You are on page 1of 2

PCM-CMM-2015 – 3rd Polish Congress of Mechanics & 21st Computer Methods in Mechanics September 8th –11th 2015, Gdansk,

mber 8th –11th 2015, Gdansk, Poland

Event driven approximate reasoning*


Krzysztof Psiuk1
1
Institut of Foudamentals of Machinery Design, Silesian University of Technology
Konarskiego 18a, 44-100 Gliwice, Poland
e-mail: krzysztof.psiuk@polsl.pl

Abstract

This paper presents a concept of using event driven architecture in expert system. The considered expert system is a dynamic expert
system based on the multimodal statement networks. It is presented a concept of statement and multimodal statement networks. The
expert system is using a notice board on which values of statements are stored. It was assumed that the change of statement values
presented on the notice boards could be described as an event. So it is proposed to use an event driven architecture technic in the
inference process in such expert system.

Keywords: artificial intelligence, event driven systems, intuitionistic network,

content is a fixed element of notice and statement value is a


1. Introduction variable element of notice.
An interesting solution which can be realized in dynamic
One of the main objectives of technical diagnostics is the expert systems could be event driven reasoning. In this solution
assessment of technical condition of devices and machines. the reasoning process will be started by different events placed
Owing to the development of methods involving both the on the notice boards.
possibility to carry out measurements and also the processing of
the existing data, the assessment of technical state can be 2. Event driven systems
carried out basing on a great amount of data. Such a large
number of input data leads to a large number of investigated The event driven systems are systems based on an Event
diagnostic symptoms. In such cases, it is recommended to apply Driven Architecture (EDA) [5]. The event driven architecture is
expert systems. Each of the expert system consists of many a programing paradigm determining the behavior of a computer
typical modules. Among others they include: program based on events [4]. An event is an occurrence or
 user interface module, happening, which originates inside or outside a system, and is
 knowledge base module significant for, and consumed by, a system’s component [1].
 database module, Events are characterized by their types. We can distinguish two
 inference module. types of events [1]:
Considering expert systems for their operation, we can  Primitive events,
distinguish two groups of such systems. The first group includes  Composite events.
static systems. Static systems are systems, working in the so Primitive events, are atomic and occur at one point in time.
called off-line mode and search for the solution in fixed Composite events include several primitive events. In this case
environment. The reasoning process in this kind of systems is primitive events occur in a time range and have a specific
realized using fixed data that are collected from the technical pattern. In composite event we can distinguish an initiator and a
object. The second group of expert systems include dynamic terminator. An initiator is a primitive event, that start the pattern
systems. Dynamic systems work in the on-line mode. The of composite event. A terminator is a primitive event that
reasoning process in this kind of systems is realized in limited completes the composite event. The occurrence time can be
time and with limited resources. So the reasoning process could defined as a time of the terminator or can be represented as a
be realized using different methods. One of the method pair of times, one for the initiator event, and the other for the
assumes, that the data acquired from the technical object are terminator event [1].
frozen for some time interval and the reasoning process is An event can be defined as a pair of two values:
realized like in static systems. e = <h, b> (1)
An interesting example of dynamic systems designed to where h is an event header and b is an event body. The event
support processes of machine monitoring are systems based on header is a vector that include information about the event such
notice boards. A special class of this kind of expert systems are as:
systems based of intuitionistic notice boards [2]. In this kind of
 event name,
systems the board consists of notices including statements. A
notice consists of a statement whose content serves as a notice  occurrence time,
content. A statement is a declarative expression, resulting from  occurrence number,
observed facts, or representing an opinion, which can be either  event source,
exclusively true, or exclusively false [2]. Results of the system  possible other elements that are included in the event
could be presented as colored rectangle on the notice board specification.
view [3]. The event body is a vector that include the details of the
The notice presented on the notice board consists of two detected state change.
elements: statement content and statement value. Statement
* Described herein are selected results of study, supported partly from the budget of Research Task No. 4 implemented under The National Centre for Research and
Development in Poland and ENERGA SA strategic program of scientific research and development entitled Advanced Technologies of Generating Energy.
PCM-CMM-2015 – 3rd Polish Congress of Mechanics & 21st Computer Methods in Mechanics September 8th –11th 2015, Gdansk, Poland

Event driven architecture is used in many domains e.g. pro- 4. Conclusions


cess management [6], programming applications or in managing
and control systems. One of the possibility of application of The application of event driven architecture in dynamic
event driven architecture is to use it in reasoning process. expert systems is an interesting solution for the operation of
dynamic expert systems. It allows better use of available
3. Event driven reasoning resources and reducing the time needed to carry out the
inference process. This is especially helpful in currently
One of the main element of each expert system is the available multithreaded operating systems, where are available
knowledge base. The accumulated knowledge is the basis for multiple processors and large memory resources. This approach
carrying out the inference process. The knowledge in expert also enables the integration of expert systems with data
systems can be described in different forms. One the most inter- acquisition systems. Changes in the value of conditioned data
esting form of knowledge representation useful for diagnostic could call the appropriate automatic reaction of expert system
systems are multimodal statement networks [7]. The value of and run an inference process.
statements that is used in the inference process in this kind of
network in many cases are evaluated using data acquisition sys- References
tems or based on values of other statements.
In dynamic expert systems the process of evaluating state- [1] Cristea V., Pop F., Dobre C., and Costan A., Distributed
ments value is realised in the on-line mode. The calculated Architectures for Event-Based Systems. Eds. Helmer S.,
statements values are the basis for the process of inference. Poulovassilis A., Xhafa F. (ed.), Introduction to Reasoning
which is realized in limited time and with limited resource. The- in Event-Based Distributed Systems, Springer Berlin
se limits are tried to be reduced in different ways. One possibil- Heidelberg pp.11-45, 2011
ity of reduction is the use of such an event driven architecture
technic. In expert systems based on the notice board, the state- [2] Cholewa W., Intuitionistic notice boards for expert systems.
ments value are written on it. According to the definition of the Eds: Gruca A., Czachórski T., Kozielski S., Advances in
event, a change in the statement value on the notice board, can Intelligent Systems and Computing, vol. 242, Berlin
be treated as an event, occurring in a given system. Springer, pp. 337-344, 2014
Let the change in statement value will be an event ei. Then [3] Cholewa W. Psiuk K. User interface of diagnostic systems.
we can define a set of events E, describing the events that can Proceedings of the Surveillance 7. International conference,
occur in the system: Chartres, France, October 29-30, 2013. Chartres : Institute
ei  E (2) of Technology, pp. 1-10, 2013
Each event can be described by a pair of vectors h and b in [4] Mani Chandy K., Charpentier M., Capponi A., Towards a
accordance with the formula (1). In this case the header of the Theory of Events, Proceeding DEBS '07 Proceedings of the
event can be described as follows: 2007 inaugural international conference on Distributed
 event type, event-based systems ACM New York, NY, USA, pp. 180-
 statement name, 187, 2007
 statement value timestamp,
[5] Michelson B., Event-driven architecture overview event-
 source of statement value,
driven soa is just part of the eda story, Patricia Seybold
 range of statement value where the event is omitted. Group, Tech. Rep., 2006
In the body of the event we can describe the following ele-
ments: [6] Estruch A., ÁlvaroJ., Event-Driven Manufacturing Process
 old value, Management Approach, Eds. Barros A., Gal A., Kindler E.
 new value. Business Process Management, Procedings of 10th
All events should be consumed by the system. So in the dy- International Conference, BPM 2012, Tallinn, Estonia,
namic system a special function which could be run when the September 3-6, pp 120-133, 2012
event occurs should be defined. The function could realized one [7] Cholewa, W.: Multimodal statement networks for
of the following scenario: diagnostic applications. Proceedings of the International
 if the difference between old value and new value is to Conference on Noise and Vibration Engineering (ISMA
small (less than range of statement value defined in event 2010), pp. 817–830, 2010
header) the event should be omitted,
 the event could initialize the process of evaluation of other
statements value, using some equations or other special
prepared procedures,
 the event could initialize the inference process.
Application of EDA provides an opportunity for the infer-
ence process to a limited extent. We can imagine, that for the
purposes of determining the value of the statement will be nec-
essary to use an inference process. In this case, we can isolate a
small statements network for which the inference process will
be conducted. This will allow better use of available resources
and reduce the time of the ongoing process of inference.

You might also like