Aksiyomlarla Tasarım Yaklaşımı İle Robot Kolu Seçimi İçin Bir Karar Destek Sistemi

Abstract

Küreselleşmenin etkisi ile birlikte yoğun rekabet ortamı, işletmelerin yönetim anlayışlarının değişmesine neden olarak, üretim faktörlerini daha etkin kullanmak amacıyla stratejiler belirlemeye ve uygulamaya zorlamaktadır. İşletmeler, sürdürülebilirliklerinin sağlanması amacıyla fiyat, kalite, verimlilik, hızlı yanıt, çevre yönetimi, ürün çeşitliliği ve esneklik üzerine stratejiler üretmeye ve bu stratejileri rekabet gücü olarak kullanmaya başlamışlardır. Bunun yanında; teknolojide meydana gelen bu hızlı gelişimin etkisi ile; teknoloji kavramı, insan yaşamının vazgeçilmez bir parçası haline gelmiş ve işletmelerin üretim faaliyetlerinde bu stratejilerin uygulanması için yer edinmiştir. Buna paralel olarak da, üretimde insan faktörünün yerini robot sistemleri almaya başlamıştır. Bütünleşik birçok alt sistemden oluşan robot sistemleri karmaşık yapıya sahiptir. Bu yapıları nedeniyle robot sistemlerinin tasarımı yapılırken, uygun alt sistemlerin tasarımı önemli ve zor bir süreçtir. Robot sistemleri tasarım sürecinin temel aşamalarından biri olan üretim sistemine uygun robot kolu seçimi işlemi, sistemin işlevselliği ve amacına ulaşması bakımından önemli bir konudur. Bu çalışmada, üretim sistemine uygun robot kolu seçimi amacıyla çok ölçütlü karar verme sürecinde kullanılan ve bilimsel bir metod olan Aksiyomlarla Tasarımdan yararlanılarak “endüstriyel robot kolu seçimi metodolojisi” önerilmiştir. Çalışmada literatürden yararlanılarak; birbirinden bağımsız olduğu ve Aksiyomlarla Tasarımın bağımsızlık aksiyomunu gerçekleştirdiği varsayılan endüstriyel robot kolu seçimi ölçütleri saptanmıştır. Belirlenen ölçütler, robot kolundan beklenen özellikler ve kısıtlar dikkate alınarak filtreleme ölçütleri ve sayısal ölçütler olarak gruplandırılıp, iki ana aşamadan oluşan metodoloji yardımıyla seçim işlemi gerçekleştirilmektedir. Birinci aşamada, robot kolu alternatifleri filtreleme ölçütlerinden yararlanılarak süzülmüş; ikinci aşamada ise elemeden geçen her bir robot kolu alternatifine ait bilgi içeriği bilgi aksiyomu prensibinden yararlanılarak hesaplanmıştır. Sayısal ölçütler, kontrol edilebilen ve kontrol edilemeyen ölçütler olarak sınıflandırılmış; tasarım ve sistem aralıkları kavramları bu ölçüt tiplerine göre tekrar tanımlanarak bilgi içeriği hesaplanmasında bu kavramlardan yararlanılmıştır. Optimum robot seçimi “en iyi tasarım minimum bilgi içeriğine sahiptir” prensibine diğer bir deyişle bilgi aksiyomuna dayalı olarak yapılmıştır. Bununla birlikte; bir tür karar destek sistemi, yazılımla bütünleştirilerek geliştirilmiştir. Bu karar destek sistemi, gıda sektöründe faaliyet gösteren bir firmanın paketleme süreci için endüstriyel robot kolu seçimine uygulanmıştır.

The intense competitive environment with impact of globalization, which causes changes in business management concepts, enforces firms to identify and apply strategies in an effort to use production factors more efficiently. In order to provide maintainability, firms start to settle strategies based on price, quality, productivity, rapid response, environmental management, product diversity and flexibility and besides, they begin to use these strategies as competitive weapons as well. Moreover, with the effect of rapid development in technology, technology concept has become an indispensable part of the human life and it takes place in production activities of enterprises. Correspondingly, robotic systems start to take a part in production area instead of the humans or with the humans. Robotic systems are preffered because of very different talents. Robotic systems consisting of several integrated sub-systems have complex structure. Due to this structure, the design of appropriate sub-systems is significant and difficult process during designing robotic systems. As one of the main design process of robotic systems, the selection process of suitable robot arm is important issue for the functionality and success of the system. With the development of robot industry, robots with various capabilities have wide range application area and there are many different robot arm alternatives having different qualifications. Large number of robot arm alternatives and differing features complicate the selection process. In the pre-selection process of robot arm, robot supplier uses intuitive methods to choose robot arm for the simulation test. Considering the number of alternatives and conflicting features, decision making process requires to be based on a scientific foundation. In this study; to select appropriate industrial robot arm for a specific production system, industrial robot arm selection methodology based on axiomatic design is proposed, which is utilized as a scientific tool in multi attribute decision making process. Axiomatic Design (AD) principles are employed for the selection of the most suitable industrial robot arm that satisfies the functional requirements of a certain design. AD establishes a scientific basis for design; improve design and decision-making by providing the designer a theoretical foundation based on logical and rational thought processes and tools. There are two axioms of AD. The information axiom states that the design having the least information content is the best one. Besides, the independence axiom requires that independence of functional requirements of a design must be maintained. Criteria for industrial robot selection assumed as realizing independent axiom of axiomatic design are determined by reviewing the literature. The robot selection criteria are classified into two groups as filtering criteria and evaluation criteria by taking into acount the expected properties of robot arm and the constraints. The design parameters which serve as prerequisites or have representational structures are classified as filtering criteria. The design parameters which ensure tradeoffs in determined ranges are classified as evaluation criteria. By using these criteria, two-staged industrial robot arm selection methodology is proposed. In the first stage, robot arm alternatives are eliminated by using filtering criteria and then optimum robot arm selection is made by calculating information content of the evaluation criteria in the second stage. In the study; filtering criteria are defined as the criteria, validity of which must be ensured and chosen as the degrees of freedom, protection class, connection type, nominal payload capacity, wrist reach distance etc. These criteria are used to eliminate the robot arm alternatives which does not satisfy the customer requirements. Because of filtering process, the information content of unnecessary robots are not calculated in the early stage, so the selection process progresses more rapidly. The evaluation criteria are the criteria whose information contents would be calculated in the assessment process. The evaluation criteria are classified as controllable and non-controllable criteria. The concepts of design range and system range are re-defined and the information contents of each of the alternatives are calculated by means of the information axiom principle. Some of the evaluation criteria can get all the desired values between the lower and upper limits, because of the programmable structure of robots. In these limits, the probability value of success to satisfy the functional requirement (FR) is one. Because the target destination can exactly be reached between the lower and upper limits depending on the nature of the work. These criteria are classified as controllable criteria. On the other hand, some of the evaluation criteria are not controllable. Due to the nature of the system, the values of criteria deviate from the target point. These criteria are given in terms of tolerance and classified as non-controllable criteria. In this study, the controllable criteria contain speed of each axis and the non-controllable criteria contain repeatability and investment cost. The optimum robot selection is done according to principle of “the design that has smallest information content is the best design” that is called the information axiom after calculating the information content of each alternatives. The proposed methodology is applied in a pilot study which contains evaluation of three industrial robot arm alternatives before the real life application. While determining the system range and design range for the criteria, the information content of the three robot arms is calculated and optimum robot arm is chosen according to information axiom principles. Besides, a kind of decision support system integrated with software has been developed and as an application, this decision support system has been implemented in packaging process of a food industry to select an industrial robot arm. Seventeen robot arms are evaluated by the decision support system based on the proposed methodology. The decision support system consists of three main components in integrated manner: database, inference mechanism, user interface. After determining the design parameters and qualification realizing independence axiom, the database is designed in Microsoft Excel by utilizing the design qualification and parameters. The database is developed to be easily updated and enlarged. Seventeen robot arm alternatives are loaded to the database with the qualification and system range of each evaluation criteria. The proposed methodology, industrial robot arm selection methodology based on axiomatic design, creates the inference mechanism. The algorithm of methodology is coded in MATLAB. By using the interface, which is designed in Microsoft Excel, the target design specifications and design range of the designer or decision-maker are transferred into MATLAB. With the help of the decision support system’s inference mechanism, the robot arm alternatives in the database are analyzed by considering design specifications and design range. The information contents of robots are presented in MATLAB interface as a report. Most suitable robot arm selection is done according to the information axiom principle. By viewing the MATLAB report, information content of each criteria can be analyzed. In conclusion, in this study by using the decision support system based on the methodology, a large number of alternatives were evaluated rapidly and optimum robot arm was offered to a food industry firm for the packaging application. By interviewing the robot supplier, it was found that the chosen robot arm was consistent with the existing robot arm in the firm. The decision making process for industrial robot selection is based on a systematic and scientific basis with the use of the methodology. The proposed methodology based on axiomatic design provides following acquisitions: It ensures more consistent and effective decision making process compared to the intuitive decision making methods which are used for pre-selection process of robot arm in the current situation. The results of incorrect decision such as cost of time, cost of human resources and loss of the company’s image are thus prevented. According to information axiom principle, if the design range of a criterion can not be satisfied by the alternatives, then it would be rejected. It is the most important advantage that differentiates from the other multi attribute decision making (MADM) techniques. Added new alternatives and criteria does not cause too much increase in the number of iterations compared to other MADM techniques. Because, alternatives are evaluated separately in AD rather than other MADM techniques that evaluate alternatives relatively. With the developed decision support system having high data processing and analyzing capacity, large number of robot arm alternatives can be analyzed rapidly. In the future study; robot selection criteria could be expanded to include special purpose application by decomposing robot components in detail. Uniform distribution is assumed for the system range in the study. Instead of uniform distribution, different distribution types that simulate the real system could be used. Moreover, the validity of distribution types could be observed and analyzed. Thus, this improved decision support system could be integrated and implemented to a software as a selection module.