Compact Genetic Algorithm Based on Roulette Selection Operator for Synthesizing thinned Arrays

In thinned array synthesis, it is easy to fall into local optimum using non-global optimization strategies due to the lack of global search capability. On the other hand, synthesis using stochastic optimization algorithms that facilitate global search, such as genetic algorithms, can easily fail in large arrays with thousands of elements. This paper proposes an optimization algorithm based on the compact genetic algorithm (cGA) to realize robust global search of thinned array synthesis that is not limited by the array size. Compared with cGA, the proposed algorithm reduces the computational effort by 17.9%/ 73.8% /87.4%, respectively, in optimizing fully populated arrays of 100/500/ 1000 elements. Multiple results show that the global search ability of the proposed algorithm is better than that of the stochastic optimization algorithms.