www.iiec2013.ir A Novel Multi-Objective Genetic Algorithm for Cell Formation Problems R. Tavakkoli-Moghaddam , R. Jafari-Marandi Faculty of Industrial Engineering College of Engineering, University of Tehran Tehran, Iran tavakoli@ut.ac.ir ruholla.jafari@ut.ac.ir Abstract — contemplating a real cell formation problem (CFP), we can see there is not just a single objective to optimize; ideally, an eligible solution should be optimized in more than one objective simultaneously. The design of manufacturing cells with respect to multiple criteria has been attractive research for more than two decades. Due to contradictory and incommensurable objectives, most of the exact and heuristic algorithms fail to solve multi-objective cell formation problems. In this paper, we propose a novel methodology based on a genetic algorithm (GA) dealing with multi objective-based CFPs. We present a new and unique chromosome inspired by the essence of the CFP. This chromosome is defined somehow by two spirits (i.e., rows), and this fact empower us to propose unique and equally adaptable crossover and mutation operators. Furthermore, the proposed methodology of dealing with the multi-objective optimization problem (MOOP) shows a highly powerful tool in finding a Pareto optimal set. We engage a dummy evolutional objective (DEO) to better inspire the essence of the MOOP. Both main contribution of this paper, the unique chromosome for the CFP and the new methodology for solving the MOOP are adaptive enough to be applied to the gamut majority of problems related to the CFP and the MOOP. Cell formation problem; Multi-objective optimization; Genetic algorithm; Meta-heuristics I. INTRODUCTION A well-built manufacturing facility strengthens manufacturing efficiency and capability by diminishing material flow, materials handling cost, lead times and work in progress. Scheduling and the control of operations can be also more effective [1]. Group technology (GT) is an approach in a manufacturing area, which has proved having the potential to contribute conclusively in batch-type production, and its endeavor is to take advantage of both the flexibility of consider job production system and productivity of the flow production system [2]. The application of GT in a manufacturing system is cellular manufacturing (CM) that is known as a production approach with ability to decode intricacy of the problems. It is a competent of reducing the length of manufacturing lead times in batch production. The inviting problem in CM is the formation of product families and machine cells, namely cell formation (CF). The CF problem appears as the first point of a cellular manufacturing system (CMS). The most known and also principle objective of CF is dispatching part families to machine cells in order to optimize the performance measures, such as inter-cell and intra-cell grouping efficacy, transportation cost, exceptional elements, and the like. The cell formation problem (CFP) commenced with Burbidge [3] who initiated the very first method of the production flow analysis (PFA). For decades, many researchers have been reported their methods for the CFP, whose methods can be catalogued below (first initiated by Felix Offodile et al. [4]): • Array-based methods [5]. • Clustering methods [6-13]. • Mathematical programming-based methods [14-17]. • Graph theoretic methods [18-22]. • Neural network-based methods [23-29]. • Search methods(meta-heuristic algorithm): – Tabu search (TS) [30, 31]. – Simulated annealing (SA) [32-35]. – Genetic algorithm (GA) [2, 27, 36-49]. Contemplating real Cell formation optimization problem, we can see there is not just a single objective to optimize; however, eligible solution must be optimized in more than one objective simultaneously. The manufacturing cell design regarding multiple criteria has been an inviting research area since 1990. Wei and Gaither [50] initiated a four-objective cell formation to optimize the bottleneck cost, intra-cell load imbalances, and inter-cell load imbalances, and the average cell utilization. Venugopal and Narendran [27] developed a bi- criteria model to minimize the total sum of inter-cell and intra- cell material handling movements as well as the total cell load variation by means of the GA simultaneously. Gupta et al. [36] invented the optimization model, which concentrates on the weighted sum of both intra-cell and inter-cell movements. They solved the model using a genetic algorithm. There has been considerable attention on the minimum acceptable