基于神经网络的新型缺陷接地结构优化设计
Optimization Design of a Novel Defected Ground Structure Based on Neural Network
-
摘要: 应用人工神经网络与单纯形优化算法相结合的方法,对一种新型组合式非周期性缺陷接地结构(CNPDGS)进行优化设计.与电磁场数值分析方法相比,以神经网络模型作分析单元,可以在保证精度的基础上大大提高分析速度,因此在优化设计中可用来替代FDTD分析方法作为结构分析的计算单元.本文中以所要求的传输系数为期望日标,以可以使误差函数达到极小化的结构尺寸为输出,经单纯形优化算法寻优,进行该具有双阻带特性CNPDGS的优化设计.仿真设计和实验的对比结果表明了这一方法的有效性.Abstract: The optimization design of novel combinatorial nonperiodic defected ground structures(CNPDGS) is accomplished applying artificial neural network(ANN) and Nelder-Mead simplex optimization algorithm.Compared with electromagnetic numerical analysis,regarding the ANN as analysis unit may bring great advantage in speed meanwhile the accuracy guaranteed.Therefore the ANN can take the place of FDTD as the calculational unit in the structure analysis.In the optimization design of the CNPDGS with the characteristic of dual stop band,dimensions that minimize the error function are obtained by simplex algorithm according to the desired parameters of transmission coefficient in the paper.The comparisons between computer simulation and experimentation demonstrate the effectiveness of the developed approach.