Use of Associative Memory and Self-Organization in Conceptual Design

Experienced human designers store known design solutions. Whenever a new design problem is encountered, the designer uses the known elemental functional requirements and associates them with the known design solutions and retrieves the most closely matching solution. This design solution can be either directly used or mutated to generate a new design solution. In this paper the authors propose a model based on human associative memory as a means for capturing the conceptual design process. The associative memory is modeled as an Artificial Neural Network. The development and implementations of the model are discussed with the help of relevant examples. The two layer perceptron is trained using the back propagation algorithm.