论文信息 - Artificial intelligence techniques for SPICE optimization of MOSFET modeling

Artificial intelligence techniques for SPICE optimization of MOSFET modeling

This paper proposes new method for optimize and verified electric characterization graph of MOSFET by using artificial neural network. Optimization using Neural Network (ONN) will compare current-voltage (I–V) Characteristic graph between the TCAD simulation and TSPICE modeling as desire data control a model parameter of BSIM. In this paper, the neural network method is dynamic feedforward Neural Network. After NN training, the best result is at Neural Network architecture of 36-30-10-5 with Mean Squared Error (MSE) of 1e-28 at epoch of 5.

[1] Mauro Mongiardo,et al. A review of artificial neural networks applications in microwave computer-aided design , 1999 .

[2] Q.J. Zhang,et al. A Neural Network Approach to the Modeling of Heterojunction Bipolar Transistors from S-Parameter Data , 1998, 1998 28th European Microwave Conference.

[3] Michel Nakhla,et al. A neural network modeling approach to circuit optimization and statistical design , 1995 .

[4] Peter B. L. Meijer,et al. Fast and smooth highly nonlinear multidimensional table models for device modeling , 1990 .

[5] Fang Wang,et al. Knowledge based neural models for microwave design , 1997, 1997 IEEE MTT-S International Microwave Symposium Digest.

[6] Kurt Hornik,et al. Degree of Approximation Results for Feedforward Networks Approximating Unknown Mappings and Their Derivatives , 1994, Neural Computation.

[7] S. Dimitrijev. Principles of semiconductor devices , 2005 .