A Digital Twin Approach Toward Lifetime Analysis and Predictive Maintenance of Power Semiconductors for Railway Application

Abstract:
This paper proposes an approach to electrical and thermal modeling of a high-voltage power semicon-ductor module with a view to set up a digital twin representation for assessing degradation and remaining life of component in railway applications. Based on experimental results obtained at test bench level where optic fibers and Negative Temperature Coefficient (NTC) thermistors have been used to monitor internal temperatures of the power semiconductor module, this paper will focus on an example of fitting methodology between digital twin and real application data from tests. An application of optimization algorithm, more specifically a genetic algorithm, will be described. Results and limitation will be then discussed.