Nonlinear Disturbance Observer with Recurrent Neural Network Compensator

  • Shihono Yamada
  • Jun IshikawaEmail author
Conference paper
Part of the Lecture Notes in Electrical Engineering book series (LNEE, volume 554)


This paper proposes a nonlinear disturbance observer with recurrent neural network compensator applicable to nonlinear systems that is velocity controlled. In the proposed method, the recurrent neural network is trained to generate nonlinear velocity defined as the difference between the measured velocity of joints and a nominal linear velocity calculated via a linear model from the reference input to the velocity control system. The training results is evaluated based on the leave-one-out method, and it is shown that the RNN can be well-trained to estimate the nonlinear velocity. Then, a common DOB can be estimate disturbance from the nominal linear velocity restored by the trained RNN output and the velocity measurement. The validity of the proposed method is evaluated by simulation for a vertically-articulated two-link manipulator, comparing a conventional disturbance observer. The simulation result showed that the proposed disturbance observer can be comparable to a conventional DOB that works in an ideal condition where all the parameters of the manipulator are known, while the data needed to construct the proposed DOB is only the velocity measurement and its reference.


Recurrent neural network Disturbance observer Nonlinear system 


  1. 1.
    Zhang, X.H., Liu, X.H., Ding, H.S., Li, H.S.: PMSM speed-adjusting system based on disturbance observer and finite-time control. Control Decis. 24(7), 1028–1032 (2009)Google Scholar
  2. 2.
    Ma, J., Yang, T., Hou, G.Z., Tan, M.: Adaptive neural network controller of a Stewart platform with unknown dynamics for active vibration isolation. In: Proceedings of the IEEE International Conference on Robotics and Biomimetics, pp. 1631–1636 (2008)Google Scholar
  3. 3.
    Nakao, M., Ohnishi, K., Miyachi, K.: A robust decentralize joint control based on interference estimation. In: Proceedings of the IEEE International Conference on Robotics and Automation, pp. 326–331 (1987)Google Scholar
  4. 4.
    Yang, J., Li, H.S., Chen, X., Li, Q.: Disturbance rejection of ball mill grinding circuits using DOB and MP C. Powder Technol. 198(2), 219–228 (2010)CrossRefGoogle Scholar
  5. 5.
    Choi, Y., Yang, K., Chung, K.W., Kim, H.R., Suh, H.I.: On the robustness and performance of disturbance observer for second-order systems. Autom. Control 48(2), 315–320 (2003)MathSciNetCrossRefGoogle Scholar
  6. 6.
    Zhou, P., Xiang, B., Chai, Y.T.: Improved disturbance observer (DOB) based advanced feedback control for optimal operation of a mineral grinding process. Chin. J. Chem. Eng. 20(6), 1206–1212 (2012)CrossRefGoogle Scholar
  7. 7.
    Li, Q.S., Qiu, H.J., Ji, H.L., Zhu, J.K., Li, J.: Piezoelectric vibration control for all-clamped panel using DOB-based optimal control. Mechatronics 21(7), 1212–1221 (2011)Google Scholar
  8. 8.
    Li, Q.S., Qiu, H.J., Li, J., Ji, H.L., Zhu, J.K.: Multi-modal vibration control using amended disturbance observer compensation. IET Control Theor. Appl. 6(1), 72–83 (2012)MathSciNetCrossRefGoogle Scholar
  9. 9.
    Li, H.S., Yang, J.: Robust autopilot design for bank-to-turn missiles using disturbance observers. IEEE Trans. Aerosp. Electron. Syst. 49(1), 558–579 (2013)CrossRefGoogle Scholar
  10. 10.
    Guo, L., Chen, H.W.: Disturbance attenuation and rejection for systems with nonlinearity via DOBC approach. Int. J. Robust Nonlinear Control 15(3), 109–125 (2005)MathSciNetCrossRefGoogle Scholar
  11. 11.
    Li, S., Liu, Z.: Adaptive speed control for permanent magnet synchronous motor system with variations of load inertia. IEEE Trans. Ind. Electron. 56(8), 3050–3059 (2009)CrossRefGoogle Scholar
  12. 12.
    Chen, W.: Disturbance observer based control for nonlinear systems. Mechatronics 9(4), 706–710 (2004)Google Scholar
  13. 13.
    Li, J., Jun, Y., Shihua, L., Xisong, C.: Design of neural network disturbance observer using RBFN for complex nonlinear systems. In: Proceedings of the IEEE International Conference on Chinese Control Conference, pp. 6187–6192 (2011)Google Scholar
  14. 14.
    Ma, J., Yang, T., Hou, Z.G., Tan, M.: Neural network disturbance observer based controller of an electrically driven stewart platform using backstepping for active vibration isolation. In: Proceedings of the IEEE International Conference on International Joint Conference on Neural Networks, pp. 1939–1944 (2009)Google Scholar
  15. 15.
    Beale, H.M., Hagan, T.M., Demuth, B.H.: MATLAB Neural Network Toolbox Getting Started Guide (R2016b). Math Works Inc. (2016)Google Scholar

Copyright information

© Springer Nature Switzerland AG 2020

Authors and Affiliations

  1. 1.Tokyo Denki UniversityTokyoJapan

Personalised recommendations