【摘要】 频率选择表面是一种二维周期阵列结构,能够有效控制电磁波的传输和反射。为了解决传统设计方法参数选择的盲目性和有效性缺陷,提出一种基于进化模糊神经网络算法的设计方法。该方法具有开放的结构,可以在线自适应并不断进化,克服普通神经网络中模型结构和参数难以设置的缺点,同时系统可以进行模糊规则插入和规则提取等。仿真结果表明,该方法具有更高的准确度,能有效地解决频率选择表面设计工作中的一些相关问题。更多还原

【Abstract】 A FSS structure is usually composed of an array of periodic patches,which can effectively control the transmission or reflection of electromagnetic waves. Proposes an evolving fuzzy neural network approach to overcome the defects in arbitrary parameters selection and validity. With an open structure and self-adapting and evolving features, EFuNN algorithm avoids the difficulty in setting the structure and parameters of ordinary neural networks; meanwhile, the system is capable of rule insertion and extraction. The simulation result shows that EFuNN algorithm is able to solve FSS design issue with a higher accuracy.更多还原