Vol.:(0123456789) 1 3 Arabian Journal for Science and Engineering https://doi.org/10.1007/s13369-020-04348-2 RESEARCH ARTICLE-COMPUTER ENGINEERING AND COMPUTER SCIENCE Integrating Advanced Harmony Search with Fuzzy Logic for Solving Bufer Allocation Problems Mahmoud Z. Mistarihi 1  · Rasha A. Okour 2  · Ghazi M. Magableh 1  · Haythem Bany Salameh 1,3 Received: 3 September 2019 / Accepted: 14 January 2020 © King Fahd University of Petroleum & Minerals 2020 Abstract This paper introduces a new fuzzy advanced harmony search algorithm for solving single-objective bufer allocation prob- lems (BAPs). The proposed algorithm represents the frst attempt at solving BAPs using a fuzzy logic system, by tuning the advanced harmony search control parameters. The main steps of the proposed algorithm included parameter initialisa- tion, harmony memory initialisation and evaluation, improvisation, harmony memory update, AHS parameter update, and termination criterion check. The aim of this approach is to achieve a better convergence rate and avoid the stacking of local optima. The performance of the proposed algorithm was compared with other methods used in solving BAPs. The proposed approach has shown a higher capability in fnding optimal solutions compared to previous methods used for two benchmark problems. Improvement of up to 94.75% in the overall throughput is reported for the 3-stage problem, while for the 12-stage problem, a slight improvement (up to 7.58%) is also reported in the overall throughput. The results achieved indicate that the proposed algorithm is an efcient and promising tool in solving BAPs. Keywords Bufer allocation problem · Fuzzy logic system · Advanced harmony search · Pitch adjusting rate · Bandwidth distance 1 Introduction The bufer allocation problem (BAP) is concerned with the allocation of a certain amount of bufer sizes in a functional way that adds value to any work or job in terms of ef- ciency and productivity [1]. It is classifed as an NP-hard combinatorial optimisation problem that utilises the discrete formulation style in order to be solved [1–3]. The BAP is a stochastic, nonlinear, conventional operations research subject that aims at fnding the optimal bufer sizes to be allocated with a suitable arrangement, in order to optimise given objectives and factors while taking into consideration various constraints. The BAP mainly arises when there are signifcant changes in demand and throughput volumes in manufacturing fow line systems. This research is concerned with solving BAPs with discrete, constrained bufer sizes in serial manufacturing fow line systems. The BAP has attracted the attention of researchers for over 50 years and is a signifcant research feld; thus, dif- ferent search algorithms have been implemented in order to solve it [1]. The conventional solution concepts that were used to solve BAPs have been reported in the literature. They were classifed into four main categories: exact procedures, heuristic methods, metaheuristics, and simulation methods [4]. Several classes of exact algorithms and heuristics have been proposed to optimise BAP [1]. Exact algorithms are based on linear programming and dynamic programming [5–7]. These methods are limited to smaller problems because of the exponential growth of the number of solu- tion stages, and this requires an exponential amount of memory. The bufer elimination algorithm (BELA), standard exchange vector algorithm (SEVA), and modifed Hooke- Jeeves algorithm are common examples of heuristics. Due to the complexity of BAPs, which are found to increase exponentially, researchers developed several metaheuristics and hybrid metaheuristics with different search algorithms such as simulated annealing (SA) [8], * Mahmoud Z. Mistarihi mahmoud.m@yu.edu.jo 1 Yarmouk University, Irbid 21163, Jordan 2 Jordan University of Science and Technology, Irbid 22110, Jordan 3 Al Ain University, Al Ain, UAE