Study of brushless excitation system parameters estimation based on improved genetic algorithm

For estimating the parameters of brushless excitation system including nonlinear blocks, in this paper, a new optimization method based on improved genetic algorithm (IGA) is developed. A practicable test method which performs the linear block parameters and nonlinear parameters estimation respectively is presented. First, the model of brushless excitation system of AC1A developed by IEEE is presented. Then an improved GA based on real-coding is detailed. In this method a modified linear cross operator is used whose merits lies in both of child generation are feasible and one must be in the vicinity paternal generation so the convergence and precision of new method are improved. Finally, the single unit simulation system and AC1A excitation system are built in Matlab/Simulink. Utilizing the improved GA the linear block parameters and nonlinear block parameters of AC1A excitation system are estimated respectively using M serial pseudo random binary signal (PRBS) and reference voltage step signals. Compared with results received from simple GA the conclusions can be obtained that the improved GA is effective and the test method is practicable.