GABSys: using genetic algorithms to breed a combustion engine

Most engineering systems are designed on the basis of a steady, or static, state. However, dynamic effects must also be considered. This paper presents GABSys (Genetic Algorithm-Bond graph System), a tool for approximating design optimality for multi-dimensional dynamic systems. Although GABSys does require the assistance of an engineer, it can be used to "automatically" generate solutions to design problems. We illustrate GABSys's utility by using it to design a two-stroke combustion engine. GABSys begins with behavioral requirements expressed in a bond graph and uses genetic algorithm techniques to produce a design that meets the specified behavioral requirements.