Hücresel imalat sistemlerinde maliyet ve sinir ağları tabanlı iki evreli bir kümelendirme yaklaşımı
Artificial neurat network x operation costs based twostage GT clusterning procedure
- Tez No: 46191
- Danışmanlar: DOÇ.DR. BÜLENT DURMUŞOĞLU
- Tez Türü: Yüksek Lisans
- Konular: Endüstri ve Endüstri Mühendisliği, Industrial and Industrial Engineering
- Anahtar Kelimeler: Belirtilmemiş.
- Yıl: 1995
- Dil: Türkçe
- Üniversite: İstanbul Teknik Üniversitesi
- Enstitü: Fen Bilimleri Enstitüsü
- Ana Bilim Dalı: Belirtilmemiş.
- Bilim Dalı: Belirtilmemiş.
- Sayfa Sayısı: 84
Özet
ÖZET Artan pazar talebinin kaçınılmaz sonucu olan büyümenin, geleneksel imalat sistemlerinde yarattığı olumsuz etkileri ortadan kaldırmak yıllardır akademik çevrelerin ve modern endüstrinin ilgi odağı olmuştur. Gerçekte fonksiyonel yerleşime sahip bu imalat birimleri planlama, çizelgeleme,organizasyon, motivasyon ve kontrol konularında yeterince etkin bir çerçeve çizememektedirler. Merkezi bir yönetim anlayışı ile tüm problemlere çözüm bulmak ve günümüz rekabet piyasasında müşterilere uygun kalitede zamanında teslim olanaklarım sunabilmek karmaşık iş akışına sahip bu tür sistemler için çok güç olmaktadır. İşte bu anlamda çözüm, fonksiyonel yerleşime sahip kesikli imalat sistemlerinin hücresel imalat sistemlerine dönüştürülmesidir. Malana hücreleri ve parça ailelerinin oluşturulması ise hücresel imalat sistemlerinin en temel sürecini oluşturmaktadır. Yapılan bu çalışmada, hücresel imalat sistemlerinde, yapay sinir ağlan tabamı benzer proses özelliklerine dayanarak parça aileleri ve makina hücrelerini oluşturan iki evreli bir kümelendirme algoritması geliştirilmiştir. Algoritmanın birinci evresi alternatif rotaları ve kapasiteyi gözönünde bulunduran bir tamsayılı programlama modeli, ikinci evresi ise hücreler arası hareket minimizasyonunu hedefleyen bir sinir ağı kümelendirme modelidir. Çalışma bir cam kalıp fabrikasında yöntemin uygulamasını ve mevcut sisteme uygulanan bir diğer sezgisel yöntemle karşılaştırılmasını bu yöndeki üstünlüklerini ve eksikliklerini ortaya koymaktadır. Tezin ilk bölümünde, genel hatları ile Grup Teknolojisi konusu ele alınmıştır. İkinci bölüm ise grup teknojisinde bu konuda geliştirilen kümelendirme yöntemlerinin ayrıntılı bir incelemesine ayrılmıştır. Üçüncü bölüm, genel hatları ile yapay sinir ağlarını işlerken, dördüncü bölümde kullanılan yöntem ayrıntılı olarak açıklanmıştır. Son bölüm ise uygulama ve sonuçlan içermektedir. vıı
Özet (Çeviri)
SUMMARY ARTIFICIAL NEURAL NETWORK & OPERATION COSTS BASED TWO STAGE GT CLUSTERING PROCEDURE Group Technology is a manufacturing philosophy which identifies and exploits the underlying sameness of items and the processes used for their manufacture. In other words, it's a method of technological process development, equipment planning and efficient setting of the machine tool, so as to insure the most profitable technical planning of production in the shortest time. [SNEAD S. Charles]. In today's complex manufacturing environment, we must constantly be aware of any tecnology that will aid us in improving or performance. GT is such a tecnology and must be applied in its brodest sense. To respond customer's orders with acceptable quality as soon as possible, manufacturing systems must be designed effectively and it is incredibly important restructuring process which changes the whole manufacturing structure. One tenet of group tecnology entails dividing the manufacturing facility into small groups or cells of machines, each cell being dedicated to a specified set of part types. The term“cellular manufacturing”is often used in this regard. The connotation of a cell is a small group of perhaps only one or two machines but seldom more than five. A typical cell might contain a machining center, on-machine inspection and monitoring devices, tool and part storage, a robot for part handling, and the associated control hardware. Configuring machines with different capabilities into a cohesive group is an alternative to process layout. This group configuration is most appropriate for medium-variety, medium-volume environments. If volumes are very large, pure item flow are possible; if volumes are small, and parte are varied to the point of only slight similarities between jobs, then there is less to be gained by grouping. Neverthless, GT can produce highly significant improvement where it is appropriate and presents important lessons that should be utilized in all manufacturing environments. The Group Tecnology (GT) problem is that of classifying parts and machines into part families and machine cells of efficient production based on certain mesurable commonalities. For example, by grouping similar parts one can take advantage of their similarities in design and manufacturing. A part family may share the same setup processing, routing and so on and may approach economies of scale of mass production. Similarly, by grouping machines together, intercellular travels can be reduced, thereby minimizing and in some cases eleminating material handling Vlllcosts. Furthermore, reductions in setup time, manufacturing lead time design variety, and work-in-process, inventory can be achieved. GT problem is therefore that of problem solving through the determination of appropriate family membership of the entities to be grouped. When two or more machines or machine centers are grouped together to form machine cells used in the manufacture of part families in group tecnology, their operations are refered to as cellular manufacturing (CM), a term which is often used interchangeable with Group Tecnology. Based on these advantages of group tecnology, identification of machine- cells is one of the most important problems in the design of cellular manufacturing systems. Several algorithms with varying degrees of success have been proposed and utilized to solve this problem. By the way, the tecnological developments in today's modern manufacturing environment have a rapid growth and new tecniques are developed for grouping parts and machines. Artificial Neural Networks are one of these, which have a recent surge of interest. The present study analyzes eleven of the promising cell formation tecniques in GT and gives a general review of artificial neural networks, information processing centers. Artificial Neural Systems (ANSs), which have parallel information processing structure, are mathematical models of theorized mind and brain activity. ANSs exploit the massively parallel local processing and distributed representation proporties that are belived to exist in the brain. The primary intent of ANNs is to explore and reproduce human information processing tasks such as speech, vision, olfaction, touch, knowledge processing and motor-control. In addition, ANNs are used for data compression, near-optimal solution to combinatorial optimization problems, pattern matching, system modeling, and function approximation. In this study, Carpenter & Grossberg 's ARTİ neural network model (and modified ARTİ developed by C. SURESH & P. CERVENY) is analysed and used in the second stage of application. In the first division of the thesis, Group Tecnology is introduced. In addition, traditional manufacturing systems and cellular manufacturing systems are disscused. Three methods for identifying machine-part families have been identified. The first of these taxonomic groups is the ocular or visual method, and the other two methods are part characteristic based and production process based methods. In the second chapter of the thesis, eleven of the promising cell formation tecniques introduced in GT. These are; Production Row Analysis, Binary Ordering Algorithm, Single-Pass Heuristic, Similarity Coefficients, Single Linkage Clustering, IXAvarage Linkage Clustering, Branch & Bound Technique. The Modified Rank Order Clustering,MROC, Algorithm, An Ideal Seed Non-Hierchial ClusteringJSNC, Algorithm, SC Seed Algorithm, The Bond Energy Clustering Algorithm, BEA. The third chapter of thesis gives a general review of Artificial Neural Networks considering processing elements and topology characteristics. In addition, memory, recall, and learning subjects in ANNs are discussed. In the fourth chapter of thesis, Carpenter - Grossberg Neural Network Model and Kaparthi, Suresh & Cerveny '93 Neural Network Procedure are introduced. In the next chapter, a two stage clustering procedure is presented. The first stage of the procedure is an integer programming model which considers alternative routings. Objective function of the model minimizes operation costs under capacity constraints. The data which is obtained from the first stage of the procedure is structured as a binary machine-part matrix and used in the second stage. The second stage of the procedure is an artificial neural network model which minimizes intercell travels considering capacity conditions with alternative routeings. An application of this two stage clustering algorithm is performed in a factory and the present method is compared with an alternative procedure in the last chapter of thesis. LINDO linear programming model is used in the first stage to select routings which have minimum operation costs. J. L. Burbidge has noted seven characteristics of successful groups. These are should be embedded into the group stucture. Size is important The group must be small enough to act as a close-knit team with a common goal, yet be large enough to contain all necessary resources for achieving that goal. Although as large as 30 workers have been successfully implemented, social scientists have learned that a range of 6 to 12 workers is best ( the magic number 7 is, of course, optimal ). As group size grows beyond a dozen, it becomes increasingly difficult to maintain sufficient camaraderie, level of cornminucation, and cohesiveness toward a comman goal. It is likewise important to remember to level the long-run work load ( utilization ) in each group and to match this to production plans. Failure to balance work load creates a strong temptation to route extra work to the cell or to off-load jobs to other cells, thus destroying the concept of Group Tecnology. To facilitate theorganizational authorization process, it may be advisable to maximize utilization of existing machines and to minimize investment requirements. Characteristics of successful groups defined by J. L. Burbidge as follows; Characteristic Team Products Facilities Group Layout Target Independence Size Desciption Specified team of dedicated workers Specified set of products, and no others Specified set of dedicated machines / equipment Dedicated contiguous space for specified facilities Common goup goal, established at start of each period Buffers between groups;groups can reach goals independently Preferably 6-15 workers;more can be accommodatecLhowever Groups should be designed with safety in mind. Painting may be required after welding, but placing these functions too close may not be the spark of enligtenment for which you are searching. Likewise, machines and processes that are imcompatible for other reasons, such as tolerance capability or environmental factors, should not be group together. A discussion of group technology principles would not be complete without some comment on the impications for organizational structure. The organization should be structured around groups. Each group performs functions mat in many cases were previously attributed to different functional departments. Labor reporting and the entire accounting system should be based on group activity. For instance, inmost situations employee bonuses should be based on group group performance. If skill or experience levels vary significantly, it is possible to assign shares of the bonus based on relevant, quantified measures, but the entire group should be rewarded from a common evaluation. It is obviously clear that classifiying parts into part families and machines into machine cells is very important for an organization. Group Tecnology has a lot of benefits. Houtzell and Brown of Organization for Industrial Research, Inc. list the following areas of savings that their clients have experienced; XIConsidering these advantages of GT, it is easy to say that the use of cellular manufacturing is a strategy for the factory of future. [DURMUŞO?LU B.] Xll
Benzer Tezler
- Design of additively manufactured hybrid structural brackets via topology optimization
Topoloji optimizasyonu ile eklemeli imalata uygun hibrit yapısal braketlerin tasarımı
CENGİZ KÖSEOĞLU
Yüksek Lisans
İngilizce
2023
Makine Mühendisliğiİstanbul Teknik ÜniversitesiMakine Mühendisliği Ana Bilim Dalı
DOÇ. DR. MESUT KIRCA
- Grup teknolojisi imalat sistemleri tasarımı için bir metodoloji ve bu metodolojinin endüstride uygulanması
Başlık çevirisi yok
NEVİN AYDIN
Doktora
Türkçe
1998
Endüstri ve Endüstri Mühendisliğiİstanbul Teknik ÜniversitesiEndüstri Mühendisliği Ana Bilim Dalı
PROF. DR. M. BÜLENT DURMUŞOĞLU
- Hücresel imalat sistemlerinde insan-makina hücreleri oluşturmak için çok amaçlı bir matematiksel programlama yaklaşımı
A multi objective mathematical programming approach for human-machine cell formation problem in cellular manufacturing systems
GÖKÇE KILIÇ
Yüksek Lisans
Türkçe
2008
Endüstri ve Endüstri MühendisliğiGazi ÜniversitesiEndüstri Mühendisliği Ana Bilim Dalı
YRD. DOÇ. DR. FEYZAN ARIKAN ÖKTEMER
- Hücresel üretimde tesis düzenlemesi ve bir uygulama
Cellular layout and an application
AHMET ÇEVİKBAŞ
Yüksek Lisans
Türkçe
2002
İşletmeGazi ÜniversitesiSayısal Yöntemler Ana Bilim Dalı
DOÇ. DR. HALUK SOYUER
- Grup teknolojisinde parça ailesi ve imalat hücresi oluşturma: Bir örnek inceleme
Part family and manufacturing cell formation in group technology: An example study
ERHAN BABALI