Stochastic Cooperative Game-Based Assistant Fault-Tolerant Control Considering Driver Behavior Uncertainty
Bohan Zhang, Shaobo Lu, Feifei Xie, Wenke Xie
Abstract
A stochastic cooperative game (SCG)-based assistant fault-tolerant control (FTC) is proposed in this article for maintaining the stability of the distributed drive electric vehicle (EV) suffering actuator failure, which also takes the driver’s uncertain behavior into account. A distributed stochastic model prediction control (DSMPC) framework is leveraged to simulate the interaction among the driver, automatic steering, and in-wheel motors. First, a stochastic driver-vehicle model that includes driver uncertainties and the potential actuator failure is established. Then, the idea of the cooperative game is introduced to mathematically describe the conflict and cooperation among the driver and automatic control subsystems. Specifically, the human driver and the automated control subsystems are modeled as players, whose interests are represented by the mathematical expectation of the cost function considering the stochastic nature of the system. Finally, the Pareto solution of the cooperative game is derived using a convex iteration algorithm. A driving simulator experiment is performed to capture the stochasticity of different drivers. Numerical simulations of two serious failures prove that the proposed SCG method can better maintain the stability of a failed vehicle while also taking account of the driver behavior uncertainty.