HHGA-RBFNN在DGS设计中的应用  

Application of HHGA-RBFNN in Design of Microstrip DGS Structure

作　　者：杜道鹏[1] 高卫东[1] 李乐[1] 

机构地区：[1]合肥电子工程学院,合肥230037

出　　处：《现代雷达》2010年第5期60-63,共4页Modern Radar

摘　　要：将递阶遗传算法和最小二乘法相结合,构成一种混合递阶遗传算法,用于优化径向基函数神经网络的结构和参数,通过MATLAB仿真实现该算法,并将该神经网络运用于微带缺陷接地结构的神经网络建模之中,实现对微带缺陷接地结构传输系数的仿真。实验结果表明,应用混合递阶遗传算法优化的径向基函数神经网络,具有很高的拟合精度和很好的泛化能力,可以与时域有限差分分析方法结合,辅助微带缺陷接地结构的设计。A novel hybrid hierarchy genetic algorithm was introduced through combining hierarchy genetic algorithm and least square method. This hybrid algorithm could determine the structure and parameters of the radial basis function neural network, and the program is implemented using MATLAB language. This neural network is then applied to the modeling of the microstrip defected ground structure, and the transmission coefficient of the microstrip defected ground structure can be obtained from the neural network model. The results of simulation demonstrate that the radial base function neural network based on hybrid hierarchy genetic algorithm has very high fitting precision and good reforming ability, and this neural network model can be used as a kind of aid tool for the design of microstrip defected ground structure in conjunction with the FDTD method.

关 键 词：混合递阶遗传算法 径向基函数神经网络 缺陷接地结构 传输系数 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] TN817[自动化与计算机技术—控制科学与工程]