Abstract
The paper proposes a frequency domain identification method for linear continuous-time systems including delay ones. The model parameters are estimated by minimizing the cost functions which are appropriate for open/closed loop. The minimization is achieved through Particle Swarm Optimization, which attracts a lot of attention recently in the evolutionary computation area due to its empirical evidence of its superiority. Its effectiveness is demonstrated by numerical examples. In addition, a comparison with the existing optimization method is given to show the robustness of the proposed method. Furthermore, an experimental evaluation using a magnetic levitation system is performed.
