欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
提出了基于Levenberg-Marquardt BP神经网络的纳米MOSFET量子更正模型，并对拥有不同隐层、不同隐层神经元数的网络的训练精度和速度进行了研究对比。结果表明，包含2个隐层的网络可以获得高的训练速度和精度。该模型可用于快速预测纳米MOSFET Si层各点载流子量子密度，并对其电容及漏电流进行量子更正，其结果与Schroedinger-Poisson方程的吻合度很高。
英文摘要：
Backpropagation neural networks using Levenberg-Marquardt algorithm are applied to make quantum correction to nanoscale MOSFETs and the efficiency and accuracy of the neural networks with different hidden layers and different neurons are studied. The research indicates that high speeds and accuracy can be obtained using neural networks with two hidden layers. The model can be used to predict quantum charge density in Si layers and make quantum correction to capacitance and drain currents of MOSFETs in very good agreement with Schroedinger- Poisson approach.
同期刊论文项目
同项目期刊论文
期刊信息
位置：