Trajectory planning and vibration control of translation flexible hinged plate based on optimization and reinforcement learning algorithm

Aiming at the flexible hinged plate with translational motion, a hybrid control strategy combining motion trajectory optimization and piezoelectric active control is developed. Combining finite element modeling (FEM) and experimental system identification, the piezoelectric driving model and motor acceleration driving model of the flexible hinged plate are obtained. On this basis, the immune optimization algorithm is adopted to obtain the optimal trajectory of vibration suppression. An Auto-Soft-Actor-Critic (Auto-SAC) reinforcement learning (RL) algorithm is designed to train modal active controllers. The experimental setup is constructed. Compared with the classical trajectory, the optimized trajectory shows a good effect in avoiding excitation vibration. For piezoelectric active control, the Auto-SAC RL modal controller is superior to the large-gain PD controller in suppressing vibration, and its nonlinearity effectively overcomes the problem that the traditional linear controller has limited ability to suppress small amplitude vibration. In addition, the hybrid control strategy of trajectory optimization and Auto-SAC RL modal controllers can effectively reduce and suppress the vibration of the flexible hinged plate in translational motion.