Nonparametric Model-Based Online Junction Temperature and State-of-Health Estimation for Insulated Gate Bipolar Transistors

Abstract

Insulated gate bipolar transistor (IGBT) is widely used in power equipment, it generally works in complex circuit profiles and it is very difficult to measure or predict the thermal parameters of the module in real-time and evaluate the corresponding health status in the transient process. This paper develops a novel approach for solder-layer condition monitoring of IGBTs. In the approach a time-series nonparametric model of a power module is constructed, the current power and ambient temperature data are used to deduce the health state junction and case temperature. Three groups of time-series insulated gate bipolar transistors (IGBTs) data are used to train and verify the time-series nonparametric model for online conditions, the results show that the developed method has high accuracy. Compared with traditional methods, the time series non-parametric model method not only saves characteristic experiments but also saves the process of mathematical model construction. Besides, the proposed method also has the advantages of strong generalization and low equipment requirements which is useful for actual working conditions. Thereafter, another nonparametric model is built, the predicted junction temperature is used to estimate the collector voltage in the health state, and the percentage deviation of the measured collector voltage from the estimated voltage is used to do the state-of-health estimation of the IGBT and its accuracy is verified by the experiment result.

Keywords

• IGBT
• time-series ANN
• state-of-health
• junction temperature
• artificial intelligence