Applying Multi-objective Optimization to Thermography : using the NSGA-II genetic algorithm

Abstract: In multi-objective system design there is rarely a clear path through the decisions an engineer hasto make since in practice, improving a single objective, or desired quality of the system, comes atthe cost of another objective. It is then the task of the engineer to match the requirements of thesystem to a balance in objective priority. Since this balance is typically difficult to quantify, thedesign options should be well understood prior to setting preference.This thesis uses a genetic algorithm in the engineering design process to generate a set offavorable design options, or the Pareto front of a system's desired behavior, prior to when theengineer decides which objectives are the most important. This optimization process searches foroptions that provide efficient configurations of the system regardless of objective priority. Theseresults are fundamental to making well informed multi-objective decisions in engineering design.This analysis suggests that successful engineering design requires a clear understanding of asystem and how the system's desired behavior is measured. It is the author's hope that the readergains understanding on this relationship and what tools and best practices can be used to aid in amultiple objective scenario for engineering design.