|
|
Application of artificial neural network for switching loss modeling in power IGBTs |
Yan Deng*, Xiang-ning He, Jing Zhao, Yan Xiong, Yan-qun Shen, Jian Jiang |
School of Electrical Engineering, Zhejiang University, Hangzhou 310027, China |
|
|
Abstract The modeling of switching loss in semiconductor power devices is important in practice for the prediction and evaluation of thermal safety and system reliability. Both simulation-based behavioral models and data processing-based empirical models are difficult and have limited applications. Although the artificial neural network (ANN) algorithm has often been used for modeling, it has never been used for modeling insulated gate bipolar transistor (IGBT) transient loss. In this paper, we attempt to use the ANN method for this purpose, using a customized switching loss test bench. We compare its performance with two conventional curve-fitting models and verify the results by experiment. Our model is generally superior in calculation speed, accuracy, and data requirement, and is also able to be extended to loss modeling for all kinds of semiconductor power devices.
|
Received: 22 July 2009
Published: 02 June 2010
|
|
Fund: Project supported by the Power Electronics Science and Education Development Program of Delta Environmental & Educational Foun-
dation (No. DREO2006022) and the National Natural Science Foun-dation of China (No. 50737002)
|
Application of artificial neural network for switching loss modeling in power IGBTs
The modeling of switching loss in semiconductor power devices is important in practice for the prediction and evaluation of thermal safety and system reliability. Both simulation-based behavioral models and data processing-based empirical models are difficult and have limited applications. Although the artificial neural network (ANN) algorithm has often been used for modeling, it has never been used for modeling insulated gate bipolar transistor (IGBT) transient loss. In this paper, we attempt to use the ANN method for this purpose, using a customized switching loss test bench. We compare its performance with two conventional curve-fitting models and verify the results by experiment. Our model is generally superior in calculation speed, accuracy, and data requirement, and is also able to be extended to loss modeling for all kinds of semiconductor power devices.
关键词:
Insulated gate bipolar transistor (IGBT),
Switching loss,
Modeling,
Artificial neural network (ANN)
|
Viewed |
|
|
|
Full text
|
|
|
|
|
Abstract
|
|
|
|
|
Cited |
|
|
|
|
|
Shared |
|
|
|
|
|
Discussed |
|
|
|
|