Modeling a simple inverted pendulum using a model-based dynamic recurrent neural network

Author

Karam, M. ; Zohdy, Mohamed A.

Author_Institution

Dept. of Electr. Eng., Tuskegee Univ., USA

fYear

2005

fDate

20-22 March 2005

Firstpage

78

Lastpage

82

Abstract

A model-based dynamic recurrent neural network (MBDRNN) is used in this paper to improve the linearized model of a simple inverted pendulum (SIP). The MBDRNN´s equations start as those of the linearized SIP model. Then, through back-propagation-based training, the MBDRNN´s activation functions´ weights are modified with the objective of improving the linearized SIP model. Simulation results show that the MBDRRN effectively improved the linearized model. By tuning several of the MBDRNN parameters, an improved configuration was found yielding a satisfactory´ small modeling approximation error.

Keywords

backpropagation; linear systems; nonlinear dynamical systems; pendulums; recurrent neural nets; back-propagation-based training; inverted pendulum; linearized model; model-based dynamic recurrent neural network; simple inverted pendulum; Approximation error; Computer networks; Mean square error methods; Modeling; Neural networks; Nonlinear dynamical systems; Nonlinear equations; Nonlinear systems; Recurrent neural networks; Systems engineering and theory;

fLanguage

English

Publisher

ieee

Conference_Titel

System Theory, 2005. SSST '05. Proceedings of the Thirty-Seventh Southeastern Symposium on

ISSN

0094-2898

Print_ISBN

0-7803-8808-9

Type

conf

DOI

10.1109/SSST.2005.1460881

Filename

1460881