knowledge structure representing the specified engineering
domain
[7]
and to support powerful tools that enable inference,
querying, and information retrieval, OWL (Web Ontology
Language) is used to conceptualize engineering design
semantics so as to share knowledge over the web
[8]
, and to
overcome a lack of interoperability by creating a collaborative
environment
[9]
.
There is a lot of work on ontology and description logics to
represent a product model from CAD systems. An early study
by Kim et al.
[10]
suggests using the ontology-based assembly
model to express assembly features and relations between
features, and using OWL triples and SWRL
[11]
(Semantic Web
Rule Language) rules to explicitly represent assembly
constraints. Moreover, to develop a consistent formal model for
product assemblies, the work of Fiorentini et al.
[12]
creates a
semantic model including non-geometry concepts from the
Core Product Model (CPM)
[13]
and the Open Assembly Model
(OAM)
[14]
developed at NIST (National Institute of Standards
and Technology). They implement the model in OWL-DL 1.0
[15]
and outline a method for evaluating the appropriate level of
expressiveness to capture both the information content and the
abstraction principles
[16]
. Besides the above work, there is also
some research on how to use ontology to capture and represent
design semantics for engineering design
[17-19]
.
Furthermore, ontology is identified as one of the promising
solutions to address the issue of product family design related
information management. Kim et al.
[20]
propose a series of
related work. In their paper, they suggest a semantically
annotated multi-faceted product family ontology (MFPFO) that
is able to comprehensively model the relationships between a
product family ontology with other related ontologies of
interest in an open and scalable manner. Moreover, they
propose the methodology for building a semantically annotated
MFPFO
[21]
and also suggest a framework of faceted
information search and retrieval for product family design
[22]
.
Then, product analysis and variant derivation based on faceted
search is presented
[23]
, while a new commonality metric and a
faceted platform selection approach is presented
[24]
. Their
work has demonstrated the effectiveness of ontology for
information representation and automatic selection.
In this paper, an automatic selection oriented ontology
model of control valves based on CPM will be constructed. It
will demonstrate how the ontology model is designed to
integrate structured and unstructured knowledge in the
selection process and to map application requirements to
product features so as to effectively support the automatic
selection of control valves. It will also show an ontology
construction methodology based on the spiral model. More
details are explained in Sections 2 to 4. Section 5 demonstrates
the experiments results. Finally, Section 6 concludes this paper
and discusses our future work.
2. DEMAND ANALYSIS OF ONTOLOGY MODELING
FOR AUTOMATIC SELECTION
In essence, the selection is a design behavior of selecting
optimized products assembly scheme according to users’
requirements, and it corresponds to a conversion process from
the knowledge of application requirements to product features.
Furthermore, the automatic selection specifically refers to
computer-aided operations instead of human expert operations
in all aspects of the selection process including analyzing,
judging, reasoning, decision-making and so on. Therefore, to
support automatic selection, modeling of application
requirements, product features and their associations is
essential.
2.1 Application Requirements Knowledge
Application requirements knowledge is the knowledge
about requirements provided by the users of control valves. It
mainly includes the following two categories:
1) Selection requirement aimed at control valve. In this
situation, the requirements are directly aimed at specific
attributes of control valves. These can be function,
material, geometry and any other related attributes of the
required control valve. For example, to select a control
valve for the heat exchange system of air separation plant,
the user provides a series of attributes of the required
control valve: nominal pressure is 12.4Mpa, working
temperature is -129℃, rated CV is 11.5 and nominal
diameter is 20mm. We define this kind of selection
requirements as the selection requirement aimed at control
valve.
2) Selection requirement oriented to application
environment. In this situation, the requirements are
oriented to application environment of the required
control valves. They are always vague and qualitative. For
example, the application environment of air separation
plant has the characteristics of high-pressure and low-
temperature, so the user requires control valves suitable
for the conditions. Here, no series, no models or any
specific technical parameters, but only its application
environment, is provided. We define the selection
requirements as the selection requirement oriented
application environment.
2.2 Product Features Knowledge
Product features knowledge is the knowledge about their
own characteristics of products provided by the suppliers of
control valves. It can be described from the following three
aspects: ① product category, aiming to record the category of
control valves; ② related attributes, aiming to record attributes
related to control valves, including both quantitative attributes
and qualitative attributes, such as function, material, geometry
and performance; ③ incidence relation of product and its
attributes. This records the specific attributes of the given type
of control valves: for example, the control valve ALS-XX has
the attribute of low temperature.