Control strategy of hydraulic cylinder based on Deep Reinforcement Learning

Authors developed a novel control strategy of hydraulic cylinder based on deep reinforcement learning. The control parameters of hydraulic cylinder are difficult to regulate for practical applications, and problems of force and oil pressure disturbance occur during the operation process. A class of reinforcement learning agents developed for hydraulic systems is designed based on the deep deterministic policy gradient and proximal policy optimization algorithms. The agents are trained by a significant number of system data. After learning completion, they can automatically control the hydraulic system online and consequently the system can always maintain a good control performance. Experiments are conducted to verify the proposed control strategy. Results show that the proposed method can achieve better performance that conventional proportional-integral-derivative regulator and effectively overcome the effects of disturbance.