An enhanced NSGA-II algorithm for fuzzy bi-objective assembly line balancing problems

• Published in: Computers & industrial engineering Vol. 123; pp. 189 - 208
• Main Authors: Babazadeh, Hossein, Alavidoost, M.H., Fazel Zarandi, M.H., Sayyari, S.T.
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.09.2018
• Subjects: Assembly line balancing; Fuzzy programming; Multi-objective genetic algorithm; NSGA-II; Multi-objective genetic algorithm; NSGA-II; Fuzzy programming; Assembly line balancing
• ISSN: 0360-8352; 1879-0550
• DOI: 10.1016/j.cie.2018.06.014

•We developed a new fuzzy mixed-integer linear programming model for SALBP and SULBP.•We proposed an enhanced NSGA-II algorithm to solve such problems.•The results demonstrate our proposed approach is capable to handle any MOLP models.•The solution approach validity is evaluated along with several benchmarks. Due to immense role of an efficient assembly line in today’s manufacturing systems, the consideration of this paper is devoted to address the straight and U-shaped assembly line balancing problems. Assembly Line Balancing Problem (ALBP) considers conflicting objective functions that should be optimized simultaneously subject to set of constraints. To this end, this paper spends endeavor to develop a new fuzzy linear programming model. Having dealt with uncertain nature of the real production system, triangular fuzzy numbers (TFNs) are employed in order to represent uncertainty and vagueness associated with the task processing times. The proposed model can be regarded as a basis of fuzzy programming for further practical development in assembly line problems. To solve the problem, an efficient Multi Objective Genetic Algorithm (MOGA) is proposed. For this purpose, having respected several special characteristics of both fuzzy straight and U-shaped assembly line balancing problems for algorithm segments, including initial generation, encoding and decoding schemes, and GA’s operations, an new approach is proposed to promote population diversity and search efficiency. Finally, the proposed algorithm is evaluated though several benchmarks as well as that of exact method. The obtained results proved high efficiency of our method over other approaches suggested in the literature.