Machine Cells Assignment In Cellular Manufacturing Systems Using Pso Algorthithm

Abstract

Cellular manufacturing (CM) is a crucial method to adapt lean manufacturing with an enormous archive of research and studies. One pivotal task of CM is the cell formation problem (CFP) that can determine how successful the formation of cells is. This study's goal is to effectively implement particle swarm optimization (PSO) as a linear integer program to solve the cell formation problem (CFP). The proposed model adopts the expression of particle velocities as proportional likelihood, uses Sigmoid function as a probability of change, and introduces a weight matrix for a repair heuristic that can effectively boost the overall performance. Different well-known benchmarks are used as a reference to compare the grouping efficacy of the proposed model with Intelligent Particle Swarm Optimization (IPSO) - a 2-phase network to evaluate the proposed model's effectiveness. Numerical results have shown that the proposed model can perform well in most cases when the best grouping efficacy values are equal or better than that of IPSO. Furthermore, the results also show significant improvements of both the convergence speed and the consistent of PSO algorithm by implementing weight matrix into the instance of assigning unassigned-parts to formed cells. Although there are still visible weaknesses in the current work, this study can be a ground work for improvements and innovations for future research.