Modeling of MOSFET Transistor by MLP Neural Networks
Modeling of MOSFET Transistor by MLP Neural Networks
复制标题
通过 MLP 神经网络对 MOSFET 晶体管进行建模
DOI：
--
复制
发表时间：
2016
期刊：
影响因子：
--
通讯作者：
S. Berrah
中科院分区：
文献类型：
--
作者：
K. Lamamra;S. Berrah
本文采用神经网络对MOSFET晶体管进行建模，神经网络模型的结构和训练采用遗传算法，该算法以使实际测量值与神经网络模型值之间的差异最小为目标进行进化。神经元模型由三层组成，输入层有两个神经元，一个用于漏极电压，另一个用于我们想要最小化的误差;输出层用于神经模型的漏极电流，隐藏层具有不同数量的神经元。在应用这种方法后，找到了几个模型，并提供了一个非常低的建模误差，所获得的结果是非常令人满意的。
In this paper neural networks are used to model a MOSFET transistor, the structure of the neural model and its training are performed by the genetic algorithm which evolves to minimize the difference between the desired values resulting from practical measurements and the neural model values. The neuronal model consists of three layers, an input layer with two neurons, one for the drain voltage and the other to the error that we want to minimize; an output layer for the drain current of the neural model and a hidden layer with varying number of neurons. After applying this approach, several models are found and that offer a very low modeling error, and the obtained results are very satisfactory.