基于PSO-BP算法的微带天线谐振频率神经网络建模

基于PSO-BP算法的微带天线谐振频率神经网络建模

内容简介：谐振频率是微带天线设计过程中最重要的一个参数,直接决定设计的成败.本文提出基于粒子群优化(PSO)算法和反向传播(BP)算法的一种混合算法(PSO-BP)来训练神经网络,并基于该网络对矩形微带天线的谐振频率进行建模.该算法充分利用了PSO的全局搜索特性和BP的局部搜索特性,可以有效地提高神经网络的建模精度.仿真试验表明,基于PSO-BP算法的神经网络所建立的微带天线的谐振频率模型好于此问题的已有结论.

Abstract：Resonant frequency is an important parameter in the design process of microstrip anten-na (MSA). A hybrid algorithm fused particle swarm optimization (PSO) and back propagation (BP) together is proposed in this paper, and model of resonant frequency of rectangular MSA is established in terms of the network. The algorithm fuses effectively the global search characteris-tic of PSO and local search characteristic of BP, and it can improve the model's accuracy definite-ly. Experiment shows that the model of resonant frequency of MSA established by use of the al-gorithm is better than available ones.

作者：董跃, 田雨波,

关键词：微带天线, 谐振频率, 神经网络, 粒子群优化, 反向传播,