A working process of oil-chemical industry is known to be complex multivariable control object with distributed parameters and the important element of the hierarchical control system is a human operator. Operator can control process and compensate the destructive internal and external actions on the control object, if it is necessary.
The control system of the working process must work efficiently, under the normal conditions of operations and acceptable deviations of parameters from their nominal values. In the case of emergency and unacceptable parameters deviations, the control system must provide transfer of control functions to operator. The purpose of control process in general is to keep process parameters in acceptable limitations. If the process parameters are outside acceptable limitations, then control system must provide their normalization for minimal interval of time.
In order to ensure the necessary level of reliability of decision making and their efficiency, the active expert systems are used. This report concerns the experience of creations of knowledge base of control for real technological process in oil-chemical industry. This process deals with separations of pyrolysis gas and has following characteristics: 200 apparatus, 150 parameters of control and inspections.
The structure of active expert system is considered, including the investigation of issues of knowledge engineering, membership functions definition on the base of expert estimates. Expert system can control about 200 parameters of process and contains about 50 fuzzy rules, that describe control system behavior, if the working parameters of process have unacceptable deviations. In some cases (e.g., the switching of apparatus under the fault of one of them or failures of information-control devices) expert system doesn’t control, but serves as supervisor, since operator’s participation in control loop is necessary according to normative documents. The control rules have standardized structure, which is determined by efficiency of mechanism of processing and implementation.
In order to increase the reliability of expert system proposed, the program language LISP is used as the development language, that can be interface of expert system due to functional models of computing, interpreting nature and simplicity of LISP-system interface. Taking into account the inertia of chemical-technological process speed of data processing by expert system proposed it is possible to control and manipulate much more complicated objects.
The prototype implementation of expert system for the control process, based on RISC-microprocessor stack-architecture is discussed. The problems of realization of knowledge base, the expert system core and operator’s interface are considered.
L.G.Muradkhanova EXPERT SYSTEM FOR BUSINESS APPLICATIONS
Expert Systems(ES) are computer systems seeking to emulate the decision behavior of a human expert. Starting from applications in medical diagnosis, mineral exploration, and computer configurations, presently ES are spreading into complex business applications such as managing assets and liabilities, corporate planning, tax advice, competitive bid preparations, internal control evalutions, and fault analysis. Major effects are now underway in industry, government, and science to exploit this technology and extend it to new applications, especially in areas where human expertise is in short supply.
ES try to emulate the decisions of an expert in some particular problem domain and include ways to automate decisions in repetitive environments. These are appropriate when rare expertise exists or when complex operations would be improved by precise actions. Rare expertise can be preserved and multiplied. The effort involved is fairly intense, requiring thorough thought for complex decisions. This makes it necessary for a repetitive decision in order to make the effort worthwhile.
Knowledge base of ES for price forecasting includes the following rules :
IF Supply>0 AND previous_price>o AND Demand=(Supply)
THEN Newprice=( previous_price)
BECAUSE “The market is efficiently clearing”;
IF Supply>0 AND previous_price>o AND Demand>(Supply)
THEN Newprice=(1.1* previous_price)
BECAUSE “There is an obvious upside movement in market prices.”;
ES for marketing application asks several questions concerning demand, supply and previous prices : “What is the demand for your product per month?”; “What is your supply of this product per month?”; “What are you currently charging per unit?” and recommends new price.
ES is realized on the base of VP-EXPERT shell.
V. E.Gliznutsin, E. S.Gliznutsina THE FORMALIZATION OF ACCOUNTING INFORMATION IN DECISION MAKING SUPPORT SYSTEM
The management of business activity of enterprise presupposes the realization of the following functions :
planning (global, strategic, current);
coordination and adjustment;
The information database is necessary for the realization of these functions. Today the accounting information is used for this database. However, the formalization methods of accounting data are intended for the special accounting tasks, but not for managerial ones. Undoubtedly, the existing formalization mechanisms of accounting information must be used in the information database. At the same time, they must uniformly define any economic essences and processes between them.
The managerial tasks as to various processes are divided into financing tasks, investing tasks and operating ones . Evidently, the financing tasks are solved in the process of global planning and deal with such essences as Capital, Current Liabilities, Long-term Liabilities and Retained Earnings. The investing tasks are solved in the process of strategic planning and deal with such essences as Human Resources, Fixed Assets and Intangible Assets. At last, the operating tasks are solved in the process of current planning with such essences as the type of purchase resource, the type of produced product and the type of saled ones. The asynchrony of business operations demands the creation of intermediary buffers.
The buffers of inventories, receivables, cash, etc. are supposed for the level of current planning. These processes are studied by the science of logistics.
The potencial of enterprise (Human Resources, Fixed Assets and Intangible Assets) are the buffers planned on the strategic level. Some processes on this level reflect the increase (decrease) of the potencial of enterprise as a result of purchase (sale) or production. The other processes reflect the amortization or depreciation of the enterprise potencial.
The buffers of global planning are Capital, Liabilities, Earnings, Monetary Assets. They characterize the aims of various groups of people dealing with the enterprise. Some processes in the global planning are connected with the establishment (extinguishment ) of liabilities, and the other processes with the distribution of earnings.
It is suggested to present the structural model of any business process by the three-dimensional matrix R (M x N x V) with elements r m, n, v, where
m=1,... M are the types of essence created by the process (of resource, of products, of capital, etc.);
n=1,.. N are the "space-time" elements of process (technological operation, the centers of responsibility, the periods of planning);
v=1,... V are the types of costs to provide the process.
1. Хан Д. Планирование и контроль: концепция контроллинга: Пер. с нем. - М.: Финансы и статистика, 1997. - 800с.: ил.
Helfert Erich A. Techniques of financial analysis / - Burr Ridge, Illinois, IRWIN, 1994.