• CN: 11-2187/TH
  • ISSN: 0577-6686

Journal of Mechanical Engineering ›› 2025, Vol. 61 ›› Issue (4): 355-364.doi: 10.3901/JME.2025.04.355

Previous Articles    

Neural Network Control of Electrohydrostatic Actuator Based on Flow Pulsation Compensation

GE Yaowen1,2, YANG Xiaowei1, Deng Wenxiang1, YAO Jianyong1   

  1. 1. School of Mechanical Engineering, Nanjing University of Science and Technology, Nanjing 210094;
    2. State Key Laboratory of Fluid Power and Mechatronic Systems, Zhejiang University, Hangzhou 310027
  • Received:2024-04-07 Revised:2024-09-17 Published:2025-04-14

Abstract: Flow pulsation is an inherent attribute of the plunger pump. Theoretically, it is generated by the reciprocating movement of multiple plungers. However, in actual operation, residual pressure will be accumulated by the extrusion and discharge of oil in the cylinder hole sealing cavity of the plunger, which is accompanied by nonlinear uncertainties such as leakage, resulting in deviation between the actual flow pulsation and the theoretical flow pulsation. In the current study of nonlinear control methods for electrohydrostatic actuators(EHA), the flow pulsation is usually treated linearly and compensated feedforward. The above processing method ignores the nonlinear characteristics and deviation effect of pump flow pulsation, thus enlarges the upper bound of the matching uncertainty, resulting in the increase of the controller gain coefficient and the amplification of noise, which limits the improvement of the control accuracy of the EHA. In order to achieve higher control accuracy, a neural network control method of the EHA based on flow pulsation compensation is proposed. Firstly, a nonlinear normal model of theoretical flow pulsation is established for feedforward compensation. Then the friction of the actuator, the interference of the second channel, the residual of the theoretical and actual flow pulsation and the interference of the third channel are approximated in real time by the two-channel neural network. The experimental results show that compared with two other control methods, the tracking accuracy of the proposed method is improved by more than 65%, and the control performance is better while reducing controller gain and tuning difficulty.

Key words: electro-hydrostatic actuator, neural network control, flow pulsation, feedforward compensation

CLC Number: