Theses supervised by Prof. Dr. Adil Baykasoğlu
15 theses · Gaziantep University, Dokuz Eylül University
Modeling and solving truck-load consolidation problems by using multi-agent technology
Logistics organizations mainly the ones providing land transportation services are facing with difficulties while making effective operational decisions. This is especially the case in making load/capacity/route planning and load consolidation decisions where customer orders are generally unpredictable and subject to sudden changes. Classical modeling and decision support systems are mostly insufficient to provide satisfactory solutions in a reasonable time for solving such dynamic problems. Agent-based approaches especially multi-agent paradigms which can be considered as relatively new members of system science and software engineering are providing effective mechanisms for modeling dynamic systems. These systems are generally operating under unpredictable environments and having high degree of complex interactions. It seems that multi-agent paradigms have a big potential to handle complex problems in land transportation logistics. Based on this motivation, in this thesis a multi-agent based load consolidation decision making approach is proposed. In the proposed approach the load consolidation decisions for the less-than-truckload orders which are dynamically dispatched to the system are made by the software agents. The less-than-truckload orders are assigned/consolidated to the trucks by the negotiation mechanism constructed within the model. The proposed approach considers the truck travel distances, number of order rejections, average costs of the transportation operations and system profitability so as to measure the performance of the proposed system. The proposed system is tested with some static problem sets and via running it under some stochastic environment.
Heuristic optimization through negotiation
Finding realistic and express solutions to several problems has been a fundamental requirement in this rapidly changing conditions and environment. Conventional approaches, which are solving dynamic problems, ?as they are static? or reinventing their models after each corresponding change, have all expired. In this respect, it has been understood that, the most effective solution strategies have been the agent-based strategies. These strategies, which are mostly constructed upon a heuristic, are capable to provide parallel and distributed solutions and furthermore they let agents to take autonomous decisions that can add value to the objectives of the problem. This PhD thesis aims both 1) setting up a representation scheme for agent-based approaches and 2) providing solutions to dynamic variants of Travelling Salesman Problem (TSP), which has been solved by static approaches up to now.Classical TSP consists of ?n number of cities? where, each city is visited once using an optimal route. Similarly, Generalized Travelling Salesman Problem (GTSP) covers ?c number clusters? where, each cluster is visited exactly once by visiting one of cities of the clusters using an optimal route. However, in some specific cases, assumptions of classical TSP and GTSP may not be satisfactorily enough to reflect physical reality and dynamism of actual systems. During solution of those problems, some cities are added to city domain or some disappear from the system. Thereby, the assumption of keeping the number of cities constant fails. Consequently, ?a solution provided for a specific to state defined for a time? may lose its superiority immediately after these unexpected changes. In this respect, these dynamic types of TSP and GTSP may not be modeled with the conventional research methods since much time and memory spaces are required to setup novel models and solve them repeatedly. With those considerations in mind, this thesis covers different agent-based solution policies/strategies for providing promising solutions to different problems in domain of dynamic TSP. In this respect, two main solution strategies are proposed for the solution of the defined problem types in this PhD thesis. One of these is competition of agents without a heuristic and the other is competition of agents using Great Deluge Algorithm (GDA). All those strategies have proved themselves competed with other findings in literature.
A simulation study to evaluate the effect of flexibility on flow time
In today?s competitive environment, companies are seeking for newer capabilities to survive. Especially, the manufacturing companies struggle to increase their inner potential by establishing the better manufacturing policies. In the increasing variety of demands and the products, it has been essential to include flexibility in the manufacturing policies. However, flexibility itself, without analyzing the type and conditions of the manufacturing system, may have no meaning. The level and the type of flexibility require to be adjusted and to be monitored for having the expected benefits fully. This thesis specifically intended to find effects of flexibility in job shop manufacturing. For this purpose, a hypothetical system with six machines and ten types of jobs at most with the six different operation types has been considered. Twelve scenarios are developed under four different cases (with/without transportation, with/without additional processing time), five level of flexibility, two different machine selection rule, and three types of dispatching rules. Developed models are experimented by the use of SIMAN simulation language. Findings have been tested statistically. Results indicated that full flexibility is a preferable state for most of the cases. However, in some cases chain configurations perform better since it combines the benefits of pooling and specialization.
Evaluating fuel alternatives for electricity generation in Turkey through multiple criteria decision support methodologies
In today's world, with development to the growing electrical energy demand is considered as a important problem. This problem became significant, to capture the new perspective about to installing environmental and environment-friendly clean fuel technology of energy generation systems not only thinking economical criteria. Turkey electric energy demand trend in is an increase 7.5% on average at a rate of speed. According to estimates to meet the increasing electricity demand, the current 42,394 MW installed capacity by 2020 of at least the period should be doubled. However Kyoto agreement was signed in 2009, within the framework need to be limit of carbon emission rates. In this case, the generation of electrical energy that are not only depend economic factors, also environment, sustainability, such as direct and indirect interactions with each other bearing in mind the many factors which need to be evaluated. For this kind of multi-criteria decision problems Analytical Hierarchy (AHP) and Analytical Network Processes (ANP) are good approaches to analysis comes to the fore. In this thesis to evaluate of electricity generation ways models built with the help of AHP and ANP methods. Models and criteria, mined from resources as follows, "Energy and Natural Resources Ministry", "International Energy Agency", "Turkey Electricity Transmission Company? issues and surveys with experts and were obtained by interviews. According to the thesis, models built and results put in place for priorisation of alternatives.
A simulated based approach to develop new approaches for ?due date assignment? and ?job release? in multi-stage job shops
Workload control (WLC) concepts are advocated as one of the new productionplanning and control (PPC) methods. In its elaborated form, WLC includes threemajor decision levels: (i) job entry, (ii) job release, (iii) priority dispatching. In eachdecision levels, several decision points which have significant impact on theeffectiveness of the PPC are defined (i.e. acceptance/rejection, due date assingmentetc.). In order to improve the effectiveness of PPC, WLC systems should consider allthese decision points simultaneously. In adddition to these decision points, flexibilityof the shop can be included as the fifth decision point which allows the shop capacityto be adjusted as new orders enters the system and as they are released to the shopfloor.In this study, simulation models which enable the effect of each decision pointswithin this WLC concept to be explored are developed and tested. The resultsrevealed that simultaneous consideration of decision points improves theeffectiveness of PPC. Moreover, a composite dispatching rule which can be used inpriority dispatching level to improve the effectiveness of other levels is developed byusing a data mining tool.Keywords: production planning and control, workload control, simulation, datamining
Employing meta-heuristics and fuzzy ranking functions for direct solution of fuzzy mathematical programs
Primary objective of this study is to present how fuzzy mathematical programmingmodels can be solved by employing metaheuristic algorithms and ranking methodsfor fuzzy numbers without requiring a transformation into a crisp model. Up to datevarious solution approaches are proposed to solve different fuzzy mathematicalprogramming models. The main difficulty in fuzzy mathematical programming is tosolve fuzzy models using the existing solution algorithms. In the existingapproaches to overcome this problem the crisp equivalent of the fuzzy models areobtained. In this study, a direct solution method is proposed to solve fuzzymathematical programming problems with different fuzzy parameters and fuzzymathematical programming problems with fuzzy decision variables. In the proposeddirect solution method, ranking methods for fuzzy numbers and metaheuristicalgorithms are used. The effectiveness of the proposed direct solution method isproved with different examples.Keywords: fuzzy mathematical programming, fuzzy decision variables, rankingmethods for fuzzy numbers, classification of fuzzy mathematical programmingmodels
Investigation of fuzzy functions approach and its possible applications in industrial engineering problems
Fuzzy set theory was introduced by Zadeh in 1965 as an extension to classical set theory. It has been a very important research subject for many researchers and has led to new developments for many fields since it enables to handle uncertainties successfully. One of these important developments is the fuzzy functions concept which was introduced by Professor I. Burhan Türkşen and combines fuzzy sets and fuzzy clustering concepts to provide an alternative solution approach to solve problems in diverse domains. The novelty of fuzzy functions is based on the fuzzy clustering concept and therefore based on fuzzy membership values. Fuzzy clustering is one of the corner stone of the fuzzy functions since finding the best partition constitutes the main problem in this approach. There are several fuzzy clustering algorithms in the literature which can be used in generating fuzzy functions. In this thesis Fuzzy c-Means (FCM) clustering algorithm is used in order to find out the membership values.One of the main motivations behind the development of the fuzzy functions approach was to overcome some of the drawbacks of the fuzzy rule bases which are one of the most frequently used fuzzy inference methods with many successful applications.As a contribution to the existing studies about fuzzy functions, first time in the present thesis we proposed to use genetic programming (GP) along with fuzzy clustering as a new approach in generating fuzzy functions. We used many data sets from the literature in order to present the application and the performance of our approach. We also performed comparisons with the existing fuzzy function generation methods like Least Square Estimation (LSE) in order to prove the validity of our approach. Based on the computational results we illustrated that fuzzy functions which are generated through genetic programming are very competitive and effective in many problem settings.Keywords: Fuzzy set theory, fuzzy rule bases (FRB), fuzzy clustering, fuzzy functions (FF), least square estimation (LSE), support vector machines (SVM), genetic programming (GP).
Modeling and solving mixed-model assembly line balancing problem with setups
This dissertation concerns the type-I mixed-model assembly line balancing problem with setup times (MMALBPS-I). MMALBPS-I is an extension of classical MMALBP-I in which sequence-dependent setup times between tasks are taken into consideration. The main goal of this dissertation is developing the mathematical formulation of the problem and solving the problem with newly proposed parallel hybrid meta-heuristic approaches.Within this context, a mixed-integer linear programming (MILP) model for the problem is developed and the capability of our MILP is tested through a set of computational experiments. Due to the complex nature of the problem, parallel hybrid algorithms are proposed in order to tackle the problem.First, a new hybrid algorithm (ACO-GA), which executes ant colony optimization in combination with genetic algorithm, is developed. The proposed ACO-GA algorithm aims at enhancing the performance of ant colony optimization by incorporating genetic algorithm as a local search strategy for MMALBPS-I. In the proposed hybrid algorithm ACO is conducted to provide diversification, while GA is conducted to provide intensification.Second, we tackled the problem with Bees Algorithm (BA), which is a relatively new member of swarm intelligence based meta-heuristics and tries to simulate the group behavior of real honey bees. However, the basic BA simulates the group behavior of real honey bees in a single colony; we aim at developing a new BA, which simulates the group behavior of honey bees in a single colony and between multiple colonies. The multiple colony type of BA is more realistic than the single colony type because of the multiple colony structure of the real honey bees.The performances of the proposed algorithms are tested through a set of computational experiments and computational results indicate that both algorithms have satisfactory performances.
An analysis of criteria interaction techniques in multiple attribute decision making and a fuzzy TOPSIS based new model proposal
The aim of this study is to shed light on the concept of criteria independence/dependence and propose a simple and effective method to deal with complex decision problems without the restrictive assumption of preferential independence. For this thought in line, we provide a comprehensive survey of the state of art methodologies used for modelling and solving MCDM problems with interdependent/interactive criteria. Initially, the independence assumptions of the classical decision theory are compiled from the multi attribute utility theory. Then focusing on the preferential independence concept, Analytical Network Process, Causal Maps, Decision-Making Trial and Evaluation Laboratory (DEMATEL), Fuzzy Cognitive Maps, Interpretive Structural Modelling, Bayesian Networks, System Dynamics, and Fuzzy Integral (specifically Choquet Integral) methods are explained and discussed. Literature survey of the pertinent methods is provided at the end of each chapter. Due to the excessive number of alternative methods capable of modelling criteria interactions, characteristics of a sufficient MCDM method are determined as a complexity management strategy. In our regard, Fuzzy Cognitive Maps has much to offer in the field of MCDM due to its capabilities in qualitative modelling and simulating the dynamic systems. We propose a novel hybrid MCDM method by combining Hierarchical Fuzzy TOPSIS and Fuzzy Cognitive Maps for modelling interdependent, complex, and uncertain problems. The proposed method is illustrated in the strategy selection problem to demonstrate practicality and applicability of the method.
Modeling and solving assembly line design problems by considering human factors with an application
Automobile cable network(harness) production area is divided into two groups as sub assembly and assembly line. Balancingprocess is generally consisted of sub assembly, assembly line, enrichment assembly andelectrical test.In this thesis, harness assebly line balancing problem is studied. While assembly line balancing is done at the same time pay attention that the assebly stations are ergonomically designed and the similiar tasks are grouped together. The goal is detected the acceptable ergonomic risk levels and made the balancing by considering ergonomic constraints. This study is a part of a San-Tez project that has a 1512.STZ.2012-2 project number which is supported by Ministry of Science, Industry and Technology of Turkey.
Karmaşık optimizasyon problemlerinin çözümü için metasezgisel algoritmaların paralel hesaplama yoluyla koalisyonu
Most of the real-life problems could be modeled as optimization problems and the need for effective solution of these problems is always in demand. In this context, efforts to develop effective approaches to solving optimization problems have been the subject of considerable research. These approaches usually combine general-purpose or particular rules in a logical scope and the methods that are formed by the logical combination of these rules are called optimization algorithms. In this study several metaheuristic algorithms are brought together to form a coalition under Weighted Superposition Attraction-Repulsion Algorithm (WSAR) in a parallel computing environment for solving complex optimization problems. The proposed approach runs different single solution based metaheuristic algorithms (SSBMAs) in parallel and employs WSAR (which a recently developed recently proposed swarm intelligence based optimizer) as controller. While SSBMAs are responsible for exploring the search space, WSAR controls the communication process between the SSBMAs. The presented method tested against some well-known complex optimization problems in three groups, namely, continuous optimization problems, binary optimization problems and combinatorial optimization problems. While CEC 2020 problems are selected as test problems for continuous optimization problems, the uncapacitated facility location problem (UFLP) and the set union knapsack problem (SUKP) are selected as test case for binary optimization problems. In addition, the Resource Constrained Project Scheduling Problem (RCPSP) and the Permutation Flow Shop Scheduling Problem (PFSP) are selected as test problems for combinatorial optimization. The obtained results are compared with some other optimization algorithms. The results of the comparison show that the proposed approach is competitive in terms of solution quality and solution time.
Bir öğrenme yönetim sisteminin aksiyomatik tasarım ile kullanılabilirlik değerlendirmesi
With the disruption of face-to-face educations after the Covid-19 pandemic worldwide, the usage of Learning Management Systems (LMS) has become widespread. Various software products usually named as LMS, that provide distance education, are used by universities and various educational institutions. However, choosing an appropriate LMS is a complex Multiple Criteria Decision Making (MCDM) problem and has also gained importance over time. Within the scope of this problem, although usability evaluation of the LMS is used, there are not many studies in the literature in which the MCDM method is used and real life case studies are conducted. In this direction, perceived usability evaluation of the SAKAI-LMS, that is in use at an academic department is performed as a real life case study. An integrated MCDM model is created from the usability criteria based on the analyses of the related literature. A questionnaire is developed and directed to three types of system users, to evaluate these criteria's importance and levels of fulfillments by the system. The criteria weights are calculated via analytical hierarchy process and detailed statistical analyzes of the data obtained from the questionnaire is made. For LMS perceived usability evaluation, Axiomatic Design Procedure (ADP) is considered as a suitable MCDM method as it allows an easy approach to data fusion and setting performance targets for decision makers. It is concluded that the proposed ADP based approach is easy to apply to practical circumstances and able to quantify perceived usability of the existing SAKAI system.
Farklı üretim ortamlarında dinamik parti büyüklüğü belirleme problemlerinin modellenmesi ve çözümü
This thesis addresses common lot sizing and scheduling problems that are encountered in the industries. Among these problems, General Lot Sizing and Scheduling Problem (GLSP) has received a lot of attention from researchers. The first aim of this Ph.D. study is to propose novel Simulated Annealing (SA) based hybrid approaches for solving the GLSP. First, a mixed-integer programming (MIP) model for the GLSP is given in order to solve smaller size problems. Afterwards, a matheuristic algorithm that integrates SA algorithm and MIP model is devised for solving larger size problems. The proposed matheuristic approach decomposes the GLSP into subproblems. The second aim of this Ph.D. study is to solve a practical production issue in the paper bag industry that contains the problem of capacitated lot sizing and scheduling with cutting plan assignment. A MIP model is proposed for modeling and solving the problem. To handle the industrial case, a problem-based matheuristic approach that resolves the lot sizing and assignment problems separately is provided based on this mathematical formulation. A similar approach is also proposed for the extended version which several real life applications such as existence backordered and defective items, external procurement, and sequence dependent setup times. The last aim of this Ph.D. study is to solve a practical production issue in the pool equipment manufacturer industry that contains the problem of the multi-machine capacitated lot sizing and scheduling problem with machine eligibility constraints. The considered problem is also formulated as a multi-objective MIP model. Computational experiments indicate that the constructed model meets the practical requirements for the pool equipment manufacturer industry.
Bulanık küme ve fonksiyonların seçilen karar mühendisliği problemlerine uygulanması
Decision engineering is an emerging multidisciplinary field in which systems analysis, uncertainty handling and decision theory have been utilized in combination. A decision engineer provides with a set of computer-assisted mathematical models to understand and model complex decision problems encountered in real-life. These problems are mostly characterized by imprecision, complexity and dynamism. Understanding of complex interrelationships among decision criteria by managing complexity of the problem is of critical importance in real-life decision engineering problems. These interrelationships are hard to be articulated by means of exact terms. Interval type-2 fuzzy sets are more appropriate to model words due to their flexibility in assigning membership grades. First, interval type-2 fuzzy DEMATEL method has been used to demonstrate how the causal dependencies among criteria be managed within a hierarchically structured problem by means of interval type-2 fuzzy sets. Then, fuzzy cognitive maps, which are another advanced tools for modeling causal dependencies, are extended to the type-1 and interval type-2 fuzzy sets by proposing new alpha-cut based methods. Another critical issue in decision engineering is about intelligent processing of past data for dynamic decision support. As ever-more data pour through the networks of organizations, considerable effort has been made in order to cultivate valuable information from these resources. In this thesis, two new models, based on fuzzy cognitive maps and interval type-2 fuzzy regression functions, have been developed to solve real-life dynamic multiple attribute decision making problems. Therefore, this thesis contributes to the decision engineering literature with four new decision models.
Dynamic manufacturing cell formation through market oriented programming
In today's competitive environment, cellular manufacturing is a promising approach providing both the flexibility of job shops and efficiency of flow lines. However, one of the drawbacks of cellular manufacturing and its algorithms is their inability to handle dynamic events, especially dynamic changes in part spectrum. Although there are various efforts in the literature, researchers still could not overcome this problem efficiently. Since handling dynamism with traditional methods is nearly impossible, and the reconfiguration of the cells according to each change is difficult and costly especially in volatile manufacturing systems. In this context, agent based modelling provides opportunities to model dynamism and to obtain efficient solutions. Since it has ability to track and evaluate the real time information if it is implemented successfully. On the other side, virtual cell formation concept provides the opportunity to create manufacturing cells without the reconfiguration. In this thesis study, it is mainly focussed on these modelling approaches to develop a dynamic cellular manufacturing system. And an integrated novel agent based virtual cellular manufacturing approach is developed. The proposed approach enables to realize part family formation, virtual cell formation, and scheduling simultaneously while considering dynamic part demand arrivals. The results are discussed and it is shown that the proposed approach is very effective.