A microstrip antenna size optimization method based on K-Nearest Neighbors (KNN) and Artificial Neural Network (ANN) algorithms is proposed to solve the problem of high optimization complexity of traditional antennas. By analyzing the surface current distribution of the antenna, high-sensitivity parameters are set as variables, while low-sensitivity parameters are set as constants. The KNN algorithm and ANN algorithm are then utilized to optimize the size parameters of the antenna, ultimately enhancing broadband performance. To validate the effectiveness of the optimization algorithms, two antennas were fabricated and tested. The results indicate that compared to traditional antenna design methods, the KNN and ANN algorithms increase the impedance bandwidth by 20.8% and 18.4%, respectively. Although the ANN algorithm requires longer training time, it demonstrates significant improvements in impedance matching across multiple frequency bands.