An improved NSGAII-SA algorithm for the cell manufacturing system layout optimization problem

To address the challenges of significant logistics crossings and low production efficiency in traditional cluster layouts, a cellular manufacturing system (CMS) is commonly employed in diverse, small-batch production processes due to its high flexibility and adaptability. This study presents a comprehensive approach to effectively transform cluster layouts into cell manufacturing layouts, addressing the associated challenges. Initially, an improved fuzzy C-means clustering algorithm, enhanced with the elbow and the dissimilarity coefficient methods, is applied for cell division. Subsequently, a bi-objective optimization model is developed to minimize both the logistics distance and the layout area, with the NSGA-II-SA algorithm specifically tailored to handle the bi-objective sampling criterion. Thereafter, the layout optimization is performed, focusing on both the order and direction of the intracellular facilities. By applying the elbow method to the part-equipment matrix across various dimensions, its effectiveness in determining the optimal number of cell partitions is validated. Finally, the whole process of transforming the cluster layout into a CMS is successfully executed. The results demonstrate that the proposed algorithm outperforms non-dominated sorting genetic algorithm II (NSGA-II), the simulated annealing (SA) algorithm using random sampling (RM_SA), and the SA algorithm using bi-objective sampling (TM_SA) algorithms in both searchability and overall performance.