Designing an Optimal Test Plan for Accelerated Life Models Through Simulation

Owen McGrath, University of Limerick

Co-authors: Kevin Burke, University of Limerick

Abstract: In this work, we discuss a simulation-based framework for optimising test plans for accelerated life models. Test plans (experimental designs) are optimised by utilising the differential evolution (DE) algorithm, where the inputs to the objective function are the test plan parameters, and the output is the RMSE of out-of-sample (extrapolated) model predictions. Our simulation framework offers an alternative to theoretical optimisation, and is particularly favourable for complex designs where analytical optimisation could prove intractable.