International Journal of Scientific and Research Publications, Volume 3, Issue 2, February 2013 1 ISSN 2250-3153 www.ijsrp.org An Artificial neural network (ANN) based solution approach to FMS loading problem Suman Kant*, Rahul O Vaishya** *,** Assistant Professor, Department of Production Engineering, PEC University of Technology Chandigarh-160012 Abstract In this Paper, the FMS loading problem is solved with the bi-criterion objective to minimize the system unbalance and maximizing the throughput by the use of artificial neural networkin the presence of available machine time and tool slots as constraints. The complexity of machine loading Problem in FMS is very high due to the different flexibility criteria such as part selection, operation allocation and various constraints involved such as availability of tool slot and time available on machines. This encourages various researchers to apply various heuristic techniquesto get optimal/ near optimal solution. Artificial Neural Network (ANN), inspired by the structure and functional aspects of biological neural networks which is an adaptive system changes its structure based on internal and external information flow during learning Phase. In this current decade ANN has emerged as one of the important problem solving tool for complex engineering problems. Keeping this views present research adopted ANN to solve FMS loading problem. It gave optimal/near optimal results in less computational time. It has also been found simple and naive tool for solving loading problems of FMSs. I. INTRODUCTION flexible manufacturing system (FMS) consists of Numerical Controlled Machine tools, automated material handling system and other production assisted equipment. It is an integrated computer-controlled configuration with a supervisory computer named as host computer, which has a control over production in FMS. Pallets and fixtures are used to reduce the set-up times on machines to zero. The objective of an FMS lies in having strategy between large volume of mass production and low volume job-shop production. Therefore Medium volume and medium variety production are used in FMS. The main purpose of the FMS is to achieve efficiency of a well-balanced transfer line while retaining the flexibility of the job shop (Stecke, 1983). There are two types of decision problems associated with FMSs: Design problems (machine selection, lay out decisions of robots, selection of AGVs and paths etc.) and Operational problems (selection of part types, determination of production ratio, allocation of resources and loading). The machine loading problem is one of the major operational problem where decision is to be made regarding part types i.e. which part types are selected from a pool of part types and assigned to be which machines in order to meet certain specified objectives while satisfying the system constraints. These objectives are: (1) Balancing the machine processing time. (2) Minimizing the number of movements. (3) Balancing the workloads per machine for a system of groups of pooled machines of equal sizes. (4) Unbalancing the load per machine for a system of groups of pooled machines of unequal sizes. (5) Filling the tool magazines as densely as possible. (6) Maximizing the sum of operation priorities. It has been found through various literatures(Stecke and Solberg 1981, Shanker and Tzen (1985),Shanker and Srinivasulu 1989,Mukhopadhyay et al. (1991 ,) that multiple objectives have been considered while solving the FMS Problem. Numerous authors solved the loading problems with different solution methodology Mukhopadhyay et al. (1991) prioritized the loading of machine tools and parts in random FMSs through eigenvalue analysis. Mukhopadhyay and Tiwari (1995) solved the machine loading problem using the principle of conjoint measurement. Mukhopadhyay et al. (1991) prioritized the loading of machine tools and parts in random FMSs through eigenvalue analysis. The machine loading problem has been solved using different solution methodologies, which are given as follows: (1) Heuristic oriented (Stecke and Solberg 1981,Stecke 1983, Mukhopadhyay et al. 1992, Morino and Ding 1993, Chen and Chung 1996, Tiwari et al. 1997). (2) Simulation based (Jain et al. 1989, Sabuneuogloand Hommerzhein 1992, Basnet and Mize1993). (3) Multi-criterion decision making (Kumar et al. 1990, Chen and Askin 1990, Kim and Yano 1997,Sawik 1998). (4) Mathematical programming (Stecke 1983, Laskari et al. 1987, Shanker and Srinivasulu 1989, Sawik 1990, Liang and Dutta 1992, 1993,Guerrero et al. 1999). Shanker and Tzen (1985) solved the machine loading problem with their objectives include balancing workloads and meeting due date of part types. Ammons et al. (1985) solved with the bi- criterion objective of minimizing workstation and balancing workloads. Rajagopalan (1986) solved the machine loading problem with other problems inherently found in the planning stage, such as job selection and production ratio determination, with the aim of achieving better production schedules without too many iterations. Mukhopadhyay et al. (1992) and Tiwari et al. (1997) made an attempt to solve machine loading with an objective of minimizing system unbalance by maximizing throughput using heuristic approaches. Although numerous researchers have solved this combinatorial problem of FMS with different perception of objective/objectives, but it has been observed that majority of the authors applied more than one heuristic to solve the same. Computer programming was also required by several authors to solve the same problem. In the present manufacturing scenario A