A Self-Organizing Map-Based Monitoring System for Insulated Gate Bipolar Transistors Operating in Fully Electric Vehicle

Abstract


 
 
 Insulated Gate Bipolar Transistors (IGBTs) are one of the most used power semiconductor devices for energy conversion applications, due to their high performance. In this work we have developed a monitoring system for IGBTs installed in Fully Electric Vehicles (FEVs), which are operating under very variable working conditions. The monitoring system is based on a Self-Organizing Map (SOM), trained considering data collected from healthy IGBTs. An indicator of the IGBT degradation is defined as the distance between the measured SOM input vector, i.e., the signal measured on the monitored IGBT, and its SOM Best Matching Unit (BMU) representative of an healthy IGBT in similar working conditions. Then, a method based on the definition of a utility function for the identification of the threshold value to be used for the classification of the IGBT degradation state is proposed. The approach is verified with respect to experimental data collected from an inverter connected to an electric motor, and is shown able to identify the IGBTs degradation state regardless of the actual operating condition.