Hybridization of modified sine cosine algorithm with tabu search for solving quadratic assignment problem

• Main Author: Nurdiyana, Jamil
• Format: Thesis
• Language: English
• Published: 2022
• Subjects: Q Science (General)
• Online Access: https://etd.uum.edu.my/10448/1/s825969_01.pdf

Sine Cosine Algorithm (SCA) is a population-based metaheuristic method that widely used to solve various optimization problem due to its ability in stabilizing between exploration and exploitation. However, SCA is rarely used to solve discrete optimization problem such as Quadratic Assignment Problem (QAP) due to the nature of its solution which produce continuous values and makes it challenging in solving discrete optimization problem. The SCA is also found to be trapped in local optima since its lacking in memorizing the moves. Besides, local search strategy is required in attaining superior results and it is usually designed based on the problem under study. Hence, this study aims to develop a hybrid modified SCA with Tabu Search (MSCA-TS) model to solve QAP. In QAP, a set of facilities is assigned to a set of locations to form a one-to-one assignment with minimum assignment cost. Firstly, the modified SCA (MSCA) model with cost-based local search strategy is developed. Then, the MSCA is hybridized with TS to prohibit revisiting the previous solutions. Finally, both designated models (MSCA and MSCA-TS) were tested on 60 QAP instances from QAPLIB. A sensitivity analysis is also performed to identify suitable parameter settings for both models. Comparison of results shows that MSCA-TS performs better than MSCA. The percentage of error and standard deviation for MSCA-TS are lower than the MSCA which are 2.4574 and 0.2968 respectively. The computational results also shows that the MSCA-TS is an effective and superior method in solving QAP when compared to the best-known solutions presented in the literature. The developed models may assist decision makers in searching the most suitable assignment for facilities and locations while minimizing cost.