The computer technology of expert rules constraction for the technical objects diagnosis systems
Description of the methods of organizing expert rules for the diagnosis of electrical machines using and without expert information. Correction of rules according to the results of causal observations. Using the intellectual potential of an expert.
| Рубрика | Программирование, компьютеры и кибернетика |
| Вид | статья |
| Язык | английский |
| Дата добавления | 18.11.2018 |
| Размер файла | 60,8 K |
Отправить свою хорошую работу в базу знаний просто. Используйте форму, расположенную ниже
Студенты, аспиранты, молодые ученые, использующие базу знаний в своей учебе и работе, будут вам очень благодарны.
Размещено на http://www.allbest.ru/
Institute of electrodynemics of NASU, Kyiv, Ukraine
The computer technology of expert rules constraction for the technical objects diagnosis systems
Shalomygin M.V.
Annotation
The expert rules organizing techniques by expert's information and without one for electrical motor's diagnosis system moreover these rules correcting technique by reason-consequences monitoring have discussed here.
Анотація
Подано методики організації експертних правил (ЕП) системи диагностики електричних машин (ЕМ) с застосуванням експертної інформації та без неї (автономно), а також методика коректування цих правил за результатами причино-слідчих спостережень.
Аннотация
Представлены методики организации экспертных правил (ЭП) системы диагностики электрических машин (ЭМ) с использованием экспертной информации и без неё (автономно), а также методика корректировки этих правил по результатам причинно-следственных наблюдений.
Ключевые слова: электрическая машина, экспертные правила, эксперт, диагностика, распределение.
The presented techniques of the expert rules organizing give a possibility to intellectualizing the technical object's diagnosis system not only by expert's intellect potential using. It is possible by techniques of the automatically rules correcting and creating new on the cause-outcome monitoring results. These techniques apply are especially important for the individual and small quantity-produced object's diagnosis, i.e. for objects without statistical dependencies between damages and checking parameters.
This diagnosis system is based on expert system procedures. The main tasks of the diagnosis system are the object's real condition's identification; the defects and damage's diagnosis on early stage; the forecasting of the future object's technical conditions; the planing of the operative actions and the maintenance planing.
To avoid of some limits on the diagnosis reliability estimation and to make open system to the list of possible damages is adding the not identificated damage. This damage is corresponds with all other unannounced damages.
Assume that the checking parameters are the diagnosis symptoms characterizing values. In a general case expert is building the knowledge base rule as if one of the checking indexes is large or less then established value. So claiming that there is one of possible damage. The entering data (established operating index value), that points to damage, has diapason from Т1 to Т2. In this case T1 0, and T1 T2 . The function of probably distribution P() of damage from correspondence operating index has visibility:
, when 0 (1)
P() = 0, when < 0
where and -factors of Veibul distribution, that estimated dependency the data distributions disperse. Those data are getting from experts about index's values characterized limits of correspondence interval. This kind of distribution is selected on base of the more than 100 Ukraine plant's high voltage AC motor's and generator's damages statistic, during 30 years.
The estimations of and factors would get from equations determinate the expectation M() and disperse D() of the correspondence operating indexes. We will assume that expectation is the maximum (minimum) long permissible value of concrete index, disperse calculating on denoted deviation.
Where Г(.) -- Gamma-function.
These nonlinear equations deciding by and are making with restriction that and could be positive and real figures.
In practice this technique realization we propose by modified equation (1) in visible
(2)
where -measured value of operating index; -the long permissible value of operating index. In this way for cases when e.g., a operating index is temperature of the motor's winding and as event -- slot insulator break-through
where Pe -the probability of damage that denoted by expert for a index value . During determine the damage on the real operating index value we take account as , so it's equal 1. For cases when e.g., for noted damage the index is wending resist
During determine the damage on the real operating index value we take account as we take account as so it's equal 0.
For the pairwise independence of events (damages) it is demands that expert's information and independently created rules could not duplicating. On this reason we did make an operative knowledge base and stationary one. The operative knowledge base is creating in next direction. For each damage Ai (i=1,2...n) and parameter Bj (j=1,2...k) by looking for the stationary knowledge base we can find the rules with same Ai and Bj. Assume we find m*i*j rules, e.g., m experts (system is an expert too) informed us about probabilities of damages Ai depending on parameter Bj value. For these rules we can receive parameters of the probability's distributions. Since for these rules describe we use a Veibul distribution (2) we have assume that index i are equal to constant e and index i, (i=1,2... m*i*j) are defining on special technique [1]. Each expert has a weight index of his information.
Analysis of the damages and damage's reasons on the Ukraine plants are showing, that a damage state is raising not immediately and has an own prehistory. On example, on the Khmelnitsky sugar plant from 1985 to 1994 years were observed the electrical motors of circular pumps, fuel pumps and irrigate station's pumps. They power are from 400 to 800 kW, on 6 kV voltage. In result we have gotten the distribution of damages on years and on kinds. In this way was fixed the operating indexes during special tests, that is, were made the cause-outcome observes. expert diagnostics electrical machine
Here we show a one of methods of expert system programming to autonomic knowledge base rules organizing by the special tests operating indexes. Those indexes are describe the degradation of observed object exploring conditions with criteria of independence [2].
It is assume we have new operating indexes values; those fixed on the periodic special tests' results of identical objects' group for certain period. For searching of the stochastic dependence of damage from list to every operating indexes we are use the independence criteria . Calculate of the independence criteria's statistics in case of the two signs; every of them has two classes [2], drawing the table for every new rule [1]
Table 1
|
Operating index is normal |
Operating index is over |
Total |
||
|
Failure is in |
d |
b |
a+d |
|
|
Failure is not in |
c |
a |
a+c |
|
|
Total |
c+d |
a+b |
where a,b,c,d -- are numericals of the correspondent damage arise. If a leaded exploration time [1] of the observed objects is not equal we can prognosis the certain index's values with smaller leaded exploration time to a damage time of the object with largest leaded exploration time to the damage. The independence criteria's statistic is computing by equation [2]
=(ad-bc)2(a+b+c+d)/[(a+b)(c+d)(a+c)(b+d)] (3)
and compare with a table value of Pirson's distribution value 2 with a number of freedom degree n=1 and to set trusting probability (for example =0,001 ) [3,6].
For example, on table 1 data will building tow rules on index -- Slot insulator's coefficient of absorbtion, and on outcome -- “winding shortage on frame” (table 2) and “mechanical bearing damage” (table 3)
Table 2
|
Operating index is normal |
Operating index is over |
Total |
||
|
Failure is in |
5 |
18 |
23 |
|
|
Failure is not in |
189 |
0 |
189 |
|
|
Total |
194 |
18 |
212 |
|
Table 3
|
Operating index is normal |
Operating index is over |
Total |
||
|
Failure is in |
39 |
3 |
42 |
|
|
Failure is not in |
155 |
15 |
170 |
|
|
Total |
194 |
18 |
212 |
|
On the first case independence criteria statistic is equal
On this case the data are contradict to independence hypothesis (probability < 0,001).
On the second case
The rule based on the table 3 data has no sense, and a probability of slot insulator damage with absorption coefficient value decreasing to 1,5 is over 0,999, i.e.,
=10,83.
After finding of the dependency between operating index's values and probably damages we can define the distribute indexes for the new rules.
The knowledge base forming result is the table of probability's distribution factors ( и i=1,...,к; j=1,2,...,m.), for the correspondence rules on the к operating indexes and m damages.
For the making equations those allowed to correct indexes of probability distribution we will use the maximum plausibility method [2]. By the known Veibul probability distribution density [3]
(4)
the plausibility function L determine. For the different values of a observed index , by that took place damages on the identical objects r or units the plausibility function has visible
,
where n -total objects number; -real controlling index value.
The estimations of values and is making by [2]
, .
In result we gotten [4] an equation system
if assume
and take account that this rule correcting method is correct when is tend to 1, we can assume
.
In practice this told that with the quite large differences between real and waiting index's values do not take a right to correct the rule. The equation system determine values and , has a visually
where m=(n-r).
Now, as take account that a determines of and values are gives to the expert's arbitrary so value's r , and become abstract figures. Will take a note that we correct factor only. This factor takes influence to the form of probability distribution's curve. Factor has no needs to correct since it's characterize the distribution's scale ad as will be shown down could be cite to any value. As after start of the object monitoring take place a damage with correspondence operating index so observed equation system take a visible
(5)
(6)
(7)
(8)
where 1 -preliminary defined value; 2 -new value. In the four equation's system we have four unknown values , , r and 2 with not varied . Will rewriting the equations (5-8) with the next changes: and the value will define by equation that determine after deciding the equations (5,6) dependable when =0
We have take account that less then 0, we can define a condition when is a positive and negative value
when
when
The deciding of equations; system (5-8) with account to the enumerated before changes is lead to equation
that is the quadratic equation dependable . To the quadratic equation would has two equal roots its discriminant must equal to 0. In this case value is determine from expression
Will use the Pierson's criteria with 2m freedom degree and confidence probability 0,01 or 0,05 for the refined factor value changing suitability estimating
The significance estimation value could not more than 0.2 to take this procedure of refining the advantageous.
Will looking for as example the procedure of rules correcting the one that describe the resource forecast of the AC motors' slot insulator condition. The insulator resource is determine by the explore regimes, temperature and number of becoming regimes. The dependence from temperature [5]
where A,B -constants which dependence from the insulator constructs and technological distinctions. AC motors' insulators could be workable when are over warming in a long time during that could be changing the working regimes those substantially influencing on the insulators' resource. That's why as a calculation of the equation we propose one takes account that in average in 700 switch's regimes per hour so the insulator resource is decrease in 2.4 times [5]
where average number of switches; - measured temperature of insulator.
For example will determine resource of the insulator band ЛС-ФЛЭП-Т based and impregnated with compound ПК-11 for electrical motor 2ЭДКОФ250М4 by this equation. The factors values in this case are А=-5,32 and В=4261,29, that correspondence to nominal resource 180,000 Hs. Adjustment explored time would determined from equation
where t -real explore time on discretizied on index's distances that characterized the explore regimes. On the figure 1 is show a result of this equation using, i.e., was determined the referenced and real time.
In this example the nominal insulator resource is 180,000 hours. However in practice are probable different explore and technological distinctions of the concrete insulator that arise in motors' exploration process. The determined curve is describing by Veilbul distribution equation with correspondence factors and . The curve bend is characterizing by factor.
Table 4
|
Adjustment insulator damage time of one of investigated slots on explore time =100,000 h. |
|||||
|
Number of investigated slots n |
1 |
2 |
5 |
10 |
|
|
Difference between prior and computed values of factor |
-16,76 |
-16,81 |
-16,84 |
-16,84 |
|
|
Significance estimation of factor value |
0,03 |
0,04 |
0,08 |
0,14 |
|
|
Adjustment insulator damage time of one of investigated slots on explore time =185,000 h. |
|||||
|
Number of investigated slots n |
1 |
2 |
5 |
10 |
|
|
Difference between prior and computed values of factor |
154,6 |
204,5 |
240,5 |
253,9 |
|
|
Significance estimation of factor value |
0,002 |
0,003 |
0,007 |
0,016 |
Now will looking for how would change the curve form if will come the insulator condition changed before 180,000 exploiting hours and after. On the figure 2 are showing dependencies of insulator wear from adjustment explore time. The curve 1 is correspondence to the calculated dependency. The curves 2 and 3 are showing the corrected dependencies in cases when take place a substantial change of insulator's condition after 100,000 and 185,000 exploiting hours consequently.
In the table are showing dependencies between corrected index's values and number of investigated slot's cases when take place substantial changes of insulator's condition after 100,000 and 185,000 exploiting hours consequently. The factor value is =39,59.
Thus was determined that using of this method take possibility to correct distribution curve of insulator resource break probability on the concrete explore conditions with its estimation. The identically operations would be use to other units of motor. There are bearings or edge pockets of powerful motors. The factor to correcting may be any other operating index.
It is important for the diagnosis system to make account that we must to estimating the real time checking values, especially in the not nominal regimes of the technical object's working. The technique for the multy factors analysis is describing in [1].
Reference
Maxim V. Shalomygin An intelligence expert system creating technique for the electric machines monitoring. -Kiev. 1996. -78pp. -(preprint/Ukrainian NAS. Institute of electrodynemics; № 785)
Fisher R.A. Статистические методы для исследователей.-М., Госстехиздат. 1958г. 268 с.
Гнеденко Б.В., Беляев Ю.К., Соловьев А.Д. Математические методы в теории надежности. М.,"Наука", 1965г., 524с.
Надёжность электронных элементов и систем. Под ред. Шнайдера Х. -Мир., 1977г., 258с.
Котеленец Н.Ф., Кузнецов Н.Л. Испытание и надёжность электрических машин. - М.: Высш. шк., 1988. 232с.
Polard J.H. Numerical and statistical techniques. -Cambridge: Cambridge University press. 1977. 349 pp.
Размещено на Allbest.ru
...Подобные документы
Призначення пакету Forecast Expert, його використання для прогнозування однофакторної залежності та оцінка її довірчого інтервалу. Створення фінансової моделі підприємства за допомогою Project Expert. Практична робота з програмою "Бізнес-прогноз 2.0".
контрольная работа [1,7 M], добавлен 24.05.2009Программа Project expert, ее сервисные возможности и удобства освоения. Разработка бизнес-планов, оценка и реализация инвестиционных проектов. Экспертные заключения, анализ изменений и автосоздаваемые таблицы. Сравнительный метод оценки стоимости бизнеса.
презентация [1,4 M], добавлен 29.11.2011Сущность методов дисконтированного периода окупаемости, чистого современного значения, внутренней нормы прибыльности. Назначение программы Project Expert, описание ее интерфейса. Основные операции: оценка рисков, анализ чувствительности и безубыточности.
курсовая работа [1,2 M], добавлен 28.11.2010Робота з фінансово-аналітичною інформаційною системою Project Expert; основні функції та модулі системи, їхній опис. Використання системи для створення інвестиційних проектів, їх аналізу та формування бізнес-плану. Опис послідовності виконання завдання.
лабораторная работа [20,5 K], добавлен 03.03.2009Построение модели инвестиционного проекта при помощи программы Project Expert. Разработка стратегии финансирования, анализ финансовых результатов. Формирование и печать отчета, ввод и анализ данных о текущем состоянии проекта в процессе его реализации.
курсовая работа [1,5 M], добавлен 12.10.2013Изучение назначения и основных задач, которые решает Project Expert - система поддержки принятия решений (СППР), предназначенная для менеджеров, проектирующих финансовую модель нового или действующего предприятия. Программные приложения, этапы работы.
реферат [30,7 K], добавлен 19.05.2010Вычислительные и графические возможности сред MS Excel и MS Word для организации производства нового вида продукции. Анализ динамики основных показателей деятельности предприятия, составление бизнес-плана его развития на основе среды Project Expert.
курсовая работа [1,5 M], добавлен 06.12.2009Назначение, основные задачи, функциональные возможности программы Project Expert. Входные и выходные данные системы. Инструментальные средства для работы пользователя. Основные программные модули и диалоги. Решение финансовых и транспортных задач в Excel.
контрольная работа [3,0 M], добавлен 07.08.2013The material and technological basis of the information society are all sorts of systems based on computers and computer networks, information technology, telecommunication. The task of Ukraine in area of information and communication technologies.
реферат [29,5 K], добавлен 10.05.2011Technical and economic characteristics of medical institutions. Development of an automation project. Justification of the methods of calculating cost-effectiveness. General information about health and organization safety. Providing electrical safety.
дипломная работа [3,7 M], добавлен 14.05.2014Создание маркетингового инструмента – анкеты для определения спроса на услуги, товар на дальневосточном рынке. Использование компьютерной программы Project Expert для создания и анализа финансовой модели нового предприятия. Разработка веб-страницы.
статья [1,6 M], добавлен 23.06.2014Information security problems of modern computer companies networks. The levels of network security of the company. Methods of protection organization's computer network from unauthorized access from the Internet. Information Security in the Internet.
реферат [20,9 K], добавлен 19.12.2013Описание принципов работы в среде Project Expert. Операционный план сбыта продукции. Отражение общих издержек. Внедрение инвестиций в процесс производства. Оценка финансовой деятельности и отчетов о прибыли, кэш-фло, эффективности и конкурентности.
курсовая работа [2,0 M], добавлен 03.11.2017A database is a store where information is kept in an organized way. Data structures consist of pointers, strings, arrays, stacks, static and dynamic data structures. A list is a set of data items stored in some order. Methods of construction of a trees.
топик [19,0 K], добавлен 29.06.2009Структура экспертных систем, их классификация и характеристики. Выбор среды разработки программирования. Этапы создания экспертных систем. Алгоритм формирования базы знаний с прямой цепочкой рассуждений. Особенности интерфейса модулей "Expert" и "Klient".
курсовая работа [1,1 M], добавлен 18.08.2009Составление плана проекта создания нового предприятия по производству автомобилей. Создание базы данных по ресурсам в программе Project Expert. Применение методики PERT для анализа проекта. Контроль выполнения задач проекта по срокам и трудозатратам.
курсовая работа [3,7 M], добавлен 11.10.2014Современная система управления проектами ProjectExpert и Microsoft Project 2007. Project Expert – разработка бизнес планов и оценка инвестиционных проектов, возможности программы. Управление проектом "ОАО Ниф-Ниф" в программной среде Microsoft Project.
курсовая работа [3,0 M], добавлен 14.05.2015Особенности составления бизнес-плана нового производства по указанным исходным данным с использованием компьютерных технологий. Методы формирования графика окупаемости проекта. Характеристика полученной отчетности. Анализ эффективности инвестиций проекта.
контрольная работа [2,9 M], добавлен 21.09.2010Загальна характеристика та характеристика функціональних можливостей програми Ргоjесt Ехрегt. Постановка задачі та варінти індивідуально–практичних завдань. Порядок виконання лабораторної роботи в Ргоjесt Ехрегt. Контрольні питання по даній темі.
методичка [2,2 M], добавлен 14.09.2010Сучасна практика складання бізнес-плану, методики та стандарти його розробки. Програмні засоби бізнес-планування. Характеристика та особливості застосування пакету деяких прикладних програм: COMFAR, Project Expert, Альт-Інвест та Альт-Інвест-Прим.
реферат [22,9 K], добавлен 03.06.2011
