Improved similarity coefficient and clustering algorithm for cell formation in cellular manufacturing systems

Cell formation plays an important role in the design of cellular manufacturing systems. Among many methods utilized to solve the cell formation problem, the similarity coefficient method is the most widely used owing to its high flexibility and low computational requirement. However, generalized similarity coefficients ignore material flow, which is closely related to operation sequence and repeated operations. Moreover, most similarity coefficient-based clustering algorithms focus on the number of inter-cell movements but disregard distinction of the movement effort. To overcome these limitations, this study improves the generalized similarity coefficient method to form part families. In addition, a new clustering algorithm is presented to assign machines to cells with minimum intensity of material inter-cell movement, which depends on the frequency, production volume and difficulty level of inter-cell movement. Experimental results demonstrate that the proposed method has superior sensitivity and effectiveness for solving the cell formation problem.

单元构建（Cell Formation）在成组制造系统（Cellular Manufacturing Systems）的设计中具有重要作用。在众多用于求解单元构建问题的方法中，相似系数法凭借其高灵活性与低计算需求得到了最为广泛的应用。然而，广义相似系数法忽略了与作业顺序及重复作业紧密相关的物料流。此外，多数基于相似系数的聚类算法仅关注单元间移动的次数，却忽视了移动作业负荷的差异。为克服上述局限，本研究对广义相似系数法进行改进以构建零件族（Part Families）。同时，本研究提出一种全新的聚类算法，可将机床分配至各单元，使单元间物料移动强度降至最低，该移动强度由单元间移动的频次、生产批量及移动难度水平共同决定。实验结果表明，所提方法在求解单元构建问题时具备更优异的敏感性与有效性。