A Novel Algorithm for Solving the Multi-Objective Assembly Line Balancing Problem

The assembly line balancing (ALB) problems address the issue of assigning work elements and workers into different workstations to optimize a predetermined set of objective parameters such as the number of workstations, the cycle time and the production efficiency. Most of these combinatorial optimization problems possess NP-hard computational complexity. This rationalizes the growing interest of researchers to use metaheuristics to solve ALB problems. The problems studied in literature over the years are often limited by simplifying assumptions, which brings in a challenge to the models developed in academia while applying to real life problems. In this thesis, we have intensively reviewed the characteristics of different types of ALB problems and existing solution approaches. Based on that, a general structure of the ALB model is developed which can be extended to any ALB problems with a reasonable effort. In traditional evolutionary optimization algorithms, encoding and decoding of solutions appear as an integral part. We eliminated the computational efforts for encoding and decoding of the solutions, by introducing a unique representation technique using an extended graph notion. This new method also helps to implement real life considerations into the algorithm in more effective manner. A novel workstation oriented line balancing algorithm is proposed to solve the multi-objective ALB model which complies with the new representation technique. This algorithm can start solving the problems at an arbitrary location compared to the conventional method of sequential forward and backward assignment. The algorithm is allowed to search in the infeasible regions while trying to minimize the degree of violation. These characteristics make the algorithm less prone to getting trapped. In addition to that, the algorithm is capable of finding potential good solutions where the user might agree to allow minor trade-off on feasibility criteria. To deal with the multi-objective aspects of the problem, Pareto fronts are generated using a robust sorting technique based on NSGA II. An interactive user interface is developed where the user is allowed to update the input parameters at any time, making it suitable for reconfigurable facilities. Finally, the effectiveness of the proposed method is verified using a series of well-known test problems from literature. The proposed novel algorithm has demonstrated promising results in all test problems. The proposed algorithm is designed to solve a representative multi-objective model from versatile types of ALB problems. Some extension of the same methodology can be used to solve many other generalized assembly line balancing problems with real life features such as – stochastic processing time, assignment restrictions, processing alternatives, U-shaped lines, mixed model lines, machine breakdown, and the like.

Name thesis:degree_name

Master of Applied Science (MASc)

Level thesis:degree_level

Master's

Discipline thesis:degree_discipline

Engineering - Industrial Systems

Grantor dc:publisher

Faculty of Graduate Studies and Research, University of Regina

Year dc:date.issued

2016

Author dc:creator

• Hossain, Sayed Kaes Maruf

Advisor dc:contributor.advisor

• Ismail, Mohamed

Committee members dc:contributor.committeemember

• Henni, Amr
• Shirif, Ezeddin

Language dc:language.iso

en

DOI

https://doi.org/10.82465/4222

OAI identifier oai:identifier

oai:uregina.scholaris.ca:10294/7691