COMPARISON OF SELECTION METHODS OF GENETIC ALGORITHMS FOR AUTOMATED COMPONENT-SELECTION OF DESIGN SYNTHESIS WITH MODEL-BASED SYSTEMS ENGINEERING

Abstract

One of the important tasks of design synthesis with the Model-Based Systems Engineering (MBSE) is the component-selection. A trade study analysis is commonly used to perform this task, but when it is used for a complex system such as a hybrid car, the analysis will be error-prone, time and cost-consuming. A Genetic Algorithm (GA) is an evolutionary searching technique that can be optimized and used to solve the selection problems. This paper compares between the GA’s Elitism and the Roulette-Wheel selection methods when performing a trade study analysis for physical components-selection based on the Systems Modeling Language (SysML) logical architecture model of a hybrid car consisting of an engine, an electric motor, and a battery; and selects the optimum method for automating the MBSE-based trade study analysis. The results indicate that the Elitism selector has a better comparative performance than the Roulette-Wheel selector.