An Improved KNN Method for Antenna Optimization

The traditional antenna optimization process usually consumes lots of time and the result maybe unsatisfactory, which could be improved with the machine-learning algorithm. In this paper, a machine-learning-assisted optimization (MLAO) algorithm is proposed based on the K-nearest neighbor (KNN) method, which can simultaneously optimize multiple objectives of antenna, e.g., s11 and gain. The MLAO algorithm can select the best K value of KNN method to realize better optimization results. To verify the effectiveness of this algorithm, a circularly polarized antenna is designed and optimized by the proposed MLAO algorithm, which has better results and less optimization time than engineer traditional methods.