Neural-Genetic Control Algorithm for Two-Link Robot

Abstract

This paper deals with soft state control of non-linear dynamic model – robot. Soft methods based on neural networks and genetic algorithms demonstrate powerful problem solving ability. They are based on quite simple principles, but take advantage of their mathematical nature: non-linear iteration computation solutions. One of the ways of control of such non-linear systems is the use of neural networks as an effective controllers. In this paper a new methodology is proposed, where for neural controller structures and parameters are computed by the genetic algorithm (GA). The proposed approach is represents by direct neural controller using multilayer perceptron (MLP) network in feedback tracking control loop. The training method using GA allows find optimal adjustment of neural network weights so that high performance is obtained. The proposed control method is realized in Matlab/Simulink and demonstrated on typical non-linear systems (two-link robot).