Minna Lanz: Logical and Semantic Foundations of Knowledge Representation for Assembly
and Manufacturing Processes
Doctoral dissertation Issue date: 2010-07-06 ISBN: 978-952-15-2393-9 ISBN (PDF): 978-952-152410-3
Partly due to the increase in global competition, the nature of manufacturing paradigms in Europe
has been undergoing continuous change in recent decades. Pressures from the market are forcing
companies to evolve into new entities of development and innovation. The trend of outsourcing to
lower-wage countries, global market situation and geographically dispersed operations are imposing
new challenges on manufacturing companies to become collaborative elements in the scheme of the
supply chain. It is no new news to anyone that problems are arising when manufacturing experts are
situated in Europe, design teams are spread around the globe and the operations are performed
somewhere in Asia.
In order to keep up with the competitors the knowledge of the product itself, possible production
facilities, and suitable processes has to be shared amongst the teams with ever-increasing speed.The
emergence of highly computerised design and control systems is changing the fundamental
assumptions of how product and production should be planned and controlled. Centralised control
and decision-making systems based on hierarchical control structures and tightly-coupled system
interfaces are giving way to globally distributed net- works that both enable and require localised
and fast adaptation to changes.
There are two problems identified in this thesis that are closely related to the decision-making
systems and knowledge share among those. The first problem is the amount of information stored
inside companies information systems and its lack of meaning. Especially in the design of products
and production systems the proprietary design models are converted from one model to another
numerous times. Because of the lack of interoperability between systems, only the basic geometry is
exchanged and the rest of the information may or may not be transferred manually to the new
design. The discontinuous knowledge flow often results in several different and contradictory
models of the same design. This also leads to a situation where the re-use of knowledge becomes a
slightly hazardous task, since before the existing knowledge is used someone must verify that all of
the disconnected pieces are valid.
The second problem emerges when the product knowledge is used as a set of requirements for the
design of the production system. In order for the production to be launched as soon as possible, the
production plan is usually verified with a decision support system. However, since decision support
systems, either on the design level or the actual production control level, rely on the input
knowledge, the lack of it naturally undermines the reliability of the decision making. This is one of
the greatest technological barriers to the implementation of the greatly desired complex and
adaptive production environment, since without the meaning of the models that are used the
reasoning becomes almost impossible.
As several other challenges to implementing an integrated collaborative and dynamically adaptive
production environment exist, only two challenges are taken into focus in this thesis. Both
challenges have multiple facets, but they do share a common need: a formal rigorous knowledge
representation. The approach introduced here aims to create a common knowledge representation
between the product, process, and system domains. The aim is not to model everything that can be
included in a knowledge representation but to model the common core components and their
relations.
The approach in this thesis starts from a feature-based modelling and analysis method and utilises
the detailed product knowledge as the core of the knowledge representation. The process-, and
system-level knowledge is represented with the necessary detail and the amount of it that is needed
that are suit- able for a given situation. The thesis introduces a knowledge representation (KR) for
combining design information from several different sources into one reference architecture, which
can be accessed via a common interface. The chosen approach is tested by means of three separate
cases, which validate the KR from different perspectives. The first case study will evaluate the
feasibility of utilising geometric and non-geometric features as the prerequisites for process
modelling. The second case study evaluates the integrity and expressiveness of the KR by
integrating the input from several commercial clients into one representation. The third case study
evaluates the KR in a real production environment, where the KR will provide the knowledge for a
holonic manufacturing system and save the events and device parameters into the history of the
product that is realised after the operations are completed.
The dissertation can be obtained here; http://URN.fi/URN:NBN:fi:tty-201006291195