An enhanced diploid genetic algorithm (GA) is introduced for optimizing antenna arrays. Initially, several indicators are developed to measure the population state, thereby increasing the algorithm's responsiveness to variations in the antenna scheme. Subsequently, prevalent issues in antenna optimization are analyzed, leading to the introduction of a diploid GA aimed at preserving the diversity of antenna schemes, especially in a smaller population. This approach not only amplifies the exploration capacity of the algorithm but also addresses the issue of extensive simulation time. Furthermore, a local radial basis function (RBF) network is implemented for the assessment of some high-quality individuals, which effectively reduces the simulation count. In this method, the position of each individual in the solution space is considered the centroid, around which samples are selected for each antenna scheme to construct an individual RBF network. This technique simplifies the correlation between antenna parameters and performance, consequently decreasing the required sample size. Additionally, a local evolution acceleration mechanism is introduced to increase the convergence rate. The efficacy of the enhanced diploid GA is demonstrated through both test function experiments and a real-world application, showcasing its capability to efficiently optimize antenna arrays.