Fuzzy Engineering in Nuclear Research Applications 英语翻译

1、英语原文：Fuzzy Engineering in Nuclear Research ApplicationsD. Ruan: L. Van Den Durpel and P. DhondtFuel Research UnitNuclear Research Centre SCKaCENBoeretang 200, B-2400 Mol, BelgiumAbstract Security, maintenance, monitoring, diagnosis, and environment are all related to humans and their soci- ety, and

2、are the most important and dificult problems of nuclear engineering. These problems are so com- plicated that they can hardly be solved without a global approach. Therefore, fuzzy engineering may be one of the most powerful tools available to us. This paper reports on the initial activity of fuzzy e

3、ngineering at the Belgian Nuclear Research Centre. It illustrates two applications, namely, nuclear emer- gency decision aiding systems, and inspection of trans- mission lines of nuclear installations with preliminary results, then states a new RtYD project concerning a fuzzy model-based control of

4、a nuclear reactor giving more practical questions lo do with fuzzy engineering rather than presenting solutions. The paper finally emphasizes that fuzzy engineering is needed for the nuclear research world through the overwhelming re- sponse to the first international FLINS workshop on fuzz# logic a

5、nd intelligent technologies in nuclear sci- ence, held September 14-16, 1994 in Mol, Belgium. 1.Introduction FLINS is an acronym for Fuzzy Logic and Intelligent Technologies in Nuclear Science. It started as a new research project, launched in line with the objective of the Belgian Nuclear Research

6、Centre (SCKoCEN) to give young talented people the op portunity to carry out future-oriented research. Now FLINS has become the name of a new international research forum aiming to promote the theory and ap plications of fuzzy logic and other novel intelligent technologies in the domain of nuclear s

7、cience and engineering.Nowadays scientists have many models at their dis- posal to treat incomplete and complex information. Undoubtedly, fuzzy set theory is one of the most widely applied of these new models. At the University of Ghent, for almost twenty years, fuzzy set theory and its applications

8、 have been extensively researched in the Centre for Fuzziness and Uncertainty Modelling under the guidance of Prof. E.E. Kerre. FLINS builds upon a strong cooperation between SCKoCEN and the University of Ghent. Invented by Prof. L.A. Zadeh at the University of California at Berkeley, U.S.A., in 196

9、5, there have been more than fifteen thousand papers published on both theoretical and practical issues of fuzzy set theory. However, as Prof. H.J. Zimmermann (ELITE, Germany) points out 21, for almost twenty years, fuzzy methods were confined to universities in most parts of the world, with the pos

10、sible exception of some fuzzy logic control applications. This is particularly true in the field of nuclear science and technologies. The best-known work in this area is particularly topi- cal, as it deals with the Chernobyl accident, in which fuzzy human-reliability analyses in man-machine sys- tem

11、s has been considered 3,4. Subsequently, Prof. T. Terano (LIFE, Japan) ad- dressed an important application of fuzzy logic to some of todays most critical problems. Security, maintenance, monitoring, diagnosis, and environment are all related to humans and their society, and are the most important a

12、nd difficult problems of nuclear engineering. These problems are so complicated that they can hardly be solved without a global approach. Therefore, fuzzy engineering may be one of the most powerful tools available to us lo. The continuous increase and remarkable success of various applications of f

13、uzzy logic prompted the foundation of the research forum FLINS at the end of 1992. At present FLINSSCKoCEN consists of several engineers, mostly from nuclear science, and scientists who are currently working on various projects for research degrees, postdoctorate studies, or other research activitie

14、s. As the name suggests, it is not only limited to fuzzy logic, but includes other intelligent technologies, such as neural networks, knowledge- based expert systems, and complex problem-solving technologies. FLINS aims at solving many intricate problem pertaining to the nuclear environment by using

15、 modern technologies as additional tools. The need for the creation of a particular forum such as FLINS is evidenced by the overwhelming response to the call for papers for the First International FLINS workshop, held September 14-16, 1994 in Mol, Belgium. About seventy papers from Austria, Belgium,

16、 China, France, Germany, Japan, Korea, Malaysia, the Netherlands, Norway, Russia, the United Kingdom, and the United States of America were accepted; they cover applications of fuzzy logic and intelligent technologies in radiation protection, nuclear safety (human factors and reliability), safe- gua

17、rds, nuclear power plant control, decision making, and nuclear reactor control. This paper reports on the initial activity of fuzzy engineering at the Belgian Nuclear Research Centre with two concrete applications, namely, nuclear emergency decision-aiding system, and inspection of trans- mission li

18、nes of nuclear installations, then states a new R&D project concerning a fuzzy model-based control of a nuclear reactor with more practical questions to do with fuzzy engineering rather than presenting solutions. The paper finally emphasizes that fuzzy engineering is needed for the nuclear research

19、world. 2 Fuzzy Engineering at the Belgian Nuclear Research Centre Nowadays scientists have many models at their disposal to treat incomplete and complex information. Undoubtedly, fuzzy engineering becomes one of the most widely applied of these new models. The Belgian Nuclear Research Centre has rad

20、iation protection, fuel, reactor materials, waste and dismantling, waste and disposal research units, as well as the safeguards project 6. Fuzzy logic applications are however at present only in the fuel, and in the radi- ation protection research units. The topics undergoing are outlined as: fuzzy

21、decision making in nuclear science 151, inspection of transmission lines of nuclear installations 15, nuclear emergency decision-aiding systems 9,11, atmospheric-stability modelling for nu- clear emergency response systems 12,13,14, fuzzy systems in nuclear applications 7, and fuzzy model- based con

22、trol of a nuclear reactor 16,17. It is expected that more and more research concerning fuzzy logic applications will soon be examined by FLINS. Among the many industrial supporter of FLINS are Belgoprocess, ECN Petten, FBFC International,Belgonucleaire, NIRAS/ONDRAF (National Agency for Radioactive

23、Waste and Fuel), etc. The OMRON Electronics Europe B.V. has sponsored the FLINS research in the area of fuzzy logic cont,rol as applied to the operation and control of reactors. It providesfuzzy logic software to FLINS, thus launching the first FLINS contact with the industrial world. Fur- ther coop

24、eration is in progress.3 Application 1: Nuclear Emergency Decision Aiding Systems The potential contribution of fuzzy set theory techniques towards the issues of safety criteria and regulatory decision making is set to become significant. As reported recently 2, the main advantage of using fuzzy set

25、 theory has been in overcoming the difficulties of decision making in a fuzzy situation represented by ill-defined terms. The inherent imprecision of such terms makes crisp ranking very difficult and application of statistical decision theory doubtful. The situation can be handled by the analyst by

26、ranking these quantities verbal, which is the normal behaviour of human beings to account for inherent imprecision. Verbal ranking is then represented by fuzzy sets. The final ranking of alternatives from best to worst can be obtained using fuzzy operations. The development of emergency response sys

27、tems, being able to support decision making in the event of a nuclear accident, is a relatively new area of R & D work, only a small number of systems are in opera- tion at present and, in general, they can only respond to a limited number of questions posed by a decision maker.During the early stag

28、es of an accidental release of radioactive material to the atmosphere. the immediate aims of the off-site emergency management scheme are twofold: firstly, to determine the extent of any contamination occurring close to the site (i.e. within a few km) for purposes of protecting the local pub lic; se

29、condly, to provide early estimates of the source term, and hence permit consequences farther afield to be assessed. The source term module has the task of supplying the system with all information concern- ing the release of radionuclides. However, the exact reconstruction of the source term is up t

30、o now an unresolved problem. Several mathematical methods areunder development for atmospheric dispersion models. Some of them already give results with acceptable ac- curacy in special but not too complex situations. A direct application of the methods used for probabilities assessment of accident

31、consequence is not suitable, since time consuming calculations are not feasible during an emergency, and the nature of the uncertainties in a large fraction of the input information and model predictions differs fundamentally. Therefore, the work ll seeks another approach based on fuzzy sets and dec

32、ision theory to treat uncertain information (incomplete and inexact). For example, an ambiguous problem of the classification with linguistic values such as very good, good, middle, bad, and very bad, according to the ratio of predictions and observation data, can be treated with membership function

33、s. Also a simple fuzzy algorithm is designed for the system to provide a reasonable solution for practical users, which could propose different solutions within a short time period. The basic idea of the system is built on the following simple mathematical model: (1)In the above equation, C is the c

34、oncentration of radioactive material, which is dependent of an unknown variable of source term Q, and another unknown variable of wind direction D (x and y are coordinates depending on the wind direction D, and the other physical parameters such as u, and , are given in this study case). In a real s

35、ituation, one can on one hand obtain by means of observation (0) measured concentration data denoted as for certain points , and on the other hand one can calculate prediction (P) denoted as , by the model if the source term Q and the wind direction D are given for the same points as used in the obs

36、ervation. For each point k, one defines a fuzzy set as (2)where the indexes i, j correspond 1.0 the grids of the wind direction and the source term, . The fuzzy set can be interpreted as in a certain point K, the prediction P is the closest to the observation 0. For a11 points, using the decision-ma

37、king theory 1,19,20 one obtains the fuzzy decision set D as: (3)The best source term qo and the best wind direction do can be obtained by using the defiizzification techniques. Practically, one can set the finite grids for wind direction since the range of it is more-or-less known. However there is

38、no idea about the source term. To set a large range of the source yields the computer timeconsuming problem for this simulation. For that reason, ratios of $ (prediction over observation) is computed, and the situation for which all these ratios (for all measurement points) are as close to 1 as poss

39、ible is being searched with respect to the different membership functions. In this way, one gets an approximate value of the source, therefore a more-or-less correct range of it. Several examples have been tested and the method proposed here seems acceptable from the point of view of the practice .4

40、 Application 2: Inspection of Transmission Lines of Nuclear Installations The safety of a nuclear power plant (XPP) and of research reactors has to be dealt in increasingly greater depth. During the construction phase of the NPP, several safety related solutions can be studied and idealljimplemented

41、. Older installations, at certain nuclear facilities, require extra safety evaluation. One of the aspects to deal with is the safety of the several transmission lines in a nuclear installation, for instance the safety of control, safety against fire, etc. The Belgian Nuclear Research Centre has to d

42、eal with existing installations, and especially the fire risk anal3sis of electrical power cable trays. In this particular case several questions arise: “To what extent can those cabinets propagate fire?”, “To what extent can those cables be the source of fire due to their previous service life?”, a

43、nd “What are the mechanisms observed?”. Clearly a strict answer to those questions cannot be obtained since it needs a lot of engineering judgement, and the practical inspection of those cable trays introduces a lot of subjective inputs. Obviously, a survey of all transmission lines against several

44、hazards, in the nuclear installation, is an enormous task as the following separate steps have to be taken: (1) Each transmission line needs to be checkedby an inspector, who has to report the observations that are made. Above all, there has also to be accordance between several inspectors themselve

45、s. (2) These observations result in a database of all lines inspected, including the remarks made for each line, in which the database forms the start of the real safety evaluation. (3) Finally a decision on the safety of each line, and moreover, of the global installation safety needs to be made. T

46、he subjectivity as noted, and the different questions involved, justify the use of fuzzy techniques. The methodology for the safety evaluation of transmission lines for a deeper case-study is based on fuzzy set theory. Its original application was in the field of medical diagnosis, and later on it w

47、as suggested to be applied in the field of safety analysis of electrical installations. Fuzzy relations and composional operations of fuzzy relations have been used in this fuzzy approach. Fuzzy relations between the symptoms of failure observed in a transmission line and safety levels are to be est

48、imated. This will be achieved from the analysis of investigated (or published) technical information, and from a survey of the views of a group of experts. These relationships are incorporated in the computer program FIRECAB (Fire Evaluation of Cables), developed to calculate the safety-value of eac

49、h line. This approach illustrates that the fuzzy technique implements a faster methodology with accordingly lower costs involved. The fuzzy approach involves the study of two fuzzy relations, and results in a reduction in a degree of subjectivity of the safety evaluation process. The basic idea of t

50、his approach is outlined in brief. Let x be a set of symptoms of a given transmission line, y a set of failure types, and z a set of safety levels. and are denoted as two (fuzzy) relations between symptoms and failure types, and between failure types and safety levels, respectively. Therefore a new

51、(fuzzy) relation for a certain trans- mission line between symptoms and safety levels denoted can be obtained as: (4)where * is a compositional operator of fuzzy relations. If one gives an inspector a checklist for each trans- mission line with the possible observable symptoms x, than one gets easil

52、y a safety level z for each transmission line. This relationship also holds for the classical approach. However, naturally the fuzzy approach seems to have an important advantage of reducing the degree of subjectivity of the human observer. Easily, one makes a checklist with symptoms x, but in thecl

53、assical approach this results directly in a list with too much detail, thereby loosing the final goal namely a safety evaluation in an economic manner. However, the fuzzy approach takes a given set of symptoms x related to a given set of failures y. Due to the fact that this relation holds for every

54、 installation, this fuzzy relation has to be made only once. Additions or changes can be modified easily. Typical elements of the set of symptoms are: no obvious isolation damage, small isolation damages, many small isolation damages, excessive isolation damages, no rust, small rust, excessive rust,

55、 no resistance change, small resistance change, and excessive resistance change. Typical elements of the set of failure types are: no loss of isolation, small isolation leaks, isolation leaks, no transmission signal change, small transmission signal changes, and high transmission signal change. As a

56、 consequence, typical elements of the set of safety levels are: unsafe against fire, questionablely safe against fire, and safe against fire. The set of symptoms needs to be described on a very comprehensive manner for all possible observable symptoms, such that an inspector can point out what is im

57、portant for each line. A good preparation of the relation helps inspectors, and for the final safety result. Clearly, this relation is common for each installation, if these x- and y-set are well chosen. The relation is however typical for each installation and known safety study. Ifone is consideri

58、ng the safety against fire, the set of safety-levels z describe the safety against fire in the installation. Therefore, this relation has to be established for each installation itself. Moreover, this relation can also describe interference between several lines, that previously made surveys installation-dependent. There is of course a quite important economical aspect involved in the relation (4). With this relation, one performs the enormous task of getting the respectiv