DUBLIN CITY UNIVERSITY SCHOOL OF ELECTRONIC ENGINEERING System Identification using Neural Networks Optimised with Genetic Algorithms

Acknowledgements I would like to thank my supervisor Jennifer Bruton for her guidance, enthusiasm and commitment to this project. Declaration I hereby declare that, except where otherwise indicated, this document is entirely my own work and has not been submitted in whole or in part to any other university. Abstract Identification is the process of modelling of a system based on its inputs and outputs. Many techniques have been developed to achieve this. The most common identification techniques are based on linear approximations of the system, and perform well for a large variety of processes, both linear and non-linear. Complex systems, however, challenge basic linear models, and more sophisticated identification techniques are required. Neural networks have been shown to out-perform traditional identification techniques on more complicated problems. They are, unfortunately, not without their problems, the design of neural networks being based largely on trial and error. Genetic algorithms have recently been applied to the design of neural networks. Based on evolution, genetic algorithms lead a more directed search than a random procedure, while still exploring the entire search space. This work looks at the identification of the 'cart pole' system, a non-linear dynamic, unstable system. Traditional linear methods, neural networks and neural networks optimised using genetic algorithms are applied to this problem.