Application of noise clustering in group technology

Cell formation is the most important problem faced in designing cellular manufacturing systems based on the principle of Group Technology. In that, parts with similar geometry, function, material and/or requiring a similar production process are grouped into part families and corresponding machines are organized as independent cells. One of the main weaknesses of the conventional grouping methods is that they implicitly assume that the components belong to one of the part families. In reality, some parts often require processing by machines from multiple cells and thereby belong to more than one-part families and appear as bottleneck parts. It is necessary to identify these bottleneck parts while grouping, and subsequently, they may be processed by alternative methods, say subcontracting. The identification of bottleneck parts may be considered equivalent to the isolation of noise and outliers in robust fuzzy classification task. In this paper, Dave's noise resistant fuzzy clustering model is applied to solve this problem.