A GENERALIZED APPROACH TO GENERATE THE COMPONENT MODELS FOR MULTI-OBJECTIVE OPTIMIZATION IN POWER ELECTRONIC APPLICATIONS

Abstract

Multi-objective optimization (MOO) in the area of power electric applications requires complete and reliable component models. This contribution presents a simple and standardized method to generate the component models. After a short introduction, the stepwise approach is presented using the example ‘dc-link capacitor’. The problem of reliable component prices at low minimum order quantities (MOQ) is addressed and a concept called ‘linear-price estimation’ is introduced to solve the problem. Results of the models are shown and discussed. It is shown that small errors between training data and the models can be achieved for the cost, size and efficiency models with the approach. Finally, the approach is classified in the research field of MOO and it is briefly outlined that the approach can also be applied to other components (e.g. semiconductors).