THE DATA FLOW AND DISTRIBUTED CALCULATIONS
INTELLIGENCE INFORMATION TECHNOLOGY FOR
DECISION SUPPORT SYSTEM IN REAL TIME
Michael Yu. Okhtilev
St.-Petersburg Institute for Informatics and Automation of Russian Academy of Sciences (SPIIRAS),
39, 14th Line, St. Petersburg,
Special Design Office “ORION”, 12, Tobolskaya str., St. Petersburg, Russia
K
eywords: Unified models of complex technological process, technical states monitoring (situation assessment).
Abstract: The aim of this investigation is to develop unified models of complex technological process as controlled
object states knowledge presentation; methods, algorithms and system of complex technological process
states monitoring (situation assessment) programs automatic synthesis according to preset target and
capability of verification and optimisation considered; special software prototype realizing controlled
objects automatic monitoring.
1 INTRODUCTION
The aim of the fulfilled investigation is to develop
methods, models and algorithms oriented to
concurrent on-line user software assurance for all
sorts of measuring information (information fusion)
specifying states of complex technological process
(situation assessment) at all phases of complex
technological process life cycle and control function
in real time.
This aim should be achieved by here suggested
artificial intelligent information technology. The
suggested newly designed information technology is
based on decision of basic research problem. This
problem is monitoring and control of states of
complex technological process (situation
assessment). The basis of this artificial intelligent
information technology is flow computing models
exploitable by state hipping (constraint
programming) in real time and in territorially
distributed computing network. At the same time
each network node represents artificial intelligent
agent.
2 THE MAIN TASKS AND
RESULTS OF INVESTIGATION
A profound study of theoretical results showed that
there exists a great number of publications in the
area of measuring information (MI) processing and
analysis methods, on the other hand, research in the
areas of design automation for monitoring software
complexes, development of techniques allowing to
arrange for parallel processing and analysis of
measuring information in computing environment
with changing structure are poorly reflected in
literature. The impact of types and structures of the
processed information on the composition and
structure of the considered software complexes is
not well investigated either.
Experimentally received positive results in
software complexes for monitoring creation and
implementation are of heuristic nature and based on
developers intuition and experience; their
elaboration requires running of time and labour
consuming experiments at the synthesis stage.
Moreover, the existing methodology and used
software packages do not meet certain requirements
for embedded special software of territorially
497
Yu. Okhtilev M. (2004).
THE DATA FLOW AND DISTRIBUTED CALCULATIONS INTELLIGENCE INFORMATION TECHNOLOGY FOR DECISION SUPPORT SYSTEM IN
REAL TIME.
In Proceedings of the Sixth International Conference on Enter prise Information Systems, pages 497-500
DOI: 10.5220/0002592804970500
Copyright
c
SciTePress
distributed complex technological process systems
with variable structure functioning in real time.
Known approaches provide ineffective solving
of the following basic problems facing to the theory
and practice of the computer-aided monitoring
system (CAMS) or do not provide them at all:
– absence of uniform conceptual basis of
construction of the information systems (IS) IF of
MI, functioning in various states of application and a
special-purpose destination system;
– impossibility of the formal description of all
possible kinds of technical states (TS) in principal in
view of their adequacy to events and processes,
taking place at Analysis Object (AO) because of the
application of the various mathematical apparatus –
for various in destination purposes of MI analysis;
– presence of a plenty of forms of data
presentation and accordingly, types of models of
knowledge representation (MKR) about AO, caused
by existence of subjective views and specialized
approaches in the different interested organizations
engaged in problems of monitoring of states of AO
that interferes with accumulation, systematisation
and distribution of a wide experience to practice of
operation of advanced IT AA MI, proved their
utility;
– impossibility, as a rule, to form automatically
correct and optimum program AA MI on any set for
concrete session of management of the purpose of
the analysis in view of specific states of its
realization with the help of existing scientific and
practical approaches:
– inability of applied theories, models and
algorithms known at the present time to carry out
AA in view of essential features of MI coming for
the analysis, among which it is possible to note its
natural parallelism, stream type, poorly predicted
intensity, presence of uncertain measurements, e
TS;
– absence of a theoretical and experimental
substantiation of structure and composition of SPO,
capable to provide the effective solving of problems
of the distributed CAMS and its basic (from the
point of view of a saturation of the used
mathematical apparatus ) element – IS IF of MI;
– presence of the big expenses for the
modernization and support of the big Program
Complex (PC) IF of MI which reliability is inversely
proportional to their volume;
– orientation of the used modelling-algorithmic
support, basically, on one parametrical optimisation
of speed of carrying out of IF, when speed of
reception of the results of AO states monitoring
depends only on capacity (and also on cost) of the
used computer complexes- independently of their
architecture and also of some others, less essential.
Monitoring automation for real-time supervision
of complex technical objects (CTO) involves several
unsolved problems:
– lack of universal methodological and
methodical basics for structure-functional synthesis
of information technologies and monitoring systems
for CTO;
– the existing models, methods, and algorithms
for processing and analysis of measuring
information (MI) do not thoroughly describe its
properties such as natural concurrency, flow
characteristics, inconsistency, and uncertainty;
– there is no theoretical foundation for computer-
aided design of real-time CTO monitoring software
in distributed control systems;
– information fusion concepts and algorithms for
integration of multi-domain telemetry data and
practical knowledge are not developed yet.
Therefore the following tasks are very actual and
are investigated (Okhtilev, 2001):
1. Review of existing research and technical
approaches to solve in real time problems of
monitoring and control for complex technical
systems.
2. Elaboration of a joint conceptual framework
to develop information systems (IS) functioning
under changing application conditions as well as
under different application missions (including
methods of modelling “the changes” for estimated
states and for controlled object based on calculus of
variations).
3. Formal description of all possible kinds of
controlled states (assessed situation) accounting for
their adequacy to actual actions and processes on
controlled object caused by application of different
mathematical apparatus for various functional
objects. Multi-model formalization intends for
describe actions and processes on the controlled
object.
4. Synthesis of automatically correct and optimal
program for monitoring any target preset for a given
control session having specific realization
conditions.
5. Development of models and algorithms for
state estimating oriented to essential features of
measuring information fusion (natural parallelism,
data-flow, weakly predictable intensity, fact of
doubtful measurement, etc.).
6. Development of software prototype using
basic suggested solutions for information fusion
system design
The fulfilled investigations towards the
development of information technologies and
monitoring systems resulted in the following items
(Okhtilev, 1999, 2000, 2001):
– system analysis of monitoring processes
providing reliable estimates of CTO states under
ICEIS 2004 - ARTIFICIAL INTELLIGENCE AND DECISION SUPPORT SYSTEMS
498
conditions of a priori uncertainty and failures was
performed;
– a unified model for description of weakly
formalized information about CTO states was
developed via declarative representation of
knowledge;
– a concept of invariance for CTO states and
monitoring process states was substantiated;
–theoretical basics for construction of
topological space of CTO states were developed;
– methods of automatic monitoring program
synthesis for a given goal were developed and
tested;
– methods of monitoring program optimisation
and verification were developed and tested;
– methods of correctness control for algebras
over computation models were justified and
investigated (Okhtilev M., Bogomolov S. and
Dmitriev A., 1990);
– unified reliable computation-model based
methods for local recognition of states were justified
and investigated (Okhtilev M., Bogomolov S. and
Dmitriev A., 1990);
– concepts of modern information technologies
were adjusted for design and maintenance of
software for state monitoring in real time (Okhtilev,
2000, 2001).
Advantages of the proposed approach to
synthesis of intellectual information technologies
and systems are based on new fundamental scientific
Subsystem of user interface
Subsystem for graphical (multimedia)
presentations
Subsystem of autonomous data operations
Subsystem for control of analysis process
Subsystem for diagnostics and supervision of CTO
Subsystem for database multi-functional queries
Subsystem for dialogue forms description
Subsystem for dynamic models of parameters
alteration
Subsystem for description of parameters value
segmentation
Subsystem for description of parameters groups
Subsystem for description of measured and
computed parameters
Databases
Special interactive
subsystems
Automatic synthesis of a
correct meta-program
GUI generation
Behavioral modeling
Conceptual modeling
M O N I T O R
of the operational
environment
Component
ware
Network
services
Operating
system
Tools
of the
programming
languages
Figure 1:
The structure of operational environme
THE DATA FLOW AND DISTRIBUTED CALCULATIONS INTELLIGENCE INFORMATION TECHNOLOGY FOR
DECISION SUPPORT SYSTEM IN REAL TIME
499
results received in topology, algebra, mathematical
logic, and the theory of artificial intelligence. This
approach provides general scientific grounds for
different important classes of CTO monitoring and
control problems. The problems that had only
heuristic solutions, if any, can be solved via the
developed unified models and methods of
information fusion, program synthesis, and
verification.
3 CONCLUSION
Suggested intelligent information technology will
allow to reduce costs and shorten complex
technological process underlying the elaboration of a
system for monitoring and control of states of
complex technological process, and will facilitate
significantly its further modification. At that systems
for real time monitoring of states of complex
technological process acquire principally new
qualities. Particularly they allow to monitor in real
time the states of complex technological process
characterized by great number of measured
parameters under reconfiguration of controlled
object structures. Based on the proposed information
technology the automation level for control of
complex technological processes could be advanced,
possibilities of control of objects at degradation of
their structures extended, reliability and efficiency of
control processes increased, possibilities of early
detection of various technical faults as well as timely
prediction of catastrophes allowing to make right
decision and take appropriate prevention measures
evolved. Moreover, suggested objects monitoring
information technology can substitute most of
currently operating software of the above analysed
application when used for a development of
embedded and scalable software. Implementation of
the special software based on the suggested
intelligent information technology looks promising
for many critical applications.
The pilot versions of CAMS for CTO states
supervision (in space systems and atomics) work in
network of IBM/PC-compatible computers; it uses
special operational environment (Okhtilev, 1999,
2000, 2001), real-time database management
system, multi-window interface, and programming
language C/C++. The structure of the operational
environment is shown in Fig. 1. The prototypes of
CMS belong under the class
MMI/CACSD/SCADA/MAIS (man-machine
interface/ computer-aided control system
design/supervisory control and data
acquisition/multi-agent intellectual system). A
command is the basic instruction that a script file
contains. Some commands require parameters that
further define what the command should do. An
expression is a combination of operators and
arguments that create a result. Expressions can be
used as values in any command. Examples of
expressions include arithmetic, relational
comparisons, and string concatenations.
REFERENCES
Okhtilev, M., 2001, Specifics of technology for
development of special computer-aided systems
canalising information measured in real-time,
Automatic Control and Computer Science, Allerton
Press Inc., New York, Vol. 39, No. 6.
Okhtilev, M., 1999, Topology in a set of measurable and
computable parameters in real-time estimation of the
state of complex technical objects, Automatic Control
and Computer Science, Allerton Press Inc., New York,
Vol. 33, No. 6, pp. 1-8.
Okhtilev M., 2000, Topological approach to construction
of computation algorithms in real-time estimation of
complex technical objects, Automatic Control and
Computer Science, Allerton Press Inc., New York,
Vol. 34, No. 1, pp. 8-16.
Okhtilev M., 2001, Construction of Programs for Real-
Time Processing and Analysis of Measuring
Information, Programming and Computer Software,
Kluwer Academic/Plenum Publishers.
Okhtilev M., Bogomolov S. and Dmitriev A., 1990,
Artificial Intelligence Methods in Problems of Control
and Technical Diagnosis In Proceeding of the IFIP.
TC5/WG5.3. International Conference in CIM,
Leningrad, USSR, 16-18 April 1990. Editors D.Michie
and L.I.Mikulich. P. 248-253.
ICEIS 2004 - ARTIFICIAL INTELLIGENCE AND DECISION SUPPORT SYSTEMS
500