Efficient Optimization of Complex Products : A Simulation and Surrogate Model Based Approach

Abstract: This thesis investigates how to use optimization efficiently when complex products are developed. Modelling and simulation are necessary to enable optimization of products, but here it is assumed that verified and validated models of the products and their subsystems are available for the optimization. The focus is instead on how to use the models properly for optimization.Knowledge about several areas is needed to enable optimization of a wide range of products. A few methods from each area are investigated and compared. Some modifications to existing methods and new methods are also proposed and compared to the previous methods.These areas includeOptimization algorithms to ensure that a suitable algorithm is used to solve the problemMulti-Objective Optimization for products with conflicting objectivesMulti-Disciplinary Optimization when analyses from several models and/or disciplines are neededSurrogate Models to enable optimization of computationally expensive modelsModern frameworks for optimization of complex products often include more than one of these areas and this is exemplified with the industrial applications that are presented in this thesis, including the design and optimization of industrial robots and aircraft systems.