Wang Xin, Zhou Zhenwei, He Shilie, Liu Junbin, Cui Wei
School of Automation and Engineering, South China University of Technology, Guangzhou 510641, China.
China Electronic Product Reliability and Environmental Testing Research Institute, Guangzhou 511370, China.
Micromachines (Basel). 2023 Apr 28;14(5):959. doi: 10.3390/mi14050959.
The problem of health status prediction of insulated-gate bipolar transistors (IGBTs) has gained significant attention in the field of health management of power electronic equipment. The performance degradation of the IGBT gate oxide layer is one of the most important failure modes. In view of failure mechanism analysis and the easy implementation of monitoring circuits, this paper selects the gate leakage current of an IGBT as the precursor parameter of gate oxide degradation, and uses time domain characteristic analysis, gray correlation degree, Mahalanobis distance, Kalman filter, and other methods to carry out feature selection and fusion. Finally, it obtains a health indicator, characterizing the degradation of IGBT gate oxide. The degradation prediction model of the IGBT gate oxide layer is constructed by the Convolutional Neural Network and Long Short-Term Memory (CNN-LSTM) Network, which show the highest fitting accuracy compared with Long Short-Term Memory (LSTM), Convolutional Neural Network (CNN), Support Vector Regression (SVR), Gaussian Process Regression (GPR), and CNN-LSTM models in our experiment. The extraction of the health indicator and the construction and verification of the degradation prediction model are carried out on the dataset released by the NASA-Ames Laboratory, and the average absolute error of performance degradation prediction is as low as 0.0216. These results show the feasibility of the gate leakage current as a precursor parameter of IGBT gate oxide layer failure, as well as the accuracy and reliability of the CNN-LSTM prediction model.
绝缘栅双极型晶体管（IGBT）健康状态预测问题在电力电子设备健康管理领域受到了广泛关注。IGBT栅氧化层的性能退化是最重要的失效模式之一。鉴于失效机理分析和监测电路易于实现，本文选取IGBT的栅极漏电流作为栅氧化层退化的前驱参数，并采用时域特征分析、灰色关联度、马氏距离、卡尔曼滤波等方法进行特征选择与融合。最终得到一个表征IGBT栅氧化层退化的健康指标。利用卷积神经网络和长短期记忆（CNN-LSTM）网络构建了IGBT栅氧化层的退化预测模型，在我们的实验中，该模型与长短期记忆（LSTM）、卷积神经网络（CNN）、支持向量回归（SVR）、高斯过程回归（GPR）以及CNN-LSTM模型相比，拟合精度最高。在NASA-艾姆斯实验室发布的数据集上进行了健康指标的提取以及退化预测模型的构建与验证，性能退化预测的平均绝对误差低至0.0216。这些结果表明了栅极漏电流作为IGBT栅氧化层失效前驱参数的可行性，以及CNN-LSTM预测模型的准确性和可靠性。
