Reliability Assessment of 6.5 kV IGBTs Considering Device Aging Using the Markov Chain Monte Carlo Method

Aiming at the reliability problem of 6.5 kV IGBT power devices in flexible direct current transmission systems under high stress, a reliability assessment framework integrating the analytical life model and the statistical sequence method is proposed. By modeling losses across multiple converter operating modes and applying a Foster thermal network, junction temperature fluctuations are calculated. The rain-flow method is used to extract thermal cycling loads for a life model. Considering parameter dispersion and statistical variability, Monte Carlo sampling is adopted to obtain probabilistic lifetime distributions, which transforms the device life distribution from a fixed value to a probabilistic distribution, approaching the actual situation. To account for long-term parameter aging and system uncertainties, the Bayerer model is optimized via Markov Chain Monte Carlo (MCMC), correcting thermal impedance drift. Case results show that device T2 suffers the highest annual life loss (0.012), with an expected life of 81.16 years. After aging correction, the lifetime fits a Weibull distribution (scale 80.898 years), and the cumulative failure probability reaches 10% after 62.7 years. This method provides theoretical support for the reliability design and operation and maintenance strategies of high-voltage IGBT systems.