APPLICATION OF RULES ENGINES IN TECHNOLOGY
MANAGEMENT
Barbara Baster, Andrzej Macioł and Bogdan Rębiasz
Faculty of Management, AGH University of Science and Technology, ul. Gramatyka 10, 30-067 Kraków, Poland
Keywords: Rules Engines, Technology Management, SQL.
Abstract: In the paper we present the initial results of our research aimed at development of the tool which will benefit
from virtues of BRMS and will enable support of technological decisions. Our task was focused on
preparation of use cases set along with precise description of rules used for solving specific decision
problems. For this purpose two decision problems were analysed which covered such issues as selection of
feedstock or executive production planning. These problems were analyzed in view of a company producing
cold-rolled strips in a wide dimension range and diversified grades of steel. The general conclusion which is
the answer to the question of the possibility to create a tool similar to BRE but capable of technological
decision supporting is a statement that it is necessary to combine two forms of knowledge presentation:
declarative and procedural. It is also necessary to ensure the possibility of communication between this type
of instrument and external data sources as well as various types of IT tools supporting specific technological
decisions.
1 INTRODUCTION
Nowadays an integrated information system
supports almost every corporation in all fields of
management. Nonetheless, such support is limited to
realization of administrative tasks and does not
influence significantly the decision making
processes. On the other hand, engineering solutions
are applied in the specific fields of management
(such as logistics, technology management,
transportation etc.), however they are not well
integrated with corporate IT systems. In the recent
years there were many attempts aimed at broadening
the functionalities of IT systems as well as their
integration with specialized tools. These work were
conducted in parallel in two fields. On one hand,
sophisticated expert systems for support of
technological decisions were developed, and on the
other the results of operational research were used
for the development of systems which support
specific decisions (e.g. advanced planning and
scheduling systems). Regrettably, no widely
accepted results were obtained in either of these
fields. A similar issue arose in case of management
problems that are not characterized by standard,
repetitive decisions. It was grounded on the fact that
development of individual IT solutions required
participation of specialists from a given domain, the
development process of application was excessively
long and ultimate solutions were hardly flexible. In
the first decade of the century in the domain of
business rules a new opportunity came to existence,
i.e. BRM systems. They allow for considerable
simplification of application development process
and enable to formulate characteristic repositories of
organizational knowledge. Nonetheless, up to now
such solutions were not successfully applied for
support of technological decisions or more complex
decisions related to management of manufacturing
processes.
Support of technological decisions may be
realized through expert systems. Presented in
literature automatic support systems of production
process design, despite different scope as well as the
manner of presenting technological knowledge, are
characterized by a stiff, a priori formulated
structure. In consequence, user has possibility
neither to develop knowledge nor introduce results
of his own research without reformulating
knowledge base – doubtlessly with cooperation of
system designer.
The research, co-funded with the European
resources, is aimed at development of the tool which
will benefit from virtues of BRMS and will enable
83
Baster B., Macioł A. and RÄ
´
Zbiasz B. (2010).
APPLICATION OF RULES ENGINES IN TECHNOLOGY MANAGEMENT.
In Proceedings of the 5th International Conference on Software and Data Technologies, pages 83-86
DOI: 10.5220/0002931700830086
Copyright
c
SciTePress
support of technological decisions. In this paper we
present results of our research on capabilities of
BRMS and characteristics of technological
decisions.
In particular our task was focused on preparation
of use cases set along with precise description of
rules used for solving specific decision problems.
For this purpose two decision problems were
analysed which covered such issues as selection of
feedstock or executive production planning. These
problems were analyzed in view of a company
producing cold-rolled strips in a wide dimension
range and diversified grades of steel.
2 RULE BASED TOOL FOR
TECHNOLOGY
MANAGEMENT
As part of our project we developed assumptions of
a new tool for decision support, including
technological decisions.
Main stream of research activities in the area of
knowledge modelling for BRE is connected with
application of description logics for ontology
formulation. Our experiences indicate that the most
effective solution, capable of direct cooperation with
majority of industrial information systems which
simultaneously provides decidability, is a
combination of relational model with inference
system that utilizes attributive logics. Our solution,
named Inference with Queries (IwQ) (Maciol,
2008), has been developed as a knowledge model
and an inference engine for formulation of Business
Rules Management Systems. Knowledge storage is
realized in accordance with principles of Variable
Set Attribute Logic (VSAL) (Ligeza, 2006). An
assumption was made, that in rule-based decisive
system, attributes as well as variables will be stored
in form of relations. Owing to the utilization of
extended selection formula, knowledge definition
process becomes simplified. SQL queries realize a
significant part of inference.
The essential functional requirements (solution
features) are as below:
1. User will be able to easily create, delete and
modify knowledge bases.
2. The way of inference (forward and backward)
and process parameters (the start-up fact, the way of
passing messages, the ending conditions) can be set
automatically by the system, the knowledge architect
at the stage of its creation or at the stage of
inference.
3. The system provides a set of tools for acquiring
knowledge of facts from external sources (databases,
procedures, programs, global network).
The inference process in our solution is controled
by variable values of attributes that indicates which
rules will be fired on current step of the reasoning.
3 EXEMPLARY DECISION
PROBLEMS ANALYSIS
In the initial phase of our research we found that the
system must combine the possibilities given by
declarative as well as procedural notation of
knowledge. Our solution gives such possibilities.
Research were carried out in a company
producing cold-rolled strip in a wide dimensional
range and variety of steel grades.
3.1 Feedstock Selection for Actual
Order
Parameters of feedstock selection may be
characterized as follows:
1. Steel grade of feedstock strip and rolled strip
should be identical. There are possible some
departures from this ‘literal’ conformity, especially
in case of low-carbon grades of steel. In many cases
chemical composition of grades of steel resemble
one another. Rules that define changing process of
steel grades may be visualised in the form of
decision grid.
2. The width of the feedstock strip must be greater
than the width of the rolled strip. In order to
minimize ‘unjustified’ loss, different orders are
combined together to prepare a milling from a single
feedstock coil.
3. The width of the feedstock strip should make
allowance for the width of the rolled strip in view of
eventual cutting of the strip into strands.
4. A specific processing rate of feedstock into given
thickness is required.
5. The weight of the feedstock coil should be in a
proper relation to the weight of the order. The
realization of order from the minimum possible
number of different feedstock charges is the most
favourable option in terms of uniformity of produced
strip.
Hereunder we present the exemplary rules stored
in a knowledge base:
ICSOFT 2010 - 5th International Conference on Software and Data Technologies
84
rule 1
if
input_material = ‘hot-rolled’
and
rolling_ratio >= 25%
and
reduction_of_thickness >= 0.5 mm
then
treatment_ratio := ‘sufficient’
rule 2
if
ordered_steel_grade =
current_coil_steel_grade
and
feedstock_coil_width >=
min_adequate_width
and
treatment_ratio := ‘sufficient’
then
feedstock_coil := ‘admissible’
The main task of the rule-based system is to
verify which of the feedstock coils are ‘suitable’, i.e.
whether they can be used as a charge for the
realization of a particular order. Along the inference
process, a given charge is marked as ‘acceptable’ if
it is characterized by the proper grade, width and
‘sufficient’ degree of material reduction. The charge
is ‘suitable’ if some additional conditions are
satisfied, i.e. those related to the mass of the charge,
the size of the crop and the number of strands being
cut. If any of these conditions is not satisfied
(‘acceptable’ charge is not ‘suitable’), possibilities to
combine orders are searched for or the coil is split.
The order which is supposed to be combined
with the one analysed must satisfy the following
conditions:
grades of materials used in the realized order and
the combined order must agree,
the width of strip in the combined order must be
greater than the width in the analysed order,
the thickness of strip in the combined order can
be smaller or equal to the thickness in the
analysed order,
the difference in weight between the coil and the
analysed order/output must be at least greater or
equal to the half of the combined order.
If there is no possibility to combine orders, the
feedstock coil is divided into two strands of equal
width.
If in this very case the conditions dictated by the
rules of knowledge base are satisfied, the coils is
designated for the realization of order.
If the mass of the coil is too big for the
realization of the analysed order, the coil should be
split into two smaller coils. The first coil is then used
for the realization of the order, and the other is sent
back to the store.
3.2 Production Scheduling
The input data for production schedule is:
1. planned sails quantity in tons,
2. production plan in tons,
3. production cards elaborated by Technology
Division containing:
parameters of actual input strip coils,
parameters of final strip;
4. working time,
5. currently active machines,
6. machines productivity and capacity,
7. currently active workers staff,
8. rolling and cutting rates,
9. time priorities of particular order.
Time priorities of order realization arise mainly
from:
confirmation date of order realization,
importance of the client for the company,
declaration of fast realization time,
acquisition of new recipients;
The executive production plan is prepared for a
given monthly production rate on the basis of annual
production plans as well as the set of orders
confirmed for the current month.
An algorithm for preparation of the executive
production plan with the use of knowledge base
must be executed in multiple phases and use both
declarative and procedural knowledge.
In the first stage the rolling mills are assigned to
realization of particular orders.
In the next phase, the individual tasks are
arranged. Formulas that are stored in the knowledge
base allow to determine whether two consecutive
orders are optimally arranged from the point of view
of criteria that define urgency of orders. Those
formulas classify current arrangement of two orders
(i and j) as ‘wrong’ or ‘correct’. If the arrangement
is ‘wrong’, a swap in their order should be
conducted. Exemplary arrangement rules are
presented below:
rule 1
if
distance_i_j = ‘long’
and
customer_i = ‘important’
and
customer_j = ‘important’
then
APPLICATION OF RULES ENGINES IN TECHNOLOGY MANAGEMENT
85
sequence := ‘correct’
rule 2
if
distance_i_j = ‘long’
and
customer_i = ‘important’
and
customer_j = ‘new’
then
sequence := ‘wrong’
where distance_i_j is the distance between time
of i order confirmation and j order confirmation.
The rules listed above are used for verification of
arrangement of two orders in a milling process. They
are applied in the algorithm presented below, which
sorts the orders from the point of view of their
urgency.
For i = 1 to OrdersCount - 1
For j = i + 1 to OrdersCount
If not Sequence(Queue(i),_
Queue(j)) then
X = Queue(i)
Queue(i) = Queue(j)
Queue(j) = X
End If
Next j
Next i
The sequence statement is the call to inference
engine that verifies if the order i should by processed
before the order j.
The successive formulas and procedures assign
the cutting operations to a specific cutting unit,
determine the number of work mode rotations for
individual rolling mills, delegate the tasks prescribed
to a given rolling mill for execution in individual
weeks, and compute a working time of the cutting
units for each week of the month. In every case
mentioned above we deal with a smooth
combination of typical procedures utilizing
numerical formulas with complex logical blocks that
can be effectively implemented by means of
declarative knowledge.
4 CONCLUSIONS
The two problems presented above are very complex
problems of decision-making. This statement
concerns especially the development of a monthly
action plan. Due to the high complexity of this
decision problems the test data will be perfectly
suitable for the verification of the inference engine
model which has been created.
The devised database compatible with the con-
cept of a knowledge base with various attribute
values will be used to verify the assumed
conceptions at the further stages of the project. The
works confirmed that it is necessary for the solution
of complex decision problems in the area of
production technology to apply hybrid systems – the
combination of a rule system and procedural
calculations. It appeared practically in each of the
analyzed problems. The rule systems check the
fulfillment of certain technological requirements
(whether the condition of an adequate degree of
processing of feedstock is fulfilled, whether
reduction of strip is sufficient, whether the
parameters of cutting unit allow to cut a strip of
defined parameters in a set number of strips etc.).
The checking of these conditions is a part of the
algorithms which support analyzed decision-making
processes.
Ultimately, the verification of the proposed
solutions will therefore need to develop programs
realizing procedural calculations and using devised
inference engine.
The general conclusion which is the answer to
the question of the possibility to create a tool similar
to BRE but capable of technological decision
supporting is a statement that it is necessary to
combine two forms of knowledge presentation:
declarative and procedural. It is also necessary to
ensure the possibility of communication between
this type of instrument and external data sources as
well as various types of IT tools supporting specific
technological decisions.
None of the currently available BRE does not
meet the requirements arising from the need to
combine declarative and procedural knowledge
while maintaining the logical consistency of the
knowledge model.
ACKNOWLEDGEMENTS
This research has been partially supported by the
Innovative Economy Operational Programme EU-
founded project (UDA-POIG.01.03.01-12-163/08-
01).
REFERENCES
Ligeza, A. (2006). Logical Foundations for Rule-Based
Systems. Berlin, Heidelberg: Springer-Verlag.
Maciol, A. (2008). An application of rule-based tool in
attributive logic for business. Expert Systems with
Applications. 34, 1825–1836.
ICSOFT 2010 - 5th International Conference on Software and Data Technologies
86