Browsing by Author "Baykasoglu, Adil"

Now showing 1 - 3 of 3

Capability-Based Machine Layout With a Matheuristic-Based Approach
(Pergamon-elsevier Science Ltd, 2022) Baykasoglu, Adil; Subulan, Kemal; Hamzadayi, Alper
Capability-based machine layout (CB-ML) problem is firstly introduced in this paper. In the conventional ma-chine layout problem, part flow matrix is generated from parts' machine routes to minimize total part flows. However, defining part flow matrix based on the machines' routes (instead of processing capability requirements of parts) restricts facility designers to utilize available flexibility in manufacturing systems. In this research, parts' processing requirements are defined in terms of Resource Elements (REs), which describe unique pro-cessing capabilities and the processing capability overlaps of machines. If part flow matrix is defined in terms of REs, it becomes possible to utilize available flexibility in a more effective manner. However, physical part flows cannot be identified directly from the RE-based flow matrices. Because, the processing requirements of manu-factured parts can be satisfied from alternative machines that contain the required REs. Therefore, RE-based part flow matrix must be mapped into the machine flow matrix, which requires defining the machine flow matrix as a decision variable. This makes the proposed CB-ML problem much more complicated than the conventional machine layout problem. We firstly developed an integer non-linear programming model for the proposed CB-ML problem. Because of its NP-completeness and nonlinear structure, a matheuristic-based solution approach is also developed. The extensive computational analysis have shown that the proposed approach is able to provide good quality solutions for the larger problem instances within reasonable computation times.
Greedy Randomized Adaptive Search for Dynamic Flexible Job-Shop Scheduling
(Elsevier Sci Ltd, 2020) Baykasoglu, Adil; Madenoglu, Fatma S.; Hamzadayi, Alper
Dynamic flexible job shop scheduling problem is studied under the events such as new order arrivals, changes in due dates, machine breakdowns, order cancellations, and appearance of urgent orders. This paper presents a constructive algorithm which can solve FJSP and DFJSP with machine capacity constraints and sequence-dependent setup times, and employs greedy randomized adaptive search procedure (GRASP). Besides, Order Review Release (ORR) mechanism and order acceptance/rejection decisions are also incorporated into the proposed method in order to adjust capacity execution considering customer due date requirements. The lexicographic method is utilized to assess the objectives: schedule instability, makespan, mean tardiness and mean flow time. A group of experiments is also carried out in order to verify the suitability of the GRASP in solving the flexible job shop scheduling problem. Benchmark problems are formed for different problem scales with dynamic events. The event-driven rescheduling strategy is also compared with periodical rescheduling strategy. Results of the extensive computational experiment presents that proposed approach is very effective and can provide reasonable schedules under event-driven and periodic scheduling scenarios.
Single Seekers Society (Sss): Bringing Together Heuristic Optimization Algorithms for Solving Complex Problems
(Elsevier, 2019) Baykasoglu, Adil; Hamzadayi, Alper; Akpinar, Sener
This paper introduces a new metaheuristic, Single Seekers Society (SSS) algorithm, for solving unconstrained and constrained continuous optimization problems. The proposed algorithm aims to simulate the behaviour of a group of people living together, both individually and holistically. The SSS algorithm brings together several single-solution based search algorithms, single seekers, while realizing an information sharing mechanism based on the superposition principle and the reproduction procedure. Each single seeker tries to improve one single solution by using randomly generated parameter set until the stopping condition is reached. Then, the SSS algorithm exchanges partial information between the best solutions identified by the single seekers via the reproduction process. This characteristic generates new solutions to set as the starting point of the single seekers for their next run and provides a satisfactory level of diversification for the SSS algorithm. Additionally, the SSS algorithm determines a target point via the superposition principle at each iteration to make the single seekers to direct their discovery towards this target point. Thus, the SSS algorithm has the feature providing to share the information produced by the single seekers through the reproduction and the superposition principle. The performance of the proposed SSS algorithm is tested on the well-known unconstrained and constrained continuous optimization problems through a set of computational studies. This paper compares SSS algorithm against 27 and 17 different search algorithms on unconstrained and constrained problems, respectively. The experimental results indicate the stability and the effectiveness of the SSS algorithm in terms of quality of produced results, achieved level of convergence and the capability of coping with trapping in local optima and stagnation problems. (C) 2018 Elsevier B.V. All rights reserved.