CC BY
5
ЭЛЕКТРОННЫЙ НАУЧНЫЙ ЖУРНАЛ «APRЮRI. CЕРИЯ: ЕСТЕСТВЕННЫЕ И ТЕХНИЧЕСКИЕ НАУКИ»
УДК 66.047
№ 2 2016
THE USE OF EXPERT SYSTEMS IN RESEARCH OF PROCESSES
OF LIQUID PRODUCTS DRYING
Skripnikova Svetlana Gennadevna student Sirotkin Aleksei Olegovich
student
Zagrebnev Roman Sergeevich
student
Tambov state technical university, Tambov
Abstract. Describes methods of work with the expert system used by the authors in the study of heat and mass transfer during drying of a liquid dispersed product.
Keywords: expert system; research; drying; liquid; dispersed product.
ИСПОЛЬЗОВАНИЕ ЭКСПЕРТНОЙ СИСТЕМЫ В ИССЛЕДОВАНИЯХ ПРОЦЕССОВ СУШКИ ЖИДКИХ ПРОДУКТОВ
Скрипникова Светлана Геннадьевна студент Сироткин Алексей Олегович
студент Загребнев Роман Сергеевич студент
Тамбовский государственный технический университет, Тамбов
Аннотация. Описаны методы работы с экспертной системой, используемые авторами при исследовании тепло и массопереноса в процессе сушки жидких дисперсных продуктов.
Ключевые слова: экспертная система; исследование; сушка; жидкость; дисперсный продукт.
Artificial intelligence system is a set of knowledge, data and system processing based on a computer that implements the process of thinking and decision making, as it makes people. This is a general theoretical concept. The main difficulty in implementing this comprehensive system lies in the modeling of human thinking. This is due to one philosophical problem which tried to resolve the outstanding thinkers of all time - how to know the man himself, his conscience, the principles of thinking, actions, feelings?
The developers of artificial intelligence systems is very soon faced with this issue and realizing that «one cannot embrace the unembraceable», went more simple way. They went towards the implementation of individual processes of human behavior and solving problems associated more with mathematical logic than from intuition, imaginative thinking, analogies [1].
This is the problem of optimal traffic management, routing of pipelines, processing equipment, processing of experimental data from instruments in real time, etc. at all of the problems that arise in scientific research and in engineering practice, chemical engineering are fertile ground for the implementation of such systems [2].
Systems that implement these decisions became known as expert systems [3]. The basis of the mathematical apparatus of such systems amounted concepts of fuzzy logic, operating with the terms «almost», «close to», «more», «less», etc.
Today expert systems are used in modeling different kinds of intellectual activity: from simple card games to systems for medical diagnostics. However, the results of such systems require constant evaluation of the accuracy of the results [4]. There are a number of objective reasons.
We use an expert system of initial level for the implementation of the simple problem of flow distribution in the drying installation of the required processing order of ten rules [5]. In the system associated with a distillation installation continuous action requires the processing of about fifty rules. In this part of the rules often provides inconsistent results, and this requires the
introduction of additional restrictive rules and facts. The algorithm of obtaining solutions becomes complicated in proportion to the increase of rules and facts [6].
Thus, the expert system at the studied, simple and consistent subject area are often not always gives the correct decision or issues an obvious solution with a large time delay or by using the operator in an automated mode.
About the correctness of the expert system is often said in the mathematical probabilities. For example: the system gives the correct decision in 70 of 100 cases (unless of course the correctness of the solution can be controlled). Usually the system is operating with a small amount of knowledge, gives the correct solution faster and with greater probability than the system operating with a large volume of knowledge. For example, the expert system layout electronic circuit boards employs more than 8.5 million claims and gives the correct solution with a probability of 98 %.
In the field of chemical technology is known at least a dozen systems operating with a volume of facts and rules, comparable to the volume of expert knowledge in a particular subject area (estimated to be about 25-40 million statements) and outstanding decision with a probability of 90-96 %.
We use expert system, allowing you to select and calculate equipment for drying the dispersed liquid products in the field of chemical and food production. A system of rules and assertions are based on data from textbooks, reference books, as well as on data obtained from experimental studies of the kinetics of drying [4].
At the moment the system works with more than 1700 rules. The main problem with this system is the need to control the intermediate and final results of the person. This feature ES especially important to consider when designing systems working in areas directly affecting the life and destiny of a person, such as medicine, production management system, transport management. The price of failure, errors, incorrect decisions is human life. Without human control such a system would do more harm than good.
Errors can occur, first, as already noted because error in the software, and secondly, due to failures in hardware. «Stuffing» of the computer chip, unstable under fluctuations of current and temperature (especially when thresholds are exceeded). Therefore, system failure can happen for the most innocuous reason, in the conventional situation. Solving these problems, science and technology are on the ways of increasing reliability of hardware, increase the performance of chips while reducing their size.
The increase in compactness leads to the increasing complexity of the internal structure of the chip, which reduces its noise immunity as magneto, and temperature. It is necessary to note one feature of the processors that is generally not spoken. None of the processors in the world can be fully tested. Simply put, while the processor of a definite combination of signals at the input must match a certain combination of signals on the output. Only in this case the processor is working correctly. However, possible combinations of the input signals so much that you should check all of them physically impossible. And this is another source of possible hardware failure.
Список использованных источников
1. Pakhomov A.N. Method of determination of adhesion of the film dries distillery grains on the substrate / A.N. Pakhomov, R.Y. Banin, E.A. Chernikh, E.Y. Loviagina, N.A. Sorokina // Proceedinge of the 5th International Academic Conference «Applied and Fundamental Studies». St. Louis, Missouri, 29-30 April 2014. S. 71-73.
2. Pakhomov A.N. Formation and behaviour of fluidized bed of inert particles / Pakhomov A.N., Volostnykh S.G., Eltsov A.G., Shuvaev L.S. // European Applied Sciences: challenges and solutions 2nd International Scientific Conference. Stuttgart, Germany, 2015. P. 119-120.
3. Pakhomov A.N. The influence of the shape of the dryer to the nature of binary fluidized bed of inert / A.N. Pakhomov, R.Y. Banin, E.A. Chernikh, E.Y. Loviagina // Applied and Fundamental Studies Proceedings of the 8th International Academic Conference. Publishing House Science and Innovation Center. St. Louis, Missouri, USA, 2015. P. 121-123.
4. Pakhomov A.N. The effect of feed slurry to the nature of the fluidized bed / A.N. Pakhomov, R.Yu. Banin, E.A. Chernikh, E.Yu. Loviagina // The Fifth International Conference on Eurasian scientific development Vienna, 2015. P. 122-123.
5. Рakhomov A. The observed heterogeneity of the fluidized bed / A. Рakhomov, R. Banin, E. Chernikh, E. Loviagina // Scientific enquiry in the contemрorary world: theoretical bas^s and innovative aррroach / ed. by A. Burkov. San Francisco, California, USA, 2015. P. 70-72.
6. Pakhomov A.N. Influence of walls of the device on frame of fluidezed bed of inert particles / A.N. Pakhomov, R.Yu. Banin, E.A. Chernikh, E.Yu. Loviagina // Europaische Fachhochschule. 2015. № 3. P. 67-68.
