Solving the problem of balancing single model assembly lines using Meta heuristic Algorithms

Abstract

Assembly lines have been widely used in various production systems. An assembly line consists of a series of work stations arranged along with a material handling system. The components are assembled with a certain sequence depending on the precedence diagram for given cycle time. The decision problem regarding balancing the assembly process optimally is known as assembly line balancing problem (ALBP). Assembly line configuration is always critical to carry out a cost-efficient production system. The configuration planning, throughout all time, consists of all elements and decisions that are related to equipment and adjusting production units for a certain production system. Simple Assembly Line Balancing Problem (SALBP) considers only one-sided single model with deterministic time. The complexity of SALBP can be reduced by some assumptions like mass-production of only one homogenous product, no restriction besides the precedence relations, equipping stations in equal manner…etc. Seeing that, ALBP is NP-hard problem, exact methods will need exhaustive enumerations to find the optimal solution especially with multi-objectives. Hence, leading the researchers to use Meta-heuristics methods like Genetic algorithms (GA) and simulated annealing (SA). The aim of the research is to solve SALBP using Meta-Heuristics e.g. GA and SA, with two objectives: minimizing the number of work stations as a first objective and minimizing smoothing index as a secondary objective, and to compare the obtained results with those obtained by exact, heuristic and Meta-heuristic methods from the literature. The Enhanced Genetic Algorithm (EGA) is based on generating population by non-traditional random method. As well, solutions generated from heuristic methods were added to the population to increase the diversity of the generated population. On the other hand, the initial solution of the Enhanced Simulated Annealing (ESA) is obtained using Ranked Positional Weight (RPW) method. The two algorithms were tested using well-known benchmark problems available in the literature. The two algorithms showed the capability of solving optimally more than 75% of 171 benchmark problems, varying in size between 11 work elements and 111 work elements. What is more, EGA and ESA outperformed Simple Assembly Line Optimization Method of type-1 (SALOME-1) for most of the benchmark problems, from Smoothing Index and efficiency perspectives.