Reinforcement Learning For Control Strategy Of Vehicle Ride Comfort Of Semi-active Suspension

With the development of economy,people put forward further requirements for the comfort of daily driving.As a key part of automobile,suspension has a profound impact on the ride comfort and operation stability of the automobile.The research on the control of suspension has always been a hot spot of automobile manufacturers and academia.With the improvement of the level of artificial intelligence technology,more and more products are gradually moving towards intelligence,and the automobile suspension is also developing towards intelligence.This paper focuses on the intelligent control algorithm of semi-active suspension based on vehicle ride comfort.The main research contents are as follows:(1)A 1/4 vehicle suspension model is built to determine the performance evaluation index of the suspension.The positive model of Bingham’s MR damper and the inverse model of multinomial MR damper are established,in which the real output damping force can be obtained through the motion state and current,and the inverse model can deduce the current needed to be controlled according to the damping force required by the suspension.(2)The control model of sky-hook,ground-hook,hybrid damping,PID are established.And analyze these traditional control methods.(3)The reinforcement learning controller based on Q-learning algorithm is established.analyze the main principles of reinforcement learning,set parameters such as learning rate and discount factor,and build a simulation model.(4)The simulation environment is established in Simulink,the continuous uneven road and discrete uneven road are established to simulate the actual road,and the vibration characteristics of the vehicle suspension are simulated and analyzed under different road conditions and vehicle speeds.In this paper,an intelligent suspension controller based on reinforcement learning is designed to improve the ride comfort of the vehicle.By contrasting the evaluation system of semi-active suspension under the traditional control algorithm of hybrid shed,PID control algorithm and reinforcement learning control algorithm,it is concluded that the semi-active suspension with reinforcement learning algorithm can effectively improve the vehicle ride comfort.