Research And Implementation Of Product Configuration For Port-and-Connector Model

Mass customization is becoming a important direction of manufacturing corporations under the pressure of changeful globality market. One of key enablers for mass customization is the Product Configurators.In 1982, Digital Equipment Corporation used the R1/XCON system to configure computer systems. Since then, there have many configuration representation and reasoning methods, such as the rule-based reasoning, the logic-based approach, the resource-based approach, the constraint-based approach and the case-based reasoning. These methods are not good for representing the production constraint, production change and customer requirments. The custom-built configurators have not the uniform knowledge representation and reasoning method, then it is difficult to share and expand the configurator's knowledge base. This paper researched the configuration knowledge representation, configuration reasoning, configurator framework and the integration of configurator.First this paper researched the configuration representation. Ontologies is the basis of knowledge representation and sharing. The existing Configuration-Design ontologies of Stanford Knowledge Systems Laboratory do not refer the resource and sharing relation between the components. In the Soininen's configuration ontologies and the configuration ontologies of Klagenfurt University Computer Science and Manufacturing Group, it is difficult to distinguish the function and the component. Then we built the new configuration ontologies for Port-and-Connector model. Now the building of ontologies is subjective. Therefor this paper introduced the Phrase Frequency Analysis(PFA), Matrix Analysis(MA), Entity-Relation Graph(E-R Graph) during the ontologies design. These methods are repeatable and mensurable. This configuration ontologies for port-and-connector model included many concepts that were not mentioned before, such as "Configuration model", "Specification", "Component collection", "substitute" etc.This paper researched the configuration representation based on the cofiguration ontologies for port-and-connector model. Kowledge is considered to be the relation between concepts. So it is natural to represent knowledge with semantic web. We studied the Sematic Web project from W3C. The Knowledge represented by RDF (Resource Description Framework) can be understanded by not only the people but also the computer. Then knowledge is easy to be communicated between softwares. The pure text XML is convenient to be communicated and shared between different operation systems in Internet. This paper defined the XML tags of configurationontologies for port-and-connector model with OWL Web Ontology Language XML Presentation Syntax (OWLX). Based on these XML tags, this paper represented product knowledge with expanded RDF/XML (Resource Description Framework / extensible Markup Language). And gave a expanded RDF/XML representation example of a configurable production. This paper also expanded RDF/XML to represent fuzzy knowledge.Configuration reasoning can be regarded as searching the component collections that match configuration requirements from finite possible component collections. Existing configuration reasoning algorithms can not give all results once while solving configuration problem in under-constrainted situation, these algorithms can not give the acceptable result in over-constrainted situation. Therefor, this paper regarded the product configuration as combinatorial optimization problem, and solved configuration problem with genetic anealing algorithms. The test results were satisfied in this paper.Distributed configuration problem usually is regarded as distributed constraint satisfaction problem (DCSP). Existing DCSP algorithms can not give satisfied result in over-constrainted situation or under-constrainted situation. Then we regarded distributed configuration problem as distributed combinatorial optimization problem. This paper gave distributed genetic anealing algorithms(DGAA) to solve distributed configuration problem. In DGAA, Agent communicated selec...