Abstrakt

A Method for System Order Reduction based on Genetic Algorithm and FFT

Farzin Asadi, Nurettin Abut

Modelling a complicated system, generally needs a high order differential equation. Designing or analysing based on such a high order equaion is cumbersome and time consuming. One idea is to use system order reduction techniques. Reduced order model must preserve original systems characteristics. Plenty of methods are proposed in the recent 50 years. Here, a method based on Genetic algorithm(GA) and Fast Fourier Transform(FFT) is suggested to the problem of system order reduction. First, a reduced order model structure must be chosen. Then, GA change Parameters of chosen model, i.e. coefficients of transfer function, in order to reach the best response(minimum cost function). In this paper a new cost function based on FFT coefficients are defined. Using this cost function and GA optimum coefficients of chosen reduced order transfer function are find. Proposed method can be used to reduce order of controllers designed by H� method due to the large order of such controllers.