The Pennsylvania State University The Graduate School Department of Electrical Engineering COMPARISON OF CAT SWARM OPTIMIZATION WITH PARTICLE SWARM OPTIMIZATION FOR IIR SYSTEM IDENTIFICATION A Thesis in Electrical Engineering by

Infinite impulse response (IIR) systems are widely used in modeling systems such as communications, bio-systems, acoustics, etc. Many algorithms in computational intelligence area have been developed to identify the systems with a novel search technique, but system identification is challenging due to non-unimodality of the error surface and the non-linear relationship between the error signal and the system parameters. Cat swarm optimization (CSO) was recently introduced to solve optimization problem with a new learning rule based on swarm intelligence to show better performance than particle swarm optimization (PSO). Also, it has been tried to be used for infinite impulse response (IIR) system identification. Optimum parameters are proposed to solve optimization problem. However, parameters adapted for the IIR system identification have not been investigated enough. This thesis examines the parameters of CSO in order to optimize them for IIR system identification with a few benchmarked IIR plants. Inertia-weighted PSO is used as a comparison for performance issue. The results demonstrate the better performance of the CSO than PSO.