Simulation Research On Coordination Control Of SAS And EPS Based On Reinforcement Learning Algorithm

Semi-active suspension (SAS) can improve vehicle ride performance and handlingstability by adjusting the damping. Electric Power Steering (EPS), which is energyconservation and environmental protection, can effectively improve the steeringportability and stability. As the principal subsystems of the vehicle, they affect the car’sride performance and handling stability. With the development of control theory, sensortechnology and electronic control technology, SAS and EPS have become researchhotspots. However, due to the coupling effects between them, simple stack control can’tobtain the optimal performance. Simple discrete control can improve part performance.Therefore, in order to improve the comprehensive performance, the integrated control isuses to eliminate the coupling between those two systems. In this paper, aiming at theimprovement of comprehensive performance, a more in-depth research on the SAS, EPSas well as integrated control is undertaken.First of all, in the view of the SAS system, a method for SAS control based onimmune algorithm is proposed to improve suspension performance. According to thisalgorithm, an immune controller is designed to research and simulation for SAS control.Simulation results show that the proposed algorithm is effective. And compared with thepassive suspension and fuzzy logic control suspension, its control capability is better.Vehicle ride performance, handling stability are effectively improved.And then, in respect of the EPS system, an EPS model is established in this paper.The steering wheel angle is taken as the input signal and the steering column angle as theoutput signal of the model. A control algorithm for EPS based on immune feedbackmechanism and fuzzy control theory is proposed. According to this algorithm, animmune fuzzy PID controller is designed to research and simulation for EPS control inMatlab/Simulink. Simulation results show that its control capability is better than theconventional PID. The system rapid response ability and follow-up characteristics arebetter. The proposed algorithm can efficiently improve steering portability.In connection with the integrated control, a layered coordination control method isproposed, and a coordinate controller is designed in order to eliminate the couplingbetween SAS and EPS. And then, a control system based on fuzzy algorithm isestablished combined with immune controller of SAS and immune fuzzy PID controllerof EPS. The results of the simulation show that, the coordination control theory can,improve vehicle’s comprehensive performance, and coordinate the ride performance andhandling stability effectively.Finally, in order to improve the control delay, adaptive capacity and robustness of theintegrated control system, a coordination control method based on Q-learning ispresented, and a Q-learning controller is designed. And then, a simulink model combined with Carsim and the proposed controller is established for co-simulation. Simulationresults show that the control performance of Q-learning algorithm is better than thecoordination control. Q-learning algorithm control can effectively reduce the yaw rateand roll angle, the Q-learning algorithm can improve the steering stability whileimproving ride comfort, its control effect is remarkable.