【摘要】 周期性缺陷接地结构(PDG S)是一种微波电路的周期性“缺陷”结构,存在明显的频率阻带特性。结构尺寸较大的PDG S,传输系数曲线会出现不同程度的微小抖动,为了减小抖动对神经网络训练样本的影响,文中采用最小二乘法和五点三次平滑法对存在不规则抖动的传输系数曲线进行了预处理,并将预处理过的传输系数作为训练样本对神经网络进行训练,结果表明,神经网络的预测结果和FDTD法计算结果一致性很好,准确反映了传输特性的变化趋势。更多还原

【Abstract】 The periodic defected ground structures(PDGS) are the structures that are etched periodically on the circuit’s ground plane.PDGS can be used to prohibit the propagation of electromagnetic waves within a certain band of frequencies.For the PDGS with biggish structure sizes,the transmission coefficients have the tiny dither owing to abnormal disturbance.In order to reduce the effect of abnormal disturbance on the training samples for ANN model,the transmission coefficients with abnormal disturbance have been smoothed by the least square method and five-spot triple smoothing in this paper.The results of the ANN model trained by the preprocessed training samples are agreement with the results of FDTD,and reflect the trend of the transmission coefficients well and truly.更多还原