A Model Predictive Control-based Energy Efficiency Control Algorithm for Power Unit of Hydraulic Humanoid Robots

Hydraulic humanoid robots, owing to their superior power-to-weight ratio of hydraulic systems compared with electric systems, exhibit considerable potential for heavy-load applications. However, the supply-demand mismatch and severe overflow losses caused by constant-current energy supply of traditional power unit substantially limit system endurance. To address this issue, a model predictive control-based energy efficiency control algorithm for a hydraulic power unit is proposed in this paper. In contrast to conventional reactive regulation strategies, the proposed method incorporates the nonlinear dynamic characteristics of the system into the prediction model and formulates a multi-objective optimization function that balances pressure-tracking accuracy against system energy consumption. The desired pressure trajectory is first determined from the robot’s motion state. A model predictive control framework is then established to predict pressure demand over a future horizon and regulate the motor-pump speed in real time through receding horizon optimization. This enables accurate matching between supply and demand of hydraulic energy while maintaining stable robot operation, thereby effectively reducing redundant power consumption of system and extending operational endurance of the robot. Numerical simulations and experimental results demonstrate that the proposed method can minimize system energy consumption while ensuring stable operation and accurate tracking of the commanded trajectories.