A Two-stage Optimal Controller For Hydraulic Loading System With Mechanical Compensation

Energy consumption and temperature rise are challenging in traditional high-power hydraulic pump/motor loading systems. This study focuses on the hydraulic loading systems that the energy can be recovered by mechanical compensation. The key challenge is to achieve pressure control accuracy while keeping the input energy low. Due to the high inertia of the loading motor and the unstableness of the system dynamics, it is difficult to obtain an optimal control performance with a single traditional controller. So a novel two-stage optimal tracking controller is proposed in this study. In the first stage, a pressure controller is employed to track the reference pressure input, and the system is transformed to be stable. In the second stage, an optimal tracking controller optimizes the infinite horizon value function and provides the reference input signal for the first-stage controller. Considering the nonlinearities of the augmented system, the technique of integral reinforcement learning (IRL) is used to evaluate the value function and obtain the improved control policy online. A simulation is carried out for the hydraulic loading system with mechanical compensation and the results show that the two-stage optimal controller outperforms the single-stage pressure controller.