Application of a cascade BAM neural expert system to conceptual design for facility layout

The major contribution of this novel application is the pilot development and feasibility study for a bank of cascade BAM (Bidirectional Associative Memories) neural networks. This improved BAM structure functions as an expert system for conceptual facility layout or for preliminary construction layout design. This application, rather than being a better analytical algorithm or a better production expert system, builds a neural expert system with the capability of incrementally learning layout design examples for a given set of constraints. The cascade BAM incremental learning methodology, which distinguishes this system from the more frequently used Backpropagation Network (BPN) learning system, creates effective multibidirectional generalization behavior from qualitative, goal-driven layout design experience. The initial tests of learnability are presented by its applicability to conceptual layout design problems, and their solutions are assessed and compared with the learning ability of a standard BAM. Issues deserving further investigation are addressed as well.