Influence of Reducing the Load Level of Mission Profiles on the Remaining Useful Life of a TO220 Analyzed with a Surrogate Model

A methodology for replacing finite element simulations with a fast-calculating surrogate model for fault tolerance in operating systems is presented. The study focuses on the TO220 rectifier system and explores methods to detect impending failures and calculate the resulting necessary load reduction. The finite element simulation model is described, highlighting the die attach as the relevant connection for failure. A surrogate model is developed using long-short-term-memory models to predict temperature and in-elastic strain. The surrogate model significantly reduces simulation time, allowing for the adjustment of load based on the system's current state of health. The rainflow counting algorithm is applied to calculate the number of cycles to failure, and the Palmgren-Miner linear damage accumulation relation is used to determine the damage and state-of-health. The dependency of the change in lifetime due to variations in scaling factor is evaluated and the results show that load reduction increases the lifetime of the system.