Session: Controls 1

Paper Number: 111276

111276 - A Feedforward Energy-Saving Control Method for Imvc Hydraulic Cylinder Systems Using the Deep Koopman Operator 

Independent metering valve-controlled (IMVC) hydraulic cylinder systems can achieve energy-efficient control more conveniently due to the fully controllable pressure in both chambers of the hydraulic actuator. However, the entire flow is adjusted by the metering valve and requires multiple input control variables, leading to stronger nonlinearity than traditional valve-controlled hydraulic systems. Thus, the accuracy of motion control is challenging. To solve this problem, this paper proposes a feedforward energy-saving control method for IMVC hydraulic cylinder systems using the Deep Koopman operator.

Firstly, a deep neural network is adopted instead of the traditional Dynamic Mode Decomposition (DMD) algorithm to calculate the Koopman operator. This operation improves the computational efficiency and approximate accuracy of the Koopman operator. Then, a high-precision linear prediction model is obtained regarding the controlled plant. The model could be used for the feedforward control compensation calculation of the IMVC hydraulic cylinder system.

On this basis, an energy-saving control strategy is proposed to reduce the energy consumption by controlling the flow and pressure in two chambers on both sides of the actuator, respectively.

Finally, the proposed controller, which introduces the Deep Koopman operator to improve feedforward items, is applied to the IMVC hydraulic cylinder system to achieve energy-saving operation. In particular, the Koopman operator is calculated by neural network training. Simulation tests are conducted to verify the control accuracy and energy consumption of such systems. The results show that the proposed controller achieves higher control accuracy and energy saving effect for the IMVC hydraulic cylinder system, compared with the PID controller and traditional robust control. Therefore, the proposed control scheme is effective and superior.

Presenting Author: Heng Liu Shanghai Jiao Tong University

Presenting Author Biography: Heng Liu received the B.S. degree in mechanical engineering from Shanghai Jiao Tong University (SJTU), China, in 2021. He is currently working toward the M.S. degree in the same major at Shanghai Jiao Tong University (SJTU), Shanghai, China. His research interests include data-driven intelligent control algorithms, hydraulic control, and 5G intelligent control.