A GRU-Based Method of IGBT Module Degradation Prediction Under Changing Working Conditions

Aiming at the problem that the traditional degradation prediction method of insulated gate bipolar transistor (IGBT) modules is difficult to be effectively applied under changing working conditions, a IGBT module degradation prediction method suitable for changing working conditions is proposed. Firstly, the aging test of IGBT modules is carried out by using a modified variable working conditions power cycle test platform. Secondly, the aging feature parameters and working condition influence parameters are selected through analyzing. Then, a degradation prediction method for IGBT modules under changing working conditions is presented based on Gated Recurrent Unit (GRU). Finally, the dataset collected by the test platform are used to verify the proposed method. The results show that compared with Long-Short-Term Memory (LSTM) network, GRU has lower complexity and greater efficiency. Meanwhile, the prediction accuracy of GRU is almost the same as LSTM.