Prediction-based stochastic dynamic programming control for excavator

Abstract

Abstract Because the fuel efficiency of excavators is low and the energy loss is considerable, a prediction-based stochastic dynamic programming (PBSDP) control strategy is proposed in this study to reduce the energy and fuel consumptions. Using the torque prediction method, the optimization control strategy can be used in real time to improve the fuel efficiency. The required torque of the proposed hydraulic hybrid excavator was analyzed. The minimum length of the required torque sample needed to estimate the trend in the required torque change was obtained using approximate mean estimation. The required torque sample was divided into complete cycles using empirical mode decomposition. The required torque of the next period was predicted using a signal superposition technique. The proposed PBSDP control strategy was then applied based on the predicted required torque. The displacement of the auxiliary pump/motor was selected as the control variable whereas the pressure of the accumulator connected to the auxiliary pump/motor was selected as the state variable. The value function was calculated based on the predicted required torque and charging state of the accumulator. The controller was used to minimize the value function by adjusting the displacement of the auxiliary pump/motor. In addition, a numerical experiment was conducted to analyze the rotation torque, energy consumption, and fuel consumption of an internal combustion engine (ICE). The numerical experiment results show that the proposed PBSDP control strategy helps in reducing the maximum torque, energy consumption, and fuel consumption of the ICE by 15, 20, and 26%, respectively.