Endüstri Mühendisliği Yüksek Lisans Tezleri / Industrial Engineering Master Theses

Permanent URI for this collectionhttps://gcris3.etu.edu.tr/handle/20.500.11851/613

Browse

Now showing 1 - 20 of 95

Acil Tıbbi Yardım İstasyonlarının Yer Seçimi ve Planlaması
(TOBB Ekonomi ve Teknoloji Üniversitesi Fen Bilimleri Enstitüsü, 2014) Konak, Ecem; Aytaç, Berrin
This study deals with the problem of optimally locating and designing emergency medical service systems. The problem is formulated as a multi-source capacitated fixed charge facility location model. The model minimizes the total cost of meeting demand while guaranteeing a maximum service response time. In order to determine the locations and capacities of emergency medical service locations, defined in terms of the number of emergency service vehicles, service requests are first assumed deterministic. The uncertainty in service requests is then introduced using a scenario-based two-stage stochastic programming approach together with the concept of service levels that allows a maximum number of service requests being not responded. Both optimization models are coded in JAVA programming language and solved using the optimization solver CPLEX. For the deterministic model, a two-stage constructive heuristic is proposed; the performance of the proposed heuristic is evaluated in terms of the solution quality and solution time using a large set of problem instances. The impact of the changes in the problem parameters on the performance of the heuristic is also investigated.
Akıllı Adapte Olabilen ve Terör Amaçlı Saldırılar Düzenleyen Düşmana Karşı Eldeki Kaynakların Optimal Kullanımının Oyun Kuramı Ekseninde İncelenmesi
(TOBB Ekonomi ve Teknoloji Üniversitesi Fen Bilimleri Enstitüsü, 2012) Birihan, Hasan Tugay; Bakır, Niyazi Onur
Terrorism has recently reached the international extents, terrorist organizations have started to target more countries to achieve their political aims in global scale. Threatened countries are obliged to defend the targets within their territory for their economic interests and societal psychology. In this study, we examined the problem of defending these targets within the framework of economic measures. Defending these targets is so costly that for decreasing the cost of defending and the risk, analytical approach is required. Game-Theory is one the most effective approach used for the enemies adapting in accordance to their targets. Government?s aim is prevent terrorist attacks within their territory or minimizing the loss in the terrorist attack. In this government-terrorist game, we assume that attacker and the defender do not move simultaneously. The State primarily allocates security resources to targets that to be protected. On the other side, the terrorist observes the state and carries out terrorist attacks.In this study a model with two players, two potential targets in the government?s territory and analyzing the possibility of attack on this targets was established. Bayesian approach was used on the gain of the attacker and loss function of defender was defined. By using this model, it is aimed to investigate the worth of intelligence and its effect on decisions to be made.
Ankara'da Toplu Taşıma için Veriye Dayalı Analiz ve Planlama
(TOBB University of Economics and Technology,Graduate School of Engineering and Science, 2018) Bakar, Merve; Kuyzu, Gültekin
Smart card automated fare collection systems are used in many transportation systems throughout the world as they provide effective and reliable payment opportunity. Smart card automated fare collection systems facilitate efficient and accurate fare collection in public transport systems. These systems enable the planners to implement more flexible pricing structures compared to traditional fare collection methods. Smart card systems record several pieces of data about the passengers, which can be used to improve the overall efficiency and service quality of the public transport network. In this work, we focus on analyzing smart card transaction data to understand spatial and temporal travel patterns of public transport passengers in Ankara, Turkey. One of our primary goals is to identify origin-destination pairs where the passengers are required to transfer through one or more intermediate points because of the lack of a direct service. We use a data set of about 30 million records corresponding to a one month period. The data includes records from bus and light rail transportation modes. Each record includes the smart card number, the transport mode, the bus/rail line, the boarding location, the boarding date and time, and the fare class of the passenger; but lacks the alighting location of the passenger. We first create a model to estimate the alighting location of each passenger. Then, we estimate origin-destination flows and their breakdown by several dimensions such as fare class, transportation mode, day of week, time of day and the frequency of the lines and the stations used. The estimation of alighting location of the passenger and the travel analysis are performed using RStudio program. Smart card data set of passenger travels on one-month period are analyzed according to smart card type; adult, student and teacher tickets. The most preferred public transportation vehicle is the bus with the usage rate of 61% in travels, the least preferred public transportation vehicle is the cable line. In addition to the travels on weekdays and weekends, density of the transit travels are analyzed according to time of day and frequency of the lines and the stations used. Keywords: Transportation, Public transport, Smart card data, Data analytics, Spatio-temporal analysis.
Approximate results for non-linear Cramér-Lundberg type risk model
(TOBB Ekonomi ve Teknoloji Üniversitesi, 2021) Allyyev, Yusup; Hanalioğlu, Tahir
Bu çalışmada doğrusal olmayan bir Cramér-Lundberg risk modeli ele alınmış, araştırılmış ve iflas olasılıkları, ?(u), hesaplanmıştır. Literatürde klasik model olarak da bilinen bu modelin doğrusal gösterimi şu şekilde tanımlanır: U(t)=u+ct-S(t) (1) Denklem (1)'deki U(t) risk süreci, belirli bir t zamanında bir sigorta şirketinin sermaye miktarını ifade eder, sabit u şirketin başlangıç sermayesidir, c – prim oranı, S(t)= ?_(i=1)^(N(t))?X_i, [0,t] aralığında meydana gelen kazalar için yapılan ödemelerden dolayı sermaye çıkışını tanımlayan bir ödüllü-yenileme sürecidir, N(t) bir yenileme süreci olup [0,t] aralığındaki toplam kaza sayısını belirtmektedir, X_i'ler ise, i. hasar için ödeme miktarını gösteren bağımsız ve aynı dağılıma sahip rasgele değişkenlerdir. Denklem (1)'de görüldüğü gibi şirketin prim gelirini ifade eden ct terimi zamanın doğrusal bir fonksiyonudur. Ancak bu varsayım gerçekçi değildir, çünkü bir sigorta şirketinin prim geliri her zaman doğrusal olarak artamaz. Bu, özellikle sigorta poliçesi sahipleri ile doymuş pazarlar için geçerlidir. Bu nedenle, prim gelirinin, monoton olarak artmasına rağmen, büyüme hızı zamanla azalan bir fonksiyon olarak modellenmesi tavsiye edilir. Bu nedenle, bu çalışmada aşağıdaki gibi ifade edilen, daha gerçekçi özel bir doğrusal olmayan matematiksel model inşa edilmiş ve incelenmiştir: V(t)=u+c?_(i=1)^(N(t))??ln(1+W_i)?+c ln(1+(?t-T?_N(t) ))-S(t) (2) Denklem (2)'de, W_i'lar (i=1,2,3… ) kazalar arasındaki süreleri gösteren pozitif, bağımsız ve aynı dağılıma sahip rastgele değişkenler dizisidir; T_N(t) = ?_(i=1)^(N(t))?W_i ise, W_i, i=1,2,3,… rastgele değişkenlerinin dizisine karşılık gelen bir ödüllü-yenileme sürecidir ve Logaritmik Risk Süreci olarak adlandırılan V(t) ise herhangi bir t zamanda şirketin sermaye dengesini tanımlar. Bu çalışmanın temel amacı, denklem (2)'deki doğrusal olmayan risk modelinin iflas etme olasılığını, ?(u), hesaplamaktır. Model oluşturulurken stokastik süreçler, yenileme süreçleri, ödüllü-yenileme süreçleri ve bu süreçlerin olasılıksal özellikleri kullanılmıştır. İlk aşamada doğrusal olmayan modelimizin iflas etme olasılığı için Lundberg tipi üst sınır bulunmuştur. Bu olasılık sınırları hesaplanmaya çalışılırken doğrusal olmayan denklemlerle karşılaşıldığında sayısal çözüm yöntemleri kullanılmıştır. Çeşitli senaryoları dikkate almak için farklı olasılık dağılımları ve parametreleri göz önünde bulundurup, regresyon modeli ile yaklaşık bir çözüm bulunmuştur. İkinci aşamada, bu doğrusal olmayan model için yukarıdan ve aşağıdan iflas olasılığının yaklaşık sınırları bulunmuştur. Bu aynı zamanda iflas olasılığı için Cramér tipi sınır olarak da bilinir. Bu amaçla, kazaları (hasarları) temsil eden {X_n } dizisi tarafından üretilen yenileme sürecinin kalan ömrünün limit dağılımını tanımlayan rastgele değişkeni X ?'ın istatistiksel özelliklerinden yararlanılmıştır. Özellikle, iflas olasılığının sınır ifadesinde bilinmeyen bir katsayı olan sabit bir C'yi belirlemek için X ?'ın moment çıkaran fonksiyonu kullanılmıştır. Bu ifadeleri sadeleştirmek ve kompakt bir forma dönüştürmek için kalkülüs yöntemleri kullanılmıştır. Benzer şekilde, iflas olasılıklarını incelemek ve hesaplamak için farklı senaryoları dikkate almak için çeşitli olasılık dağılımları ve parametreler kullanılmıştır.
Ara Stok Alanı Bulunan Tek Robotlu Üretim Hücrelerinde Çizelgeleme
(TOBB Ekonomi ve Teknoloji Üniversitesi Fen Bilimleri Enstitüsü, 2013) Gündoğdu, Emine; Gültekin, Hakan
This thesis considers the scheduling problem occurring in robotic cells consisting of a number of machines and a material handling robot which is responsible for loading/unloading of machines. Identical parts are produced with the same machine route, so flow shop assumption is valid. This is the first study considering a self-buffered robotic cell where the robot has a buffer space moving with it. The study is categorized into 3 main sections in terms of the buffer capacity. The first section, where the buffer capacity is 1, considers only 1-unit cycle in which one part is produced in one repetition. For this section, parameters for which each non-dominated cycle is optimal are determined. Moreover, the benefit obtained by using a self-buffered robot over a classical robot is found. As done in the first part, we determined non-dominated 1-unit cycle for double capacity buffer space and chose two of these cycles. Furthermore, to determine the performance of these cycles, a lower bound for the cycle time of optimal robot move cycle is developed. The final section of the study where the buffer capacity is assumed to be infinite, a new class of cycle called q-accumulated cycle is defined to provide more benefit from the buffer area. To find optimal q-accumulated cycle, optimal buffer capacity for given cell parameters and make comparison of a self-buffered robot and classical one, a computational study is conducted.
Aras Kargo Ankara Bölgesi Ana Transfer Merkezi Çizelgelemesi ve Araç Çıkış Zamanları Koordinasyonu
(TOBB Ekonomi ve Teknoloji Üniversitesi Fen Bilimleri Enstitüsü, 2010) Dikmen, Onur; Ertoğral, Kadir
The problems related to parcel delivery have started receiving attention in the literature in recent years. Determination of the vehicle routes for desired service levels, network design, consolidation decisions, operational planning of transfer centers are commonly studied problems. This study addresses the coordination of the vehicle departure times from branches and the scheduling of the inbound vehicles to unloading gates at a central cross docking terminal of Aras Kargo. The objective is to maximize the weighted sum of vehicle departure times. The main constraint is to finish up the unloading operation of all vehicles until a target completion time. Integer programming is used for modeling the problem. The mathematical model, which is developed for different confidence levels, is based on a previously proposed model for the parallel machine scheduling problem. Because of the complexity of the problem, a two-phase heuristic approach is proposed. The heuristic approach first solves the assignment problem of vehicles to unloading gates using the linear relaxation of the model, and then it addresses the scheduling problem for each door separately. The efficiency of the heuristic approach is shown by numerical experiments. Additionally, for the purpose of observing the effects of uncertainty in the system and including the dynamic structure of the queue formation, a simulation model is developed. Using the simulation model, the effects of delays on truck departure times, optimization of the truck departure times, the increase in demand and relevant door assignment alternatives are examined. The results of the mathematical and simulation model are compared with each other as well.
Ardışık Hidroelektrik Santrallerden Oluşan Bir Sistemde Üretim Planlama ve Gün Öncesi Enerji Piyasasında Teklif Oluşturmak için Entegre Bir Optimizasyon Modeli ve Analizi
(TOBB ETÜ, 2021) Özbolat, Sevde Nur; Ertoğral, Kadir
Tüm dünyada olduğu gibi ülkemizde de elektriğe olan talep gün geçtikçe artmaktadır, bunun bir sonucu olarak yenilenebilir enerjiye olan ilgi ve yatırımlar da artmıştır. Türkiye'nin önemli enerji kaynaklarından biri olan hidroelektrik enerji, diğer enerji kaynakları ile karşılaştırıldığında daha kolay başlatılıp durdurulabilir bir yapıya sahiptir. Esnek ve kontrol edilebilir olduğundan üreticilere doğru zamanda üretim yapma fırsatı sağlamaktadır. Bu nedenle, Hidroelektrik Santraline sahip bir şirket sistemin durumunu göz önünde bulundurarak elektrik üretmek için en iyi zamana karar vererek gelirini artırabilir. Bu çalışmada, birden fazla enerji santraline sahip ardışık HES'lerden oluşan bir sistem için üretim çizelgeleme ve gün öncesi piyasasında teklif kararlarının verilmesi probleminin modellenip çözülmesi amaçlanmıştır. Matematiksel model, üretim modeli ve teklif modeli olarak iki ana kısmı içeren entegre bir modeldir. Literatürdeki diğer çalışmalardan farklı olarak bu entegre model üretim planlamada planlama ufkunu gün alıp saatlik bazda üretim miktarlarına karar verirken aynı zamanda gün öncesi enerji piyasasında yine saatlik miktar ve fiyat teklifleri de oluşturmaktadır. Üretim modeli kısmında santralin çalışabilme kapasiteleri ve rezervuardaki su miktarları dengesi dikkate alınmıştır. Teklif modelinde kullanılan piyasa takas fiyatları, fiziksel elektrik ticaretinin yapıldığı gün öncesi piyasasında günlük olarak saatlik bazda belirlenir. Piyasa katılımcıları için doğru teklif miktarı vermek önemlidir, çünkü piyasa fiyatının üzerinde olan teklifler kabul edilmez. Piyasa takas fiyatları (PTF) teklifler verildikten sonra belirlendiği için modelin PTF belirsizliği altında çözülmesi gerekmektedir. Bu yüzden, modelde kullanılmak üzere PTF tahminleri yapmak için bir tahmin modeli geliştirilmiştir. Bu tahmin modelinin hata payına dayalı olarak PTF'ler için senaryo bazlı bir çözüm elde edilmesi amaçlanmıştır. Gün içerisinde değişkenlik gösteren PTF değerleri için saatlik bazda olacak şekilde regresyon analizi yapılmıştır ve bulunan beklenen değer ve standart sapmaya göre farklı senaryolar üretilmiştir. Geliştirilen senaryo bazlı entegre modelin performansı sadece beklenen PTF değerleriyle çalışan entegre modelle ve problemin ilk önce üretim çizelgesine karar verilip sonra teklif optimizasyonu yapıldığı çözümleri ile karşılaştırılmıştır. Ayrıca tahmin hatasındaki artışın entegre optimizasyon modelinin performansına etkisini gösteren bir numerik çalışma yapılmıştır. Ek olarak seçilen parametrelerin değişiminin entegre modelin amaç fonksiyon değerini nasıl etkilediğini görmek için duyarlılık analizi yapılmıştır.
Beklenmedik Uçak Yönlendirmelerini Azaltma: Zaman Serisi Analizi ve Yapay Sinir Ağları ile Modelleme
(TOBB University of Economics and Technology,Graduate School of Engineering and Science, 2019-01-01) Doğan, Hazal Berve; Hanalioğlu, Tahir
In this study, a decision support system is designed in order to minimize the number of flights that are diverted unexpectedly. The aim is to reduce the expenses that arise when the aircraft is not able to land on the targeted airport due to the unfavorable weather conditions, such as rescheduling the timetable, overuse of aircraft fuel than planned, passengers' accommodation and ticket reissue. In order to reduce such temporal and financial losses caused by diverted flights, decision to take off or not is made before departure, while the decision to land or not is made during flight, after a brief analysis based on weather data of target airport. For the aircraft to land on target airport as scheduled, it is crucial that the weather forecasts for visibility range, ceiling and wind speed are within the limits of the safe flight requirements. Considering the significance of this decision regarding by finance, there is a need for a decision support system that is capable of boosting the process through optimal decision-making by forecasting airport weather conditions. In the first part of the study, weather is forecast using regression and time series analysis, of which methods can be detailed as auto regressive (AR), moving average (MA), auto regressive integrated moving average (ARIMA) and vector auto regressive (VAR). Although such forecast methods are relatively effective in achieving the desired result, neural network and fuzzy logic techniques are expected to present more accurate forecast with their complicated and advanced algorithm structure. In the second part of the study, neural networks are created with using MATLAB. The results which is obtained with these methods are compared time series analysis results. Improvement is measured by accuracy of the decisions of diverted flights. The measurements are recorded on the confusion matrix.
Bir Açık İnovasyon Aracının Sistematik Tasarımı
(TOBB Ekonomi ve Teknoloji Üniversitesi - Fen Bilimleri Enstitüsü - Endüstri Mühendisliği Anabilim Dalı, 2011) Çubukçu, Ahmet; Gümüş, Bülent
The importance of the concept of innovation is being emphasized in all over the world. With the spread of information and communication technologies, the concept of open innovation began to be more popular as a subtitle of innovation. Hence, studies and projects on this issue have gained momentum. Open innovation practices are attempted to be performed by different techniques and approaches in different sectors, but there are still no standards or commonality established in this area. Although some companies can successfully use open innovation without relying too much on their internal R&D to achieve technological progress, the methods and tools are not well-defined yet. In addition, research shows that the managers and practitioners in companies who are practically using the open innovation techniques and approaches are not very well aware of the open innovation concept itself. Since the amount of knowledge has increased drastically and also knowledge can be accessed from practically anywhere, taking advantage of open innovation has become a necessity for companies. The validation of the required knowledge and the decisions of how and where to use the knowledge can be supported by effective open innovation tools. Companies can improve their competitiveness with the use of the open innovation web portals as an effective open innovation tool. Open innovation web portals are one of open innovation tools that bring together companies and innovators on Web 2.0 platforms to reach the goal of innovation-based results. In this thesis, a structure of an open innovation web portal is designed using Quality Function Deployment (QFD) and Systematic Design methodologies. In addition, a detailed literature review about innovation, open innovation and design methodologies is presented. A field study on the open innovation awareness and benchmark study of various open innovation web portals are also presented.
Bir Dinamik Hibrid Esnek Akış Atölyesi Çizelgeleme Problemi
(TOBB University of Economics and Technology,Graduate School of Engineering and Science, 2015) Uzel, Simay Sezgi; Kuyzu, Gültekin
This thesis is based on a real life hybrid flexible flowshop scheduling problem. Hybrid flowshop consists of multiple identical parallel machines in addition to characteristics of standard flowshop problem. Furthermore, with the possibility of skipping stages of jobs in the current manufacturing system is defined as hybrid flexible flowshop. The flexibility and being hybrid features of the system is usually ignored and is rarely discussed in the literature. In the phase of getting results for experimental studies of heuristic methods , the other papers in the literature that handle the hybrid flexible flowshop scheduling problems deal with the problems as much more limited than the real life. Aim of this thesis is to handle hybrid flexible flowshop scheduling problem with real life data. According to literature, hybrid flexible flowshop problem is already complex and studying with real life data increases the level of complexity. Additionally, while the problem is being analyzed, it is aimed to solve the dynamic scheduling problem to meet uncertain demands by digressing the assumption of being deterministic. Minimization of makespan is defined as the objective function. Heuristic methods are improved as a solution that is based on NEH algorithm with taking advantages of dynamic dispatching rules. Experimental studies are performed with open source library that is managed by Java based discrete event simulation. Thus, experimental data is generated and the effectiveness of heuristic algorithm is analyzed. Moreover, advised heuristic algorithms are studied with real life data. Consequently, the developed methods and the current methods, of which superiority were proved in literature, are discussed.
Bir Savunma Sanayii Firmasında Tedarikçi Seçimi Çalışması
(TOBB University of Economics and Technology,Graduate School of Engineering and Science, 2018) Demir, Kübra Nur; Tekin, Salih; Hanalioğlu; İç, Yusuf Tansel
Supplier selection is one of the most important task for a company. In order to reach manufacturing aims and ensure customer satisfaction, every company would like to work with a trustworthy business partner. Working with a strong supplier mitigates risks in the manufacturing process and enhances competitive capacity and profit. Therefore, supplier selection is a strategical decision making problem. In today's world producing all the sub-parts in the same company is an inefficient planning. For this reason, companies are in need of different suppliers. In this study, supplier selection problem for a project which will kick off in defence company producing air vehicles is considered. Our aim is to minimize purchasing cost and maximize quality level and total weight of the purchased products. Supplier selection problem is addressed by a three-phased method. In the first step, candidate suppliers are evaluated for their technical competences with TOPSIS (Technique for Order Preference by Similarity to Ideal Solution). Appropriate suppliers are scored subjectively with AHP (Analytic Hierarchy Process) method in the second step. During the assessment, quality, compliance requirements, price, risk, safety, technological level and brand image are considered. One of the most important criteria for the companies in the aerospace or space sector is the quality. Supplier firms have to satisfy certain quality management system standards. In the last phase, orders are assigned to suppliers using objective programming with company's aims and constraints. Material costs are handled with a piecewise linear function in programming and lead times with total demand requirements are also included in the problem constraints. For the first model, lead times consist of mean values, in the second model. Leads times are considered as stochastic, that are handled with an heuristic approach inspired from Markowitz method. Company's previous orders are solved with newly developed method and results are compared. In addition, we also solve the supplier selection problem with another method from literature and compare the results. Sensitivity and scenario analyses are done in order to evaluate situations that may occur in the future or decrease value at risk in decision processes assessed either objectively or subjectively. During analyses, weights of criterias in AHP method, priorities in goal programming and confidence levels are investigated. Keywords: Multiple criteria decision making problem, Supplier selection, Multi objective programming, TOPSIS, AHP, Taguchi method, Markowitz method
Bir Savunma Sanayii Firmasında Uçak Komponenti Montaj Hattı Dengeleme ve İşgücü Atama Problemi için Genetik Algoritma Uygulaması
(TOBB ETÜ, 2021) Kılıç, İrem; Hanalioğlu, Tahir; Erdebilli, Babek
Montaj hattı, belli bir büyüklükteki üretim kapasitesine sahip, nihai veya ara ürünlerin üretilmesi için gerekli operasyonların belirli bir sırayla gerçekleştirildiği iş istasyonlarından oluşan üretim yapısıdır. Montaj Hattı Dengeleme, üretimde gerçekleştirilmesi gereken operasyonlar arası öncelik ilişkisini sağlayacak şekilde, önceden belirlenen diğer kaynak, kapasite, işgücü, yerleşim gibi kısıtları göz önünde bulundurarak operasyonların farklı iş istasyonlarına atanması yaklaşımıdır. Bu çalışmada bir savunma sanayii firmasında tek modelli bir uçak komponenti montaj hattı ele alınmıştır. Operasyon süreleri zaman etüdü çalışmaları ile ölçülmüştür ve uygulama kapsamında deterministik olarak kabul edilmiştir. Yapılan iş etüdü çalışmaları ile ele alınan tek modelli uçak komponenti montaj hattında gerçekleştirilecek operasyonların hangi sıra ile gerçekleştirileceği bilgisi elde edilmiştir.Montaj hatlarının tasarımı ve dengelenmesi başta otomotiv olmak üzere, beyaz eşya ve elektronik sektöründe de üretim verimliliği açısından oldukça önemlidir. Uçak komponenti montaj hattı, diğer sektörlerdeki montaj hatlarından farklı olarak daha fazla emek yoğunluklu çalışmaktadır. İşgücü atamasının verimliliği arttıracak şekilde yapılması kritik öneme sahiptir. Her bir operasyon için farklı araç/gereçlerei, takımlara ihtiyaç duyulmaktadır. Bu kısıt ile birlikte her montaj operasyonunun her montaj istasyonunda gerçekleştirilemeyeceği bilinmektedir. Çalışmada konu alınan tek modelli uçak komponenti montaj hattında, montaj hattı dengeleme ve işgücü atama problemi için yapılan uygulamada her bir operasyon ve her bir operasyonun gerçekleştirilebileceği potansiyel montaj istasyonları kısıt olarak yansıtılmıştır. Firmanın ele alınan uçak komponenti montaj hattı için ayıracağı işgücü ve bütçe önceden belirlenmiş olup kısıt olarak uygulamaya yansıtılmıştır. Operasyon sürelerinin operasyonu gerçekleştiren operatörün beceri seviyesiyle ilişkili olarak farklılık göstermesi dikkate alındığında problem için bütçe kısıtlı montaj hattı çevrim süresi minimizasyonu amaçlayan bir matematiksel model geliştirilmiştir. Belirlenen kısıtlar ve varsayımlar göz önünde bulundurularak, görevler arası öncelik ilişkisi sağlanacak şekilde, çevrim süresi minimizasyonu hedefleyen matematiksel modelin doğruluğu küçük boyutlardaki problem verileriyle GAMS programında modellenerek test edilmiş ve en iyi çözüm elde edilmiştir. Büyük boyuttaki gerçek hayat verileri, Python 3.7.6 programı aracılığıyla geliştirilen genetik algoritma ile uygun çözüme ulaşılmıştır. Böylelikle minimum çevrim süresi hedeflenen uçak komponenti montaj hattında, hangi istasyonda hangi operasyonunun çalışılacağı ve hangi operasyonda hangi operatörün çalışacağı belirlenmiştir.
Bir Traktör Fabrikasında Karışık Modelli Montaj Hattı Dengeleme – Deterministik ve Stokastik Ölçümlere Göre Analizler
(TOBB ETÜ Fen Bilimleri Enstitüsü, 2020) Uyanık, Aysın Şenel; Tekin, Salih
Assembly lines are systems that allow the products with multiple components and, mass demands, to be manufactured together in sequential workstations. Assembly lines and design are among the main factors that increase the efficiency of production in factories in the white goods and electronics sectors, especially in the automotive sector. Numerous studies have been carried out using deterministic, stochastic heuristic and fuzzy methods related to the design and balancing problems of assembly lines. In this study, the distribution of work steps to the operators during the daily working period was carried out in a real assembly line of a tractor factory. Since the parts belonging to more than one type of tractor model are assembled in the factory assembly line, mixed model assembly lines are examined. The duration of the operations for the assembly was measured by time study studies. Accordingly, it was observed that the working times were random and dependent on the worker. In a parallel type assembly line, worker-dependent working times are also taken into account randomly. Line stoppages are experienced on the assembly line due to factors such as lack of training, the presence of complex jobs, and reduced motivation of work. Line stops cause deviations in the number of tractors that need to be produced daily, thus causing losses in the factory. In a mixed model assembly line where multiple work steps take place, there may be minor variations between operating times. Balancing analyzes were used on the assembly line to measure all losses. Within the scope of the study, considering the work step differences in tractor models, a priority diagram was obtained with the model-based unified priority diagram method. Stochastic assembly line balancing study was performed according to the stochastic working times on the line and the analysis results were compared with the deterministic results. In addition to the advantages of using stochastic working times, heuristic results have been obtained for the solution of the assembly line balancing problem with the java algorithm, which gives the minimum number of stations to be opened in the cycle time determined using the intuitive method Arcus (1966) proposed COMSOAL method. The number of stations to be opened according to the algorithm is determined at different cycle times. Thus, a line structure prepared for sudden plan changes at the factory was obtained. According to the station distribution obtained as a result of balancing and job assignment on the assembly line, efficiency analyzes are presented and the results are evaluated.
Bir Yüksek Hızlı Tren Hattında Yolcu Sayısı Tahmin Modelleri Geliştirilmesi
(TOBB ETÜ Fen Bilimleri Enstitüsü, 2020) Ertem, Beyza Nur; Ertoğral, Kadir
Correct and reliable order forecasting has an important role in increasing the efficiency of companies. Similar to other sectors, the demand forecasts are very important in Food industry. Especially in one day shelf life products, it is a must to make accurate order forecast and prevent food waste excess. In this study an approach is developed for forecasting the passenger numbers in the the high speed train leg between Ankara and İstanbuls. The firms is a subcontractor to The Republic of Turkey State Railways. As a first step, it was aimed to create a daily order system and the forecasting with multiple regression of the passenger numbers who traveled Ankara-İstanbul and İstanbul-Ankara route in 'Business Plus' and 'Economy Plus' coaches using data from a catering company between 2016-2019 year. The accurate Passenger number estimation will create a better meal order and will lead into a financial savings. In our study several forecasting models are constructed based on different ways of data groping and the choices of independent regression variables. We selected the best three models among the models develop and we found the mean absolute percent error performance for selected models, which turn out to be quite accurate. In the second stage our study, using the same passenger data the Winters exponential smoothing method is studied to forecast the weekly meal consumption and the forecast accuracy is tested using MAPE values.
Bulanık Fonksiyonlar ile Bulanık Sistem Modelleme
(TOBB Ekonomi ve Teknoloji Üniversitesi, 2009) Ünal, Başak; Türkşen, İsmail Burhan
Bu çalışmada, yeni bulanık sistem modelleme yaklaşımları önerilmiştir. Önerilen yöntemlerin ve bu çalışmaya referans teşkil eden Türkşen'in [2008] ve Çelikyılmaz ve Türkşen'in [2007, 2008, 2009] önerdiği geliştirilmiş bulanık fonksiyonlar ile geliştirilmiş bulanık öbekleme yaklaşımının performansları yapılan deneyler ile analiz edilmiştir. Bu çalışma ile Hataway ve Bezdek [1993] tarafından önerilen bulanık öbek regresyon modeli ve Höppner ve Klawonn [2003] tarafından önerilen bulanık model öbekleme algoritmalarına sistemi modellemek için iki yeni çıkarsama algoritması önerilmiştir. Buna ek olarak, Höppner ve Klawonn'un [2003] önerdikleri bulanık model geliştirilerek alternatif bir öbekleme algoritması (Alternatif Bulanık Model - ABM) önerilmiştir. Buna bağlı olarak, sistemi modellemek için ABM'e de bir çıkarsama algoritması önerilmiştir. Öneriler ışığında geliştirilmiş modeller, önceki modeller ile karşılaştırılarak, önerilerin yeni sonuçları vurgulanmıştır.
Bulanık Regresyon Modellerinin Tutarlılığı Üzerine Çalışmalar
(TOBB Ekonomi ve Teknoloji Üniversitesi Fen Bilimleri Enstitüsü, 2014) Özşahin, Selcen Gülsüm Aslan; Türkşen, İsmail
In today's World, collecting data is not a big issue in a proper and processable format with the help of recent technological developments but it is a big deal not only for the government bodies, private sector and also for the individuals. The main idea behind keeping data in a processable format is to discover the existing data patterns to estimate the future with the highest consistency. Enhancements in data collection, processing and estimation technics bring the strong knowledge from the past, ability to control processes and dominate the future based on the historical data. In other words, technical control on data management promises the power of knowledge to the all stakeholders in management and in many areas mainly in economics, engineering, and medical sciences as well as politics and public relations. One of the leading research area in data management is modeling which are namely stochastic models, statistical models, lineer modeling or fuzzy systems. Fuzzy Models are one of the most trendy and consistently estimating approach for modeling. In this study, the data set of Denmark which contains financial indicators as input and import amounts between 1993 and 2013 as output were used to create 221 different models by using Fuzzy C-Means Clustering Algorithm and Fuzzy Regression Analiysis with different number of clusters and degree of fuzziness in each model. With the help of high number of experimental models, evaluation of model consistency parameter, R-square, depending on the changes of number of clusters and degree of fuzziness have done and several inferences have been achieved. Last but not least, outstanding approach has been developed in this study which is named as Multi-Layer Fuzzy. In this approach more than one membership matrix calculated with at least two different number of cluster and degree of fuzziness values have been used in a single model and higher consistency in the estimations have been achieved as expected
Bulanık Sistem Modelleme
(TOBB Ekonomi ve Teknoloji Üniversitesi Fen Bilimleri Enstitüsü, 2014) Dilden, Elif Burcu; Türkşen, İsmail
While constructing a system model to find out input/ output relation, fuzzy functions (FF) method use Fuzzy c- means (FCM) algorithm to determine membership values. Instead of determining fuzzy rule bases (FRB) for each cluster, by using FF method each cluster is explained by functions. Proposed FF approach says that membership values and their user defined transformations should be add into the input matrices as independent variable with the real inputs. This suggestion is heuristic. In this study, we defined Fuzzy Normal Equations (FNE). Our proposed fuzzy normal equations explain the detail of how to add various transformations of membership values into the input matrices in fuzzy functions approach. We have modelled two real life data sets. One of them is Concrete Compressive Strength and the other one is Apple's Closing Stock Price data set. Fuzzy model performances compared with two benchmark strategies: Artificial neural networks (ANN) and adaptive neuro fuzzy systems (ANFIS).
Bütünleşik Performans Değerlendirme Yöntemi Önerisi ve Performans Tabanlı Bölüm Başarı Değerlendirmesi: Kurumsal Bir Şirkette Örnek Çalışma
(TOBB ETÜ Fen Bilimleri Enstitüsü, 2020) Şenel, Tahsin Uğur; Tekin, Salih
Enterprise companies, which make future strategic plan, need to view overall performance. In this direction, monitoring decision-making unit (DMU) performances is a critical issue. Therefore, it should be fair and consistent performance evaluation and prepared open and clear reports. To handle with these requirements, this study focusses on establishing a comprehensive method of performance evaluations (PE). As a framework of the study, in scope of effectivity and efficiency aspects, five main solutions are suggested to solve five main problems in PE. In the paper, football theorem is selected as a strategic concept. Then, right PE tool are selected in terms of strategic frame of companies firstly. In this study, we examine Balanced Scorecard (BSC) approach that considers not only financial but also non-financial topics to watch overall performance. BSC has key performance indicators (KPI) that show objectives and actualizations belonging to related DMU, and BSC also provides proper and summary reports including different perspectives. The next step is determination of KPI's weights. Analytic Hierarchy Process (AHP) is applied to determine KPI priorities. As a result of first two steps, effectivity could be followed easily for a company. Because BSC evaluates DMU'S separately, it does not provide satisfied comparison among different DMU's. Therefore, we propose Data Envelopment Analyses (DEA) that is linear program based non-parametric approach to follow efficiency. However, DEA works correctly for only homogeneous DMU's. As a third step, classification process is applied to ensure homogeneity. Then, using BSC KPI's as outputs and DMU budgets as input, DEA model is run for each class. As a last step, we separate DMUs into categories using effectivity and efficiency score obtained from previous stages. To determine category numbers, Hierarchical Clustering Analyze (AHCA) method is used and group elements are selected with applying K-Means Clustering Analyze technique. At the end, two case studies are given to show how developed model is applied within an enterprise company.
Çift Tutuculu Malzeme Taşıyıcı Robotlu Hücrelerde Enerji Duyarlı Çizelgeleme
(TOBB University of Economics and Technology,Graduate School of Engineering and Science, 2019) Emiroğlu, Nurdan; Gültekin, Hakan
Robotic cells are serial production systems that consist of a number of machines and a material handling robot that transfers parts between the machines. At the same time, the robots in the system function to load/unload the machines. In the academic literature and the business world, minimization of the cycle time is the dominant objective for the optimization of production schedules. In parallel to this objective, robot movements are assumed to be at their maximum speeds and the energy savings that can be attained with the dynamic adjustment of the speeds is not taken into account. Within the scope of this study, considering the controllability of the move speeds of the dual-gripper robot, it is aimed to determine the optimal robot move sequence and the operation parameters of the robot. In other words, we considered the energy consumption and throughput rate objectives simultaneously as a bi-criteria optimization model. In order to solve this bi-criteria model we used the epsilon-constraint approach, one of the objectives is written as a constraint. For the problem, a mixed integer non-linear mathemetical model (MINLP) is developed. In order to increase the solution efficiency, the problem is reformulated as a mixed integer second order constrained mathematical model (MISOCP). Computational tests revealed that the MINLP model cannot guarantee the optimal solution in reasonable times for systems with two or more machines. Since it is also observed that the MISOCP model is not efficient for the systems with six or more machines, a heuristic solution method called Efficient Solution Generation algorithm (ETA) is developed. By using the created data sets, these three solution approaches are investigated through a computational study. Additionally, as a result of these tests, it is observed that an average energy savings of 17.7\% can be attained by considering controllability of robot move speeds.
Çift Tutuculu Robotik Hücrelerde Üretim Hızının En Büyüklenmesi: Sade Çevrimler
(TOBB Ekonomi ve Teknoloji Üniversitesi Fen Bilimleri Enstitüsü, 2013) Dalgıç, Özden Onur; Gültekin, Hakan
In this study, we discuss pure cycles in $m$-machine dual gripper robotic cells. A material handling robot loads/unloads the machines and transports the parts between the machines. The robot is assumed to have dual grippers so that it can carry two parts simultaneously. Identical parts are to processed on machines which have the ability to perform all the operations that a part requires. Manufacturing systems in which there are flexible machines like CNC machines have this kind of ability. We consider pure cycles where each part is processed by only one of the machines and each machine processes exactly one part in a cycle. We first determine the feasibility conditions for the pure cycles and an algorithm that generates all feasible pure cycles for a given number of machines. We analyse 2-machine robotic cells in detail and prove that one of the particular five pure cycles among a huge number of feasible pure cycles always provides the optimal solution. We determine the parameter values in which each of the five cycles is optimal. In 2-machine robotic cells, we show that pure cycles in dual gripper robotic cells always dominate pure cycles in single gripper robotic cells. By using problem parameters, an experimental study is performed and the average and maximum benefits of using a dual gripper robot instead of a single gripper robot are determined. The effects of problem parameters are also examined.

Browse

Browsing Endüstri Mühendisliği Yüksek Lisans Tezleri / Industrial Engineering Master Theses by Publication Category "Tez"