论文部分内容阅读
随着微电子工艺技术的发展,硅基CMOS器件的截止频率已经达到毫米波频段,使硅基微波单片集成电路实现成为可能。因此,建立硅基毫米波频段共面波导结构模型使准确设计硅基微波单片集成电路成为必要。文章提出了一种基于神经网络技术的共面波导结构(CPW)毫米波可缩放模型,采用3层神经网络结构,根据共面波导的测试结果,用神经网络来学习其物理变量和测试的相应S参数空间映射关系。仿真与测试结果比较表明:基于神经网络方法建立的毫米波共面波导可缩放模型对不同几何参数CPW能够快速和准确地给出对应的CPW的S参数结果。
With the development of microelectronic technology, the cut-off frequency of silicon-based CMOS devices has reached the millimeter-wave frequency band, making it possible to realize silicon-based microwave monolithic integrated circuits. Therefore, the establishment of silicon-based millimeter-wave band coplanar waveguide structure model to accurately design silicon-based microwave monolithic integrated circuits become necessary. In this paper, we propose a CPW millimeter-wave scalable model based on neural network technology, using a 3-layer neural network structure. According to the CPW test results, neural networks are used to learn the physical variables and the corresponding test S-parameter space mapping. The comparison between simulation and test results shows that the scalable model of CPW based on neural network method can give the corresponding SSP results of CPW for different geometric parameters quickly and accurately.