ANN based intelligent controller for inverted pendulum system

Abstract

Artificial neural network as an artificial intelligence technique is used in this paper to control the angle with position of a non- linear inverted pendulum system. The ANN controller here is a specified three layered feed forward network having, Input, Hidden and Output layers. 'Trainlm' network function that is used to update weights and bias states according to Levenberg- Marquardt (LM) back-propagation, is used here for training. The dynamic modelling of Inverted pendulum system is done using Euler-Langerange equation and the main task is to control the angle with position of non-linear system by the neural network controller and to compare the response with response of conventional controller using MATLAB simulations and to show the improved response. © 2013 IEEE.