SAPFOCS: a metaheuristic based approach to part family formation problems in group technology

TL;DR

This paper introduces SAPFOCS, a novel metaheuristic based on simulated annealing combined with Part Coding Analysis, to improve part family formation in Group Technology, outperforming traditional clustering methods in solution quality.

Contribution

The paper presents a new metaheuristic approach, SAPFOCS, integrating PCA with simulated annealing for more effective part family formation in Group Technology.

Findings

SAPFOCS outperforms C-Linkage clustering in most tested cases.

The method effectively identifies high-quality part families.

Experimental results demonstrate the approach's efficiency and robustness.

Abstract

This article deals with Part family formation problem which is believed to be moderately complicated to be solved in polynomial time in the vicinity of Group Technology (GT). In the past literature researchers investigated that the part family formation techniques are principally based on production flow analysis (PFA) which usually considers operational requirements, sequences and time. Part Coding Analysis (PCA) is merely considered in GT which is believed to be the proficient method to identify the part families. PCA classifies parts by allotting them to different families based on their resemblances in: (1) design characteristics such as shape and size, and/or (2) manufacturing characteristics (machining requirements). A novel approach based on simulated annealing namely SAPFOCS is adopted in this study to develop effective part families exploiting the PCA technique. Thereafter Taguchi's orthogonal design method is employed to solve the critical issues on the subject of parameters selection for the proposed metaheuristic algorithm. The adopted technique is therefore tested on 5 different datasets of size 5 {\times} 9 to 27 {\times} 9 and the obtained results are compared with C-Linkage clustering technique. The experimental results reported that the proposed metaheuristic algorithm is extremely effective in terms of the quality of the solution obtained and has outperformed C-Linkage algorithm in most instances.