Multi-objective Optimal Dimensioning of Modular Power Converters

Abstract: This master thesis was carried out at Lund University in collaboration with the European Spallation Source (ESS). It brings up the question of whether it’s possible to find an optimal solution for complex power electronic topologies, using a well formulated global optimization methodology. The commercially available software tool MATLAB was adopted for these studies, including its own toolbox for optimization. These tools and associated solvers will be described for the optimal design of two different problems:- a power inductor and a multilevel DC/DC power converter. In both cases, the objective function consists of the sum of the capital costs (price of the systems in question) and of the long term operating costs (power losses and maintenance cots). A set of constraints to which the design should comply with is also defined for each case and these will comprise parameters like maximum volume of the system, minimal output performance (value of inductance, maximal saturation field density, maximal current density for the first example or output voltage ripple, volume for the second example), etc. The methodology consists of five steps, the first one is to describe the problem to be optimized. In the second step all the data and information needed to give an accurate view of the problem is to be collected. The third step contains a description of the problem with an objective function that should be optimized. The fourth step corresponds to the definition of the boundaries or constraints. Finally, the fifth step consists on the evaluation of the obtained results.