A NEW GENETIC ALGORITHM FOR THE MACHINE GROUPING PROBLEM WITH MULTIPLE OBJECTIVES

This paper presents a new genetic algorithm to address the cell formation problem in Group Technology. Instead of the conventional binary part-machine incidence matrix, information about the sequence of operations corresponding to each part type has been used. A multicriteria formulation with three objective functions, namely minimization of the total cost of intracellular and intercellular movements, minimization of cell load imbalance and the minimization of within-cell load imbalance is presented. Some new genetic operators (crossover-with-cell-reassignment operator, rolling mutation operator, cell split operator, cell fusion operator and reparation operator) are proposed. A new way to select the population based on the concept of ideal goal has been developed. An efficient set composed of all non-dominated (i. e. Pareto optimal) solutions are collected and keep updated during the exploration. Experiments illustrate the behaviour of the proposed algorithm.