the four wheels without an extra drive shaft. The mechanical and structural nature enables reduction of energy loss during power transmission and securing extra interior space. In addition, independent control of wheel torques can provide better yaw motion stability and improved energy efficiency. This paper proposes two model predictive control (MPC) methods for stability-constrained energy-efficient torque vectoring of four in-wheel motor electric vehicles. For the adaptive weighting factors of multiple objective functions of reference tracking and energy saving, we use exponential functions that vary with the lateral motion and steering input. Depending on the optimal control problem formulation with different dynamical system equations and constraints, the associated predictive controller can be represented as either a linear parameter-varying MPC (LPV-MPC) or nonlinear MPC (NMPC). For LPV-MPC, longitudinal and lateral motions are decoupled, whereas the coupled dynamics of the two-track model are exploited in NMPC. For comparisons and demonstrations of LPV-MPC and NMPC in the MPC of torque vectoring, three driving scenarios are simulated with a high-fidelity vehicle simulation solution, CarMaker (IPG Automotive). In comparison with the built-in IPG driver implemented in CarMaker, we demonstrate fuel efficiency improvements of over 2–3% on average with guaranteed yaw stability.