An Efficient Antenna Optimization Method Based on a Parameter Dimensionality Reduction Approach

As antenna structure grow in complexity, the problem of time-consuming in electromagnetic (EM) simulation have become more common. This problem undoubtedly brings a series of the challenges in raising the effectiveness of antenna optimization. In order to overcome the existing difficulties, an antenna optimization method based on RF-GWO method is proposed. This approach consists of three major components. Firstly, an artificial neural network (ANN) is used to build the dataset. Compared with the traditional sampling method, the data filtered by the artificial neural network (ANN) are closer to the optimization objective. Consequently, it can better responsive to the relationship between the antenna parameters and the optimization objective. Then, the Random Forest method (RF) is utilized to analyze the impact of the ten selected parameters based on the electromagnetic (EM) simulation results on the antenna performance. Random Forest Algorithm (RF) has strong interpretability, allowing for the assessment of the contribution of different parameters to the optimization objective, which can effectively reduce the number of design variables. Finally, the antenna optimization is performed utilizing the Grey Wolf Optimization (GWO) algorithm. Combined with the sensitivity analysis results, the Gray Wolf Algorithm (GWO) can effectively search for improved results in a faster manner. In order to evaluate the algorithm’s performance, the proposed method is employed to optimize a multiband miniaturized microstrip antenna. The result demonstrates that substantial improvement in impedance bandwidth is achieved by the proposed method, with an approximate 29% increase compared to conventional manual optimization methods. This outcome has proved potential and effectiveness of the proposed method in practical implications of antenna optimization.