Genetic algorithms as a tool for phased array radar design

Abstract

The United States Navy needs creative ways to design multi-function phased array radars. This thesis proposes that Genetic Algorithms, computer programs that mimic natural selection to arrive at innovative solutions to complex problems, would be particularly well suited to this task. The ability of a Genetic Algorithm to properly predict the behavior of an array antenna with randomly located elements was examined with encouraging results through the construction and measurement of a test array. Comparison of test data to Genetic Algorithm and Method of Moments calculations showed significant qualitative agreement in the antenna test patterns of a thin, randomly distributed array. Areas of disagreement between the test article pattern and the calculated ones were traced to systematic errors in the anechoic chamber and alignment error during antenna positioning. The final experiment to demonstrate beam steering was not completed due to lack of time and poor response of mechanical phase shifters. Despite the inability to demonstrate beam steering, the early experiments demonstrate the significant potential for using Genetic Algorithms for complex shipboard phased array radar antenna design.