بهینه سازی همکاری دسته ذرات با یک تکنیک داده کاوی برای طراحی ساخت و تولید سلول

In recent years, different metaheuristic methods have been used to solve clustering problems. This paper addresses the problem of manufacturing cell formation using a modified particle swarm optimization (PSO) algorithm. The main modification that this work made to the original PSO algorithm consists in not using the vector of velocities that the standard PSO algorithm does. The proposed algorithm uses the concept of proportional likelihood with modifications, a technique that is used in data mining applications. Some simulation results are presented and compared with results from literature. The criterion used to group the machines into cells is based on the minimization of intercell movements. The computational results show that the PSO algorithm is able to find the optimal solutions in almost all instances, and its use in machine grouping problems is feasible.

Cellular manufacturing is an organizational approach based on group technology (GT). Cellular manufacturing aims to divide the plant into a certain number of cells. Each cell contains machines that process similar types or families of products. Manufacturing cells (MC) provide considerable cost and productivity benefits to practical manufacturing environments. Other considerable benefits to be gained by grouping machines into cells can be found in literature (Selim, Askin, & Vakharia, 1998). The major issue in the design of manufacturing cells is the identification of machine and component groups. This identification process requires an effective approach to form part families so that similarity within a part family can be maximized. According to Selim et al. (1998), clustering analysis is the most frequently used method for MC design. However, because the cellular formation problem (CFP) is a NP-complete problem, there is still the challenge of creating an efficient clustering method. This paper deals with the use of a discrete particle swarm optimization algorithm in clustering problems for cell formations. The remainder of the paper is organized as follows. In Section 2, we discuss related works concerning clustering, machine grouping and particle swarm optimization techniques. In Section 3, the statement of the problem is presented. The proposed algorithm is presented in Section 4. Experimental results are presented in Section 5. Finally, we give some conclusions and suggest some lines for future research.

In this paper, a new approach based on particle swarm optimization algorithm for clustering problems has been proposed. A novel discrete PSO algorithm is proposed and applied to the computation of the velocity vector as in the traditional PSO algorithm. The proposed method works with a new notion of the velocity component. The algorithm is based on an approach called the proportional likelihoods, which is used in data mining problems. The algorithm and its theoretical concepts are explained and illustrated for a cell-formation problem with five instances of initial incidence matrix, 12 machines and 12 components. A set of experiments was performed to show that the algorithm is stable and presents low variability. The results obtained in this research are auspicious. In subsequent steps, the proposed algorithm will be applied to a variety of different test problems. In addition, parameter optimization and hybrid approaches are also topics for future research.