FACT - A NEW NEURAL-NETWORK-BASED CLUSTERING-ALGORITHM FOR GROUP TECHNOLOGY

This paper introduces the FACT (Fuzzy art with Add Clustering Techniqu e) algorithm which is a new neural network-based clustering technique. FACT can be trained to cluster machines and parts for cellular manufa cturing under a multiple objective environment. The existing GT cluste ring techniques are mainly concerned with grouping parts and machines based on only one criterion which is the parts' processing routes. The FACT algorithm is able to consider several similarity criteria such a s parts' processing routes, design requirements of parts, processing t ime on each machine, and demand for each part. The FACT algorithm, whi ch is based on the fuzzy ART neural network, is powerful enough to sol ve problems of real-world sized complexity.