Ideas-Generator for Conceptual Design using Evolution Strategies

Conceptual Design is one of the most important phases during the product development process, as the main characteristics, properties and costs of the solution and consequently also the main portions of success of the new product are pre-determined there. During this design phase the assistance of a computer aided software tool to the design engineer is desirable for the task of finding principal solutions. In our contribution the application of evolution strategies to typical conceptual design problems is examined using standard algorithms and tools. As a result of this approach, two important advantages arise. First of all, alternative effect chains may be established and secondly, it is possible to evaluate the concepts found, and to select the best of them with respect to a suitable choice of weighting factors for the relevant criteria. Only those solutions are generated and selected, which correspond best to the specified criteria of the multi-requirements problem. The presented approach was realized in a first software prototype using Matlab. It is shown by means of some significant examples, that evolutionary algorithms are useful for the generation and pre-evaluation of principal solutions during the concept phase of product development processes. The possibilities of generating different design concepts via evolution strategies will be demonstrated by a case study.