Genetic Algoritms for the Synthesis of Innovative Engineering Solutions

An intelligent system for the synthesis of compound objects is described. This system is based on the approach using natural evolution mechanisms to combine elements and to select viable solutions. A complete object is generated from a set of elements only if their mutual internal requirements are satisfied. The knowledge of experts represented as rules and functions is applied to form descriptions of generated objects. Selection of the best solutions is based on the generalized measure of similarity to the task that should be satisfied by the synthesized objects. The system allows the generation of solutions that have no internal contradictions and satisfy external requirements.