论文部分内容阅读
运用人工神经网络技术建立了高电子迁移率晶体管(HEMT)的大信号模型。通过脉冲I-V测试和测量不同偏置条件下的S参数,获得了大信号等效电路模型中寄生参数和非线性本征元件的数值。通过BP神经网络,利用偏置相关的非线性元件值作为训练样本,利用误差反向传播的Levenberg-Maquardt方法训练神经网络并得到了网络权重数据。模型中的非线性元件在CAD软件中用神经网络实现,并将权重数据和CAD软件结合进行仿真。测试和仿真结果表明模型具有很高的精度。
A large signal model of high electron mobility transistor (HEMT) is established by artificial neural network technology. Through the pulse I-V test and measurement of S parameters under different bias conditions, the values of parasitic parameters and nonlinear eigen-elements in the large-signal equivalent circuit model are obtained. The BP neural network is used to train the neural network using the non-linear component values of the bias as the training samples and the Levenberg-Maquardt method of error back propagation is used to obtain the network weight data. The nonlinear elements in the model are implemented in the CAD software using neural networks, and the weight data and CAD software are combined to simulate. The test and simulation results show that the model has high accuracy.