Motion Control of Servo Cylinder Using Neural Network

신경회로망을 이용한 서보 실린더의 운동제어

  • 황운규 (홍익대학교 대학원 기계공학과) ;
  • 조승호 (홍익대학교 기계시스템 디자인공학과)
  • Published : 2004.07.01


In this paper, a neural network controller that can be implemented in parallel with a PD controller is suggested for motion control of a hydraulic servo cylinder. By applying a self-excited oscillation method, the system design parameters of open loop transfer function of servo cylinder system are identified. Based on system design parameters, the PD gains are determined for the desired closed loop characteristics. The Neural Network is incorporated with PD control in order to compensate the inherent nonlinearities of hydraulic servo system. As an application example, a motion control using PD-NN has been performed and proved its superior performance by comparing with that of a PD control.


Neural Network;Servo Cylinder;PD-NN Control;Self-Excited Oscillation


  1. Sanada,K. and Kitagawa.A, 1993, 'An application of a Neural Network to adaptive Control of Servo system,' Proc. Of the 2nd JHPS international Conference on Fluid Power, pp. 303-308
  2. Hwang, U. K., 2002, 'Motion Control of a Hydraulic/Pneumatic Servocylinder Using Neural Network,' Ph.D Thesis, Hongik University, Department of Mechanical Engineering
  3. Nishiumi, T. and Konami,S., 1997, 'An Application of the Identification Method Using Self Excited Oscillation to Hydraulic Motor/Load System,' 10th Bath International Fluid Power Workshop, pp. 381-395
  4. Takao NISHlUMI, Hiroshi KATOH and Shizurou KONAMI, 2001, 'Appliction of Neural Network for Hydraulic Motor/Load System,' Trans of JSME (C), Vol. 67-7, p. 659
  5. Konami,S., Nishiumi,T. and Hata,K., 1996, 'Identification of Linearized Electro-Hydraulic Servovalve Dynamics by Analyzing Self-Excited Oscillations,' Journal of the Japan Hydraulics and Pneumatics Society, Vol. 27-4, pp. 143-149