Design and analysis of computer experiments for efficient model-based active thermography in the agro-food sector

Abstract

Finite element models simulating the heat transfer in active thermography are studied intensively. Together with improved model accuracy, the related complexity and required computational power are increasing. In the agro-food sector, the material properties of the inspected objects, and thus the input parameters of heat transfer models, are prone to biological variability. In this work, a strategy called 'Design and Analysis of Computer Experiments' was applied to efficiently obtain the heat transfer output for an arbitrary combination of input parameters without running every single simulation. A metamodel was set up, interpolating between the output of a few carefully selected simulations.