An Improved Taguchi Algorithm Based on Fitting and Prediction for Linear Antenna Array Synthesis

Abstract

In this paper, a Taguchi method based on fitting and prediction is proposed to accelerate the optimization process in antenna array synthesis. The implementation procedure combines the normal Taguchi method and the curve fitting technique. A possible solution is determined by prediction based on fitting curves. Specifically, the fitting curves are obtained by using the dynamic points calculated and updated as the Taguchi method progress and recorded in the response table necessarily produced in the procedure. Test functions are used for conducting some confirmation experiments, and the results verify the validity of the proposed method. In order to illustrate its good practicability, two linear antenna arrays with a null controlled pattern and a flat top pattern, respectively, are successfully optimized by using both of the normal Taguchi method and the proposed one. Some comparisons and discussions of their results are given in the paper, which proves that the proposed method has a better practicability, not only because it inherits the global optimization characteristics of the normal Taguchi method but also because it accelerates the convergence process.