Sliding mode reinforcement learning vibration control of a translational coupled double flexible beam system

The coupled vibration control of a translational double flexible beam system with high nonlinearity and strong coupling is studied. The dynamic model of the system is established based on the finite element method (FEM), and the model parameters are identified by acquiring the experimental data of the system. To overcome the chattering problem of sliding mode control (SMC), a sliding mode reinforcement learning control (SMRLC) method is investigated. Taking the sliding surface as the state input, the asynchronous advantage actor-critic (A3C) algorithm is adopted to learn the optimal control policy through repeated trial and error interaction with the environment, so as to achieve robust control performance. The identification model of the double flexible beam system is used to pre-learn the SMRLC scheme to balance the learning efficiency and precision, and the effectiveness of the pre-learned SMRLC method is further verified by numerical simulation. The experimental setup is constructed, and the vibration control experiments of the first two bending modes are conducted, including set-point vibration control and simultaneous translation and vibration control under trapezoidal trajectory. SMRLC is compared with PD control and SMC. The experimental results demonstrate that the designed SMRLC method can effectively reduce the chattering of control signals and suppress the large and small amplitude vibration of the double flexible beam system, and there is no saturation in the control value.