Component‐Selection is an important task in design synthesis of MBSE. A trade study is commonly used to help systems engineers and stakeholders selecting the components of a systems design. A simple analysis may be sufficient when it involves only two parameters. However, when the components and their integration become more complex, the trade study also becomes harder, time‐ and cost‐consuming, and error‐prone. This paper aims to propose a method to automatically generate the solution by performing an evolutionary search. Sample components of a hybrid car which consists of an engine, an electric motor, and a battery are used in our initial prototype. The logical architecture is represented in the OMG SysMLTM via CSMTM. Through the experimental result, this paper shows that the proposed technique allowed the system design to be efficiently selected.

No abstract

Genetic algorithms can be used to perform an automated trade study analysis for component selection with the OMG Systems Modeling Language or OMG SysML™. However, the genetic algorithm encoding based on a single string with real‐value encoding failed to include the relationships in the OMG SysML™ model for the solution space. The genetic algorithm was then unable to perform an automated reasoning through the OMG SysML™ as a result. To solve this problem, this study proposes tree encoding as the encoding technique for the OMG SysML™ elements. Tree encoding offers a simple way to include blocks and their relationships into a genotype without altering its semantics. The scope of this study is limited to a preliminary investigation of the block definition diagram. The result from this study suggests that the conceptual idea can be applied to the genetic algorithm and the potential use of the approach in the trade study analysis, reasoning, and machine learning.

No abstract

No abstract