Optimal Design of Conical Horn Antenna Based on GP Model with Coarse Mesh

Gauss process (GP) is a learning machine which has developed rapidly in recent years. Compared with the methods of artificial neural network and support vector machine, GP is easy to implement and has the advantages of adaptive acquisition and predictive output. This paper presents a modeling method based on GP. When constructing the proposed GP model, the input samples of the model calculates the results of electromagnetic simulation software with coarse mesh, and the corresponding output samples are these of electromagnetic simulation software with precise mesh. Based on the proposed GP model exploiting particle swarm optimization (PSO) algorithm, a dual-mode conical horn antenna is optimized, and the optimization results are perfect. The modeling process and computing results show that the proposed GP model based on the coarse mesh can greatly reduce the mapping relation between the input and output, and its generalization ability is excellent. Moreover, in the optimization process, the trained GP model can be used to evaluate the fitness function of PSO, and the time required for optimization is obviously reduced because it can give the fitness function value rapidly.