| Distributed drive electric vehicles(DDEVs)are regarded as excellent carriage for intelligent driving due to their flexible layout,rapid response,and controllability.The scale of the active safety system and the variety of electronic components on the vehicle are becoming increasingly complex as intelligent driving evolves,causing the vehicle to respond to emergencies with increasing safety.However,as vehicles become more electrified,the likelihood of actuator failure has increased.Fault-tolerant control(FTC)for actuator function failure has emerged as an acritical safety mechanism for DDEVs safety and a hot research topic in this field.The current research on FTC is primarily focused on the motion control of the vehicle itself,while insufficient attention is paid to the role of the human driver,and there is a lack of theoretical modeling methods for the interaction behavior of multiple objectives and multiple actuators.Therefore,the thesis addresses the FTC problem of DDEVs experiencing irreversible actuator failures,considers the role of the driver in FTC,and focuses on nonlinear FTC,driver uncertainty behavior,and theoretical modeling of its conflict with the vehicle using differential cooperative game theory.The details of this thesis are as follows:(1)Improved feedback linearized passive fault-tolerant control(PFTC)is investigated by cross-fusing cooperative game and terminal sliding mode control to address the issue of strong nonlinear dynamics of vehicles suffering from actuator failure.Hardware-in-the-loop(HIL)tests are used to verify the real-time performance and effectiveness of the proposed control algorithm.The research results achieve the computational efficiency and accuracy of nonlinear PFTC simultaneously.(2)To address the interaction modeling problem of actuators in PFTC of driver-vehicle systems,a Takagi-Sugeno(T-S)fuzzy method is used to approximate the nonlinearity of the system,and a T-S cooperative game model with chassis actuators as players is developed to theoretically describe the goal and strategy interactions among the actuators in FTC.The asymptotic stability of the system is demonstrated using an infinite time-domain approach.The HIL test results show that the designed method can guarantee the stability of the vehicle driven by the different styles of drivers in case of actuator failure.The proposed method provides a theoretical tool for modeling cooperative games of T-S systems.(3)To improve the identification accuracy of in-wheel motor fault parameters,a neural network tire model combining physical model and data,and a fault diagnosis(FD)method based on interactive multi-model(IMM)moving horizon estimation(MHE)are designed from the perspectives of improving the accuracy of tire model and improving the accuracy of the observer,respectively.It overcomes the disadvantage that the accuracy of the traditional tire model decreases with the change of usage conditions,and improves the observation accuracy of the faulty motor torque by using multi-model interaction correction,system history observation information,and constraint information.The method can obtain high-precision in-wheel motor fault parameter identification results under severe working conditions,and lay the groundwork for further research into active fault-tolerant control(AFTC).(4)The single-leader-multiple-follower hybrid game AFTC method is designed for the sequential interaction process in which the driver decides first and then the actuator responds to the driver’s decision to compensate for its negative effects.The driver is mapped as the leader in the Stackelberg leader-follower game to decide first,and the chassis actuator is mapped as multiple followers in the leader-follower game,which achieve the common pursuit of stability and respond optimally to the leader’s decision through the cooperative game.The effectiveness of the method is verified using HIL tests.The method provides a research method for the sequential interaction process among the driver and the chassis actuators and the different interaction modes.(5)An AFTC strategy based on the stochastic cooperative game(SCG)is proposed for the problem that the driver’s behavioral uncertainty is more accentuated under the actuator failure stimulus.The driver and chassis actuator are mapped as players in an SCG,and the coupled stochastic optimization problem with chance constraints is solved using a probabilistic constraint tightening method and a convex iterative algorithm,the computational efficiency is improved based on a parallel computing architecture.Simulation results show that the designed method can maintain the vehicle in the stability zone after the actuator fails under stochastic driver perturbations.The designed method provides a theoretical framework for investigating subsystem interactions with stochastic properties. |
