İnsan hatalarını sayısallaştırmak için bir model
A model for quantifying of human error
- Tez No: 21812
- Danışmanlar: PROF. DR. AHMET F. ÖZOK
- Tez Türü: Doktora
- Konular: Endüstri ve Endüstri Mühendisliği, Industrial and Industrial Engineering
- Anahtar Kelimeler: Belirtilmemiş.
- Yıl: 1992
- Dil: Türkçe
- Üniversite: İstanbul Teknik Üniversitesi
- Enstitü: Fen Bilimleri Enstitüsü
- Ana Bilim Dalı: Belirtilmemiş.
- Bilim Dalı: Belirtilmemiş.
- Sayfa Sayısı: 150
Özet
ÖZET Bu çalışmanın amacı, İ-M sistemlerinde insan hatalarına yönelik literatürü incelemek ve insan hatalarının sayısallaştırılmasına yönelik bir model geliştirmektir. ilk bölümde insan hatalarına yönelik çalışmalara değinilmektedir. ikinci bölüm insan çalışmasının gerçekleştiği ve insan hatalarının meydana geldiği İ-M sistemini tanımaya ayrılmaktadır. Üçüncü bölümün konusunu insan davranışları ve bu davranışların İ-M sistemindeki sonucu olan insan performansı oluşturmaktadır. İnsan davranışının esasları, performansın ortaya çıkışı ve değerlendirilmesi bu bölümde incelenmektedir. İnsan performansının bir parçası olan insan hatalarını inceleyen nedensel yaklaşım dördüncü bölümde ele alınmaktadır. İnsan hatası nedir, nasıl ortaya çıkar, sebepleri-sonuçları nelerdir, ve nasıl sınıflanır, sorularına yönelik literatür bu bölümde araştırılmaktadır. Beşinci bölüm insan hatalarını sayısal olarak inceleyen insan güvenilirliği konusuna ayrılmıştır. İnsan güvenilirliği kavramı, konunun esasları tanıtılarak, insan güvenilirliği analizi için geliştirilmiş yöntemler bu bölümde tartışılmaktadır. Altıncı bölüm insan güvenilirliği alanında önemli yer tutan veri problemi ile ilgilidir. Güvenilirlik alanında veri elde edilebilen kaynaklar ile veri tabanı oluşturma girişimleri tanıtılmaktadır. Yedinci bölümde, bu çalışmadaki kullanımına ışık tutmak üzere, bulanık küme kavramı, bazı özellikleri ve dilsel değişken kavramı açıklanmakta ve güvenilirlik alanındaki bulanık küme uygulamaları belirlenmektedir. insan hatalarını sayısal olarak incelemek üzere geliştirilen model sekizinci bölümde yer almaktadır. Modelin amacı, kapsamı ve dayandığı gerekçeler tespit edilmektedir. Üç aşamadan oluşan model açıklanarak.fiktif bir örnek üzerinde uygulama yapılmaktadır. Model matematik araç olarak bulanık kümeler kuramını kullanmaktadır. Sayısal değerlendirme, geliştirilen bilgisayar programı ile gerçekleştirilmektedir. Son kısım çalışmanın sonuçlarını, bu alanda mevcut problemleri ve potansiyel çalışma konularına dair önerileri içermektedir.
Özet (Çeviri)
SUMMARY A MODEL FOR QUANTIFYING OF HUMAN ERROR The word machine is used in a generic sense to imply any piece of device or facility with some useful purpose. The pencil we write with, the woods and irons we play golf with are, in this sense, just as much machines as the lathe in a factory or the plant itself. The same applies for aircrafts, nuclear reactors, space ships, etc., up in the spectrum of technical complexities. Since machines are designed and built for human use, humans are involved in every man-machine system. Thus, man is not a being independent of the system, an entity that only operates or controls a machine aloof from the rest of the system“. Rather, he is an active part of the system, at least even as a monitor, and as such he contributes those capabilities which are uniquely his. The entire man-machine system operates in an environment of temperature, humidity, illumination, noise, vibration, acceleration, circadian rhythm, etc. These environmental factors, to one degree or another, affect the performance of the system components, man or machine. Single man-machine systems can be combined into a larger system. In such a system, the several man-machine subsystems may interact with each other, the functions of one subsystem depending upon the efficient functioning of another. Alternatively, a system may consist of one machine and several humans, for example, as an aircraft navigated by two pilots (personnel redundancy); or one human may form the link between several machines. For a man-machine system to function well, its purpose should be clearly defined and ”understood“, not only by the operator, but by the various machine components involved. This implies that individual components are well matched to each other by carefully balancing the functions of man and machine, that their contributions to overall performance are clearly defined. And for this purpose, the various capabilities and limitations of man should be carefully taken into consideration. The current trend toward automation is altering the nature of human involvement in the man-machine systems. For example, some of the functions formerly carried out exclusively by hand or with hand tools have become mechanized, thereby shifting the human contribution more towards the operation and maintenance of machines. In automatic operations, the consequences of a human, error., an incorrect reading of a display, a minor lapse of memory, a failure to notice a warning signal, a skipping of a small procedural detail, or a misunderstanding of instructions., are usually much greater than in a conventional operation. Because of the high capital investment required for an automatic installation, downtime can be very expensive. Because of the faster pace of the machines, time spent in producing unusable products can be very costly, from the excessive waste of ximaterial. Because of the large scale, damage or injury can range widely, even precipitating the loss of many lives. Every man-machine system contains certain functions that must be performed by man. Even the so-called fully automated systems need human interventions in monitoring and maintaining. If the variability in human performance is recognized as inevitable, then it is easy to understand that when humans are involved, errors will be made, regardless of the level of training, experience, or skill. As the man-machine systems are required to become more reliable, human influence becomes more and more important. The effort that is sometimes spent in designing ultrareliable equipment is often negated by human error. Two major approaches can be taken to characterizing human error: probabilistic and causal. Human reliability analysis is concerned with the qualitative and quantitative analysis of human errors and its subsequent reduction. This approah is typically pursued by those who are interested in the human reliability aspects of risk analysis. In contrast to reliability analysis the causal approach to characterizing human error is based on the premise that errors are seldom random and, in fact, can be traced to causes and contributing factors which, once isolated, can perhaps be eliminated or at least amelioriated and subsequently modifying system designs and training programs. Human errors are said to occur when the performance is outside of predefined tolerance limits. One should not assume that all errors are alike in terms of their causes or their effects on the systems or their mechanisms. Errors may be classified in various ways. 1. In terms of what caused the error 2. In terms of what the error consequences are 3. In terms of the stage of systems development in which the errors occurred. 4. Behavior-oriented. Frequently, successful man-machine system performance depends on the successful performance of each and every component, man or machine. The human operator is inherently less stable than the machine. His performance is influenced by many more conditions: his physiological condition, fatigue, amount of learning, and incentives, the work environment (e.g., temperature, humidity), among numerous other factors. As above cited not all human errors, however, are the same in terms of their cause, mechanism, and consequences. Another measure often used to indicate the effectiveness of the human component in a man-machine system is the frequency of the system or machine failures caused by various human errors. XllAlthough human errors generally affect the speed and accuracy of mission accomplishment, they do not necessarily affect the condition of the machines, dont do they necessarily cause the failure of the system mission. On the other hand, slightly different from the trivial errors of omission or commission are the human initiated failures, which typically prevent the system from accomplishing its mission, if the nature of the failure is serious enough. Human errors that do not result in system failure are often reversible, whereas errors causing human initiated failures cannot be reversed, because failed machines usually cannot restore themselves. The role played by human errors in overall system unreliability cannot be taken lightly. The high incidence of human error in the operation of man- machine systems is well documented by many investigators. Reliability is generally defined as ”the probability that an item will operate adequately for a specified period of time in its intended application". Although the main focus of quantitative reliability has been on hardware portion of man-machine systems applied to the performance of functional objects, the general concept of reliability can be extended naturally and easily also to human activity. The purpose of human reliability analysis is to analyze the man-machine system and predict the potential for human error. Human error probabilities or human reliabilities are useful not only in estimating man-machine system reliability, but also in other Human Engineering activities such as allocating functions between man and machine (based on reliabilities with and without the human), quantifying the error likelihood and consequences of human engineered equipment, and estimating the success of personnel training programs. Reliability is the antithesis of error likelihood. Human reliability is then defined as the probability that his performance will be error-free for a specified duration. Human performance concerns both time-discrete and time-continuous task. Data on reliability are expressed in terms of failures per event for the first and of failures per unit time for the second. By combining these measures in assorted ways, the analyst can calculate a total reliability figure for human performance in some specific task or job. The best way of obtaining an human error probability in a task is to directly observe the errors committed in the task. One major problem with this approach is that it is only applicable to those installations and tasks for which empirical data are available. Further, the trend toward large, centralized systems does not foster the availability of direct empirical probability data on human involvement in complex, rare event. Numerous techniques have been proposed for predicting the performance of the human component of the man-machine system. The techniques take one of two general approaches: analytical or simulation. Bince each technique is specifically applicable to some particular types of tasks and systems, there is no general purpose methodology. XlllThe basic theoretical underpinnings of human reliability derive from conventional hardware reliability theory. Individual component reliabilities are combined according to the system configuration and its operation sequences. Human task sequences can be modeled in a similar manner. Human reliability figures for component task are combined mathematically to synthesize the error probability for the entire task sequence. Thus, the general approach is to divide human behavior in a system into small behavioral units, find data for these subdivisions and then to estimate the error probabilities for the task. Until one have the error frequency data on the whole task in question, this type of decomposition and recomposition is often necessary. The general procedure common to most of the predictive techniques are as follows: 1- As the bases for prediction, the behavioral units (e.g. subtasks, tasks, functions) are determined from the function/task analysis of the system. 2- Important parameters for the behavioral units are determined, such as operator function, display/control organization, availability of feedback information, required response accuracy, etc. 3- Input data are assigned to the behavioral units. 4- The measures for the combined behavioral units are derived either through probabilistic analysis or simulation of the behavioral operations. 5- The overall system reliability is predicted by combining the final predicted behavioral-reliability and the machine reliability. Monte-Carlo modeling of man-machine systems can include human and equipment variables. The purpose of the technique is to indicate where the system may overload or underload its personnel and to determine whether the average operator can complete all required tasks on time. Although the combinative techniques are very similar for the hardware and human reliability analysis, estimating the human task element reliabilities are not straightforward. The human is inherently less stable than the machine; he is influenced by various behavioral factors such as physiological condition, fatigue, the work environment, amount of learning, and incentives, etc. One of the complications in modeling tasks as sequences of behaviors is the dependence among the task elements. Two events are independent if the conditional probability of one event is the same whether the other event has occurred or not. Dependence can occur between people working on a task and also between an individual as several related tasks are performed. Failure to take dependence into account can underestimate task error probabilities. The most serious problem in the rather young field of human reliability is the lack of data. All human reliability analysis techniques require human error data at some point in their analysis-prediction processes. XIVThe basic theoretical underpinnings of human reliability derive from conventional hardware reliability theory. Individual component reliabilities are combined according to the system configuration and its operation sequences. Human task sequences can be modeled in a similar manner. Human reliability figures for component task are combined mathematically to synthesize the error probability for the entire task sequence. Thus, the general approach is to divide human behavior in a system into small behavioral units, find data for these subdivisions and then to estimate the error probabilities for the task. Until one have the error frequency data on the whole task in question, this type of decomposition and recomposition is often necessary. The general procedure common to most of the predictive techniques are as follows: 1- As the bases for prediction, the behavioral units (e.g. subtasks, tasks, functions) are determined from the function/task analysis of the system. 2- Important parameters for the behavioral units are determined, such as operator function, display/control organization, availability of feedback information, required response accuracy, etc. 3- Input data are assigned to the behavioral units. 4- The measures for the combined behavioral units are derived either through probabilistic analysis or simulation of the behavioral operations. 5- The overall system reliability is predicted by combining the final predicted behavioral-reliability and the machine reliability. Monte-Carlo modeling of man-machine systems can include human and equipment variables. The purpose of the technique is to indicate where the system may overload or underload its personnel and to determine whether the average operator can complete all required tasks on time. Although the combinative techniques are very similar for the hardware and human reliability analysis, estimating the human task element reliabilities are not straightforward. The human is inherently less stable than the machine; he is influenced by various behavioral factors such as physiological condition, fatigue, the work environment, amount of learning, and incentives, etc. One of the complications in modeling tasks as sequences of behaviors is the dependence among the task elements. Two events are independent if the conditional probability of one event is the same whether the other event has occurred or not. Dependence can occur between people working on a task and also between an individual as several related tasks are performed. Failure to take dependence into account can underestimate task error probabilities. The most serious problem in the rather young field of human reliability is the lack of data. All human reliability analysis techniques require human error data at some point in their analysis-prediction processes. XIV
Benzer Tezler
- İnsan kinematik verilerinin kalman filtresi kullanarak iyileştirilmesi
Improving human kinematic data using kalman filter
HÜSEYİN ESKİ
Doktora
Türkçe
2020
Bilgisayar Mühendisliği Bilimleri-Bilgisayar ve KontrolSakarya ÜniversitesiBilgisayar ve Bilişim Mühendisliği Ana Bilim Dalı
PROF. DR. ÜMİT KOCABIÇAK
- Positional uncertainty analysis using data uncertainy engine-A case study on agricultural land parcels
Veri belirsizlik motoru kullanılarak konumsal belirsizlik analizi tarımsal arazi parselleri örneği
İLKSEN URGANCI
Yüksek Lisans
İngilizce
2009
Jeodezi ve FotogrametriOrta Doğu Teknik ÜniversitesiJeodezi ve Coğrafi Bilgi Teknolojileri Ana Bilim Dalı
DOÇ. DR. ZUHAL AKYÜREK
- Digital topografik haritalar ile digital ortofoto haritaların doğruluk, maliyet ve üretim zamanı açısından karşılaştırılması
Comparison of digital orthophoto maps with digital topographic maps
OKTAY EKER
- Robot kolu ve lazer profil algılayıcı ile otonom 3B nesne modelleme
Autonomous 3-D object modeling with robot manipulator and laser profile sensor
KAYA TURGUT
Yüksek Lisans
Türkçe
2016
Elektrik ve Elektronik MühendisliğiEskişehir Osmangazi ÜniversitesiElektrik-Elektronik Mühendisliği Ana Bilim Dalı
YRD. DOÇ. DR. HELİN DUTAĞACI
- Yapı elemanı tasarım süreci için bulanık çok ölçütlü bir değerlendirme modeli önerisi
A proposed fuzzy multi-criteria evaluation method for the design process of building elements
SEHER GÜZELÇOBAN MAYUK
Doktora
Türkçe
2015
MimarlıkMimar Sinan Güzel Sanatlar ÜniversitesiMimarlık Ana Bilim Dalı
YRD. DOÇ. BERRİN ŞAHİN DİRİ