A new linguistic based algorithm for creating manufacturing cell

This paper presents a new approach for modeling formation of machine groups and the associated parts families using a linguistic model. The method outlines a hierarchical approach for deciding the number of clusters/machine groups. The solution is improved further by utilizing a genetic algorithm with the objective of decreasing actual job-shop inter-cell distance moved. A unique feature of the proposed method is that it recognizes the operation sequence of manufacturing the parts specified in the process sheets. The concept of null machine is introduced in the linguistic model to calculate the dissimilarity among parts. The movement of parts can be either flow type or random job-shop type for the discrete manufacturing shop, required to be partitioned. Back flow/tracking are minimized. The flexibility of changing number of machine groups is also built-in the model. The model has been found useful and flexible enough to solve any realistic discrete manufacturing system and is found to work comfortably in medium to large job shop situations when the machine part incidence matrix shows sparse density.