This paper was produced for the 2019 NAFEMS World Congress in Quebec Canada

Resource Abstract

Decision making is a difficult and important step in design processes. It aims at guiding designer in the selection of design solutions between numerous alternatives. A specific process, derived from the OIA ontology and applied to vehicle embedded systems, is presented in this paper in order to find the optimal solution that responds to several demanded objectives. This process is based on an optimization algorithm, coupling models of both physical behaviors and designers’ subjective reasoning. Generally, the computations time of simulation forces designers to treat non-flexible design specifications, which generally leads to an absence of satisfactory solution or, in contrast, to a huge set of unclassified solutions. In order to decrease computations time and to improve decision making, this paper introduces a dynamic optimization process which adds flexibility to design approach in several different manners. This process lies on a dynamic vision of specifications, scenarios, client needs and preferences; it aims at integrating a machine learning algorithm in a global evolutionary optimization algorithm generating reduced models directly in an online mode with a continuous adaptation to changes in the knowledge of the system. This method is applied to optimize the powertrain of an electric car, which includes battery, inverter, electric motor and gearbox, responding to three major objectives: autonomy, performance and cost.