An ANN Assisted Reverse Recovery of Diode Model for Switchingon Characteristics of IGBT Devices
Conference: PCIM Europe 2023 - International Exhibition and Conference for Power Electronics, Intelligent Motion, Renewable Energy and Energy Management
05/09/2023 - 05/11/2023 at Nürnberg, Germany
doi:10.30420/566091012
Proceedings: PCIM Europe 2023
Pages: 9Language: englishTyp: PDF
Authors:
Zhang, Huaiyuan; Lee, Steven (Keysight Technologies, USA)
Shih, Abby (Keysight Technologies, Germany)
Haensel, Stefan; Umar, Zeeshan; Zeyss, Felix (Siemens, Germany)
Abstract:
This paper presents a new Artificial Neural Network (ANN) assisted in reverse recovery of a diode model for insulated-gate bipolar transistor (IGBT) devices at the turn-on process. The dependence of both carrier lifetime (τ) and diffusion transit time (TM) on current, voltage and temperature in a diode model is modeled by ANN after a training process. The proposed diode model implemented in the Keysight IGBT model can accurately fit the transient iC of an IGBT at a turn-on process for a wide range of currents at three voltages (100, 500 and 600 V) and three temperatures (25, 75 and 100 °C). With a new parameter extraction workflow for transient simulation, an overall good fitting of the switching-on characteristics is achieved for an IGBT power module.