IGBT Junction Temperature Estimation Based on Machine Learning Method

In IGBT power module, nearly 55% of the decrease of device reliability is caused by the increase of junction temperature. IGBT junction temperature estimation is a useful way to minimize failure rate. In this paper, three nonlinear machine learning-based models including Support Vector Regression (SVR), Random Forest (RF) and Back-Propagation Neural Network (BPNN) are constructed to automatically estimate the junction temperature under different parameter values. The simulation results are compared and prove that this machine learning method can be used and have a good prospect of industrial application.

[1]  Mohsen Vatani,et al.  State of Health Prediction of Li-ion Batteries using Incremental Capacity Analysis and Support Vector Regression , 2019, 2019 IEEE Milan PowerTech.

[2]  Hui Li,et al.  A Novel IGBT Health Evaluation Method Based on Multi-Label Classification , 2019, IEEE Access.

[3]  Min Ren,et al.  A Novel Virtual Sensing With Artificial Neural Network and K-Means Clustering for IGBT Current Measuring , 2018, IEEE Transactions on Industrial Electronics.

[4]  Yennun Huang,et al.  Application of BPNN algorithm to learning high dimensional features for solving induction motor defect problems , 2017, 2017 International Conference on System Science and Engineering (ICSSE).

[5]  C. Damásio,et al.  Stator winding short-circuit fault diagnosis in induction motors using random forest , 2017, 2017 IEEE International Electric Machines and Drives Conference (IEMDC).

[6]  Shalabh Gupta,et al.  Fault diagnosis in electric drives using machine learning approaches , 2013, 2013 International Electric Machines & Drives Conference.