APPLICATION OF ARTIFICIAL NEURAL NETWORKS FOR THE MODELING AND CONTROL OF NONLINEAR SYSTEMS

Abstract

In this thesis, the wide area of recurrent neural networks is explored to propose a novel structure for the purpose of prediction and control of nonlinear systems with unknown dynamics. The proposed structure is a locally recurrent neural network with input feed through (LRNNIFT) which consists of locally recurrent loops along with input fed through weights directly to the output. The proposed network model parameters are tuned using a Back-propagation (BP) algorithm. The performance of the proposed model is compared with the state-of-the-art recurrent Elman neural network (ENN) and a single layer feed-forward neural network (FFNN). The simulation results showed that the proposed model has shown better accuracy as compared to the other two models. Furthermore, the above network is presented for control of nonlinear dynamical systems. The rationale of using LRNNIFT is due to its modest structure and proven superiority in mathematical modeling. Results from simulation showed that the LRNNIFT based controller is able to achieve adaptive control in a nonlinear system. It is also tested and observed to counterbalance the effects of disturbances. A comparative analysis is presented with the help of simulation, and it is deduced that overall performance of the LRNNIFT controller is better than that of FFNN and ENN controllers.