Approach to Knowledge Based Man-Machine Communication for BWR Start-Up Guidance

  • Published on
    12-Feb-2017

  • View
    212

  • Download
    0

Embed Size (px)

Transcript

  • This article was downloaded by: [University of Nebraska, Lincoln]On: 09 October 2014, At: 15:54Publisher: Taylor & FrancisInforma Ltd Registered in England and Wales Registered Number: 1072954 Registeredoffice: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

    Journal of Nuclear Science andTechnologyPublication details, including instructions for authors andsubscription information:http://www.tandfonline.com/loi/tnst20

    Approach to Knowledge Based Man-Machine Communication for BWR Start-Up GuidanceYasuo NISHIZAWA a , Hiroshi MOTODA a , Naoyuki YAMADA a & YutakaWADA aa Energy Research Laboratory , Hitachi Ltd. , Moriyama-cho, Hitachi-shi , 316Published online: 15 Mar 2012.

    To cite this article: Yasuo NISHIZAWA , Hiroshi MOTODA , Naoyuki YAMADA & Yutaka WADA (1983)Approach to Knowledge Based Man-Machine Communication for BWR Start-Up Guidance, Journal ofNuclear Science and Technology, 20:10, 877-879, DOI: 10.1080/18811248.1983.9733482

    To link to this article: http://dx.doi.org/10.1080/18811248.1983.9733482

    PLEASE SCROLL DOWN FOR ARTICLE

    Taylor & Francis makes every effort to ensure the accuracy of all the information (theContent) contained in the publications on our platform. However, Taylor & Francis,our agents, and our licensors make no representations or warranties whatsoever as tothe accuracy, completeness, or suitability for any purpose of the Content. Any opinionsand views expressed in this publication are the opinions and views of the authors,and are not the views of or endorsed by Taylor & Francis. The accuracy of the Contentshould not be relied upon and should be independently verified with primary sourcesof information. Taylor and Francis shall not be liable for any losses, actions, claims,proceedings, demands, costs, expenses, damages, and other liabilities whatsoever orhowsoever caused arising directly or indirectly in connection with, in relation to or arisingout of the use of the Content.

    This article may be used for research, teaching, and private study purposes. Anysubstantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

    http://www.tandfonline.com/loi/tnst20http://www.tandfonline.com/action/showCitFormats?doi=10.1080/18811248.1983.9733482http://dx.doi.org/10.1080/18811248.1983.9733482http://www.tandfonline.com/page/terms-and-conditionshttp://www.tandfonline.com/page/terms-and-conditions

  • Journal of NUCLEAR SCIENCE and TECHNOLOGY, 20[10], pp. 877-879 (October 1983) 877

    SHORT NOTE Approach to Knowledge Based Man-Machime Communication for BWR Start-up Guidance

    Yasuo NISHIZAWA, Hiroshi MOTODA, Naoyuki YAMADA and Yutaka WADA

    E n e r g y Research Laboratory, Hitachi Ltd.*

    Received April 13, 1983 Revised July 25, 1983

    KEYWORDS: BWR type reactors, start-up, core management, infer- ence engine, predicate calculus, man-machine communication, CAD

    Computer has so far been used to solve a problem by numerical computation. Many of the engineering problems, however, can be solved more efficiently by the so-called expert knowledge pertinent to the problem. The knowledge used there is sometimes heuristic and not handled by numerical com- putation. Knowledge engineering (K.E.) ()() provides a method to incorporate such knowl- edge in problem solving.

    Core management is a field in which much experience useful for decision making has been accumulated. This note introduces a computer program that receives such ex- perience as pieces of knowledge and makes inference for problem solving with an ex- ample applied to a man-machine communica- tion for BWR start-up guidance.

    1. Summary of Program Experts knowledge comprises not only

    individual fact and relations among them but also more general or abstract concept. To express such abstract knowledge, the program developed here adopts first order predicate calculus that allows to use variables for in- definite object. In the inference process, variables are compared with individual object and bound to them by unification. Using predicate calculus, individual fact is expressed as a proposition, and a relation among the facts is represented as a production rule that

    is sometimes called if-then rule. Two methods of inference are prepared,

    that are, forward and backward inferences. Forward inference finds new facts that are concluded by combining rules and previously proved facts. Backward inference proves the solution of a problem by tracing the logics in a backward direction starting from an assumed conclusion. The assumed conclusion can contain variables. These methods can be used in combination.

    Experts knowledge further includes the knowledge about when and how to use knowledge. Such knowledge is called meta- knowledge. Meta-knowledge can also be stored in the knowledge base and is used for inference control such as to proceed numerical computation on the way of inference, or to generate a question to users before looking into the knowledge base. Hierachical structure is employed for context of knowledge. Knowl- edge in the higher level context can be referred from the lower level context but not for reverse. The program has also the func- tion to explain the process of inference, that is, the way through which it has reached a conclusion. The language used is UTILISP and the computer M-200H.

    2. Application to Start-up Guidance The program has been applied to a man-

    machine communication for BWR start-up guidance. Start-up scheduling is a problem to find a way to bring reactor power to the rated value without violating any of the operating constraint^'^)'^). In the start-up process, the most difficult and troublesome problem is estimation of Xe transient state caused by reactor power change. The Xe behavior brings about a fairly complicated effect on core state. Engineers however can estimate the Xe change by reference to the power history and their experience. Although such estimation is not so accurate as that by numerical simulation, it is useful to decide the reactor operating strategy.

    Utilization of such experts knowledge makes it possible to give a guidance for BWR

    * Moriyama-cho, Hitachi-shi 316.

    - 81

    Dow

    nloa

    ded

    by [

    Uni

    vers

    ity o

    f N

    ebra

    ska,

    Lin

    coln

    ] at

    15:

    54 0

    9 O

    ctob

    er 2

    014

  • 878

    Pattern of Xenon Change

    Monotonically Increasing Monotonically Decreasing Equilibrium Overshooting Undershooting

    ~- -

    SlioRT NOTE (Y. Nishizawa et al.) J. Nucl. Sci. Technol.,

    Degree of xenon Change

    Remarkably Increasing Gradually Increasing Equilibrium Gradually Decreasing

    c

    1 I

    start-up through man-machine communication in which the program analyses what is neces- sary to solve a problem by itself. In the following, described is an attempt of such a knowledge based guidance to the decision making of reactor operation.

    The man-machine communication system consists of a knowledge base, an inference engine and a communication equipment. The rules stored in the knowledge base are those of operating constraints, corrective actions for constraint violations, Xe transient state and meta-knowledge for inference control.

    Structure of the knowledge base and the flow of information are shown in Fig. 1. The amount of rules is 55, out of which, 10 is for constraints, 20 for corrective actions,

    Power History

    Power Change o Degree o Elapsed Time

    Operating Status

    History

    -

    20 for Xe transient and 5 for inference con- trolling meta-knowledge.

    The Xe transient state is characterized by the two factors :

    (1) Xenon concentration changing pattern (2) Degree of the change.

    The former is determined by reactor power history and the latter by the degree of the power change that brought about the Xe change together with the elapsed time after the power change. The process of the Xe transient state estimation is shown in Fig. 2 with an example of the rules.

    The characteristic of this system is that it adopts backward inference in combination with meta-knowledge to generate questions to the users about what computer needs to

    Xnowledge Base

    Xnowledge of Xnowledge of Constraints Corrective A c t i o n

    Question Driving Functions Questions

    Fig. 1 Structure of knowledge base

    I

    Example of Rule5

    Action

    riitermediate

    (If [and (Xenon-transient Overshoot) (Power-change Great I

    Fig. 2 Knowledge construction about Xe (Elapsed-trme Short 1 ) (Then (Xenon Increasing Remarkab1y)I)l

    - 82 -

    Dow

    nloa

    ded

    by [

    Uni

    vers

    ity o

    f N

    ebra

    ska,

    Lin

    coln

    ] at

    15:

    54 0

    9 O

    ctob

    er 2

    014

  • Vol. 20, No. 10 (Oct. 1983) SHORT NOTE (Y. Nishizawa e t a l . ) 879

    Communication

    t Control Rod or Flow? J C R 1 Pz;;-up, Down or Hold? How Many Power-up Loops?

    Xenon State?

    Set Degree of Power Chang

    I Inference Process

    Final Power-up Increasing Remarkably

    Power Change 3 3 0 1 % )

    Set Time aftez Power Change ~1 Ihr)

    t I I

    orrective Actio

    own by Xenon.

    Fig. 3 Example of man-machine communication and inference process (in case of rod block line violation)

    know. The inference starts by assuming a conclusion about corrective action if a con- straint is violated. It first finds conditions required to prove the assumed conclusion by tracing the logics in the rules and tries to prove each of them. Some of the conditions can be asked to the user for the truth by the meta-knowledge. When the answer is not known, it then looks for a chain to prove the condition as a subgoal.

    Figure 3 shows an example of man- machine communication with the inference path to reach the final goal. This is a case to decide the corrective action when the reactor power exceeded its constraint Rod Block Line. By answering to the questions raised by the program, it has been concluded that (1) the Xe transient state is in the initial stage of overshooting introduced by a great power change and the Xe is increasing re- markably and, thus, (2) the corrective action is simply to wait for the power to decrease.

    3. Discussions The study indicates that some complicated

    problems as those encountered in a Xe trans-

    ient state can be solved by using the knowl- edge about reactor operation, expressed in simplified logical statements.

    For practical use, more detail and abundant knowledge should be prepared and the guid- ance should be more quantitative. Know- ledge to meet such a requirement includes numerical simulation. It is, therefore, neces- sary to couple symbolic manipulation to numeric computation.

    The approach introduced here will be a useful technique for intelligent problem solv- ing. Other tasks of core management such as refueling scheduling and long term control rod programming are also handled by this technique.

    -REFERENCES-

    (1) FEIGENBAUM, E. A. : Stanford Univ., S T A N -

    (2) OHSUGA, S. : Joho Shori, (in Japanese), 23

    (3) BESSHO, Y., et a l . : J . Nucl, Sci. Technol.,

    (4) NIsHizAwA, Y., et 01.: [Nucl. Technol 58,

    CS-77-621, (1977).

    (JO], 967-974 (1982).

    lE(9], 697-704 (1981).

    9-22 (1982).

    - 83 -

    Dow

    nloa

    ded

    by [

    Uni

    vers

    ity o

    f N

    ebra

    ska,

    Lin

    coln

    ] at

    15:

    54 0

    9 O

    ctob

    er 2

    014

Recommended

View more >