Modern information
technologies. /1. Computer engineering.
Doctor of Technical Science Samigulina G.
A.
Institute of Informatics and Control Problems, Kazakhstan, Almaty
Intellectual
expert system of estimation and forecasting of the risks at realization of the
complex projects
One of the actual problems of headily
developing modern information society is processing and the analysis of the huge
flows data. Today there are many traditional approaches to decision of the
given problem; however all of them have the limited scope. Development of new untraditional
biological approaches of imitating modeling produce a great interest in the
world. The most widespread are cellular automatic machines, genetic algorithms and
artificial neural networks. Recently the artificial immune system (AIS)
attracts the special attention, one based on processing of the information by
molecules of proteins and immune reaction of an organism on introduction of the
alien antigens.
Advantage AIS are:
distribution; learning capability, absence of the centralized control; self-organizing
and evolution; and small computing resources. The basic problems arising at realizations
of the artificial immune systems are: the errors of power estimations because of the data deficits,
their correlation and the measurement errors; the problems connected with localizations
of the errors; absence of effective algorithms of training AIS.
In the approach AIS a base
element is formal peptid [1]. Proteins play an exclusive role in the life of all
organisms. Reaction of an organism on the majority of outside influences is
reduced to code conversion of external signals on the language of albumins
interactions.
The fields of application of
the various applied problems based on principles of immunology constantly grow.
The given approach is used for recognition of images; in the intellectual
systems of forecasting, support of decision-making and control [2]; in the systems
of computer safety; in the systems control of remote educations in the sphere
of Internet; forecasting of the pharmacological activity of the organic
compounds at the medicines production etc.
Application of the given
systems is especially interesting for an expert estimation and forecasting of the
risks at realization of complex projects. In this case it is necessary to
collect and analyses of the opinions of considerable quantity of the people on the
many criteria from the different areas of a science for the decision of the given
problem.
The statement of the problem is
formulated as follows: it is necessary to develop the system of the
intellectual analysis of the multidimensional data and forecasting of the risks
at successful realization of the complex project on the basis immune net
modelling for support of decision-making and the operative control.
Following procedures are realized.
Procedure 1.
The groups of the risks which
influence on the project realization are allocated. For example: the macroeconomic risks; the technological risks; the time risks; the financial and economic
risks; the commercial risks; the organizational
risks; the normative-lawful risks; and social risks.
Further the list of the
reasons (subgroups) which cause these risks is made:
- Commercial risks as an economic crisis, difficulties of the project
introduction, the project recoupment etc.;
- Normative-lawful risks as
absence of the uniform standards, absence of the legislative base, difference
from the world standards etc.;
- Organizational risks as
absence of the qualified experts for development and introduction of the project,
the bad organization of the supplies, absence of the operative control etc.
Then with the help of the various
specialist-experts (economists, mathematicians, lawyers, sociologists etc.) make
an examination for the given groups of risks and an estimation of the reasons
on the offered scale.
After that the all information
is brought in a database. Preliminary data processing is carried out: normalization,
filling of the missing data. Then the factorial analysis of the data (the method
of the main component) is applied to separation of the most significant
(informative) risks. The obtained data allow to deeply analyses of the reasons
of the beginnings factors which the negatively influence the purposes and realization
of the project.
Procedure 2.
The etalon matrixes are formed
on the allocated risks for two classes, corresponding to a favorable outcome of
the project and not to a favorable outcome of the project. The immune network
is trained.
Further the matrixes of images
are formed. On the basis of the singular
decomposition matrixes and definition of the minimum energy of bond between
the peptids recognition of images is carried out.
Procedure 3.
Estimation of power errors on
the basis of properties of homologous proteins is realized. Forecasting of
risks of the project and decision-making for control of the project on the
basis of the received forecast are made.
The given approach allows to
process a considerable quantity of the expert’s opinions on the set theme in the
real time at parallel processing of the multidimensional data and to carry out of
the operative control at project realization. The given approach actually
applies at realization of very complex and big projects when the analysis of
risks is complicated.
The
literature
1. Tarakanov A.O. Formal
peptide as a basic of agent of immune networks: from natural prototype to
mathematical theory and applications. Proceedings of the I International
Workshop of Central and Eastern Europe on Multi-Agent Systems, 1999.
2. Samigulina G. A.
Development of intellectual expert systems of forecasting and control on the
basis of artificial immune systems // Computer science problems. - Novosibirsk,
2010. - ¹ 1. - P.P. 15-22.