A METHODOLOGY. FOR SELECTION AND - SEQUENCING OF PART FAINILIES IN R.M.S

Abstract

Manufacturing companies in the 21st century are facing highly unpredictable and high frequency market changes driven by global competition. Customers are demanding highly customized and complex products. To survive in today's global competitive environment a cost-effective and responsive manufacturing system is required. In the background of inability of existing manufacturing systems (such as DMS, CMS, FMS etc.) to face these global business challenges, a new manufacturing system paradigm namely `Reconfigurable Manufacturing System' (RMS) has been envisaged. It is designed for rapid adjustment of production capacity and functionality, in response to new market conditions. The issue of recognition of appropriate part families is central to the RMS design as each part family requires a particular manufacturing system configuration to manufacture all its member parts. One part family is manufactured at a time by constituting suitably configured Reconfigurable Machine Tools (RMTs) catering to its operations. The production of next family starts only when the system is reconfigured after the completion of present family. This task of recognition of part families is accomplished by applying the philosophy of group technology that takes advantage of the similarities between design and/or manufacturing attributes of the given set of parts. The similarity between parts is calculated using some appropriately defined similarity coefficient such as Jaccard's similarity coefficient. Then `Hierarchical Agglomerative Average Linkage Clustering Algorithm' is applied to similarity matrix in order to obtain a dendogram that shows the diverse sets of part families that may be formed. For, the final selection of a set of families, the cost of each level in the dendogram is evaluated, and the set of part families at the level with the lowest cost will be selected. In this work a methodology has been proposed for selection and sequencing of part families for meeting R.M.S objectives economically. The problem has been formulated- as a Traveling Salesman Problem (T.S.P) and it is proposed to solve it through Ant Colony Optimization algorithm "Ant Colony System" (A.C.S). The objective function minimizes the sum of reconfiguration cost and machine idle cost. The algorithms developed are coded in MATLAB software. iii