Reinforcement-Learning-Based Vibration Control for a Vehicle Semi-Active Suspension System via the PPO Approach

The vehicle semi-active suspension system plays an important role in improving the driving safety and ride comfort by adjusting coefficients of damping spring. main contribution this paper is proposal a PPO-based vibration control strategy for system, which designed reward function realizes dynamic adjustment according to road condition changes. More specifically, different performances caused conditions, three body acceleration, deflection, tire load, were taken as state space PPO algorithm, value was set numerical results passive suspension, so that corresponding force selected action space, weight matrix dynamically adjusted agent could have better improvement effect at speeds conditions. In paper, quarter–car model analyzed simulated, simulations performed using stochastic excitation classes roads, models, continuously changing simulation showed acceleration reduced 46.93% under road, proved effectively improve performance combining changes with function.