Concurrent Engineering Workflow

CONFLOW - CP96-243

Keywords: workflow design, intelligent workflow management, concurrent engineering, autonomous knowledge acquisition, inductive learning

Start Date: open / Duration: 36 months

[ participants / contact]

Objectivies and Approach

The project has two major objectives:

In the engineering domain the product structure fixes the tasks which have to be performed to achieve a certain goal. The product components (objects), their characteristics (attributes) and their relationships within assemblies (interference) determine the planning and operational processes needed to manufacture an optimal product in terms of price, time to market and quality. Therefore, reference processes have to be defined at the component level that describe which workflows (activities and work items) within the product and process definition have to be conducted according to the component attributes. At the assembly level, the relationship between the components has to be represented in order to assess the interdependence between the reference processes at the component level. The process definition module of the workflow management system has to be enabled to connect the reference processes with the product structure. Modifications in the product structure must automatically result in adapted process definitions. According to the resulting design, the workflow management system executes the workflow, manages the engineering work units involved, provides them with the required data and establishes efficient concurrent engineering.

In order to achieve a dynamic link between the product structure and the workflow design, rules are required which adapt the workflows to particular component attributes and interference. The manual definition of these rules is a very tedious work and requires an enormous effort. In addition, conventionally established rule bases are very difficult to manage. Therefore, artificial intelligence techniques that allow the automatic set up and administration of rule bases via autonomous knowledge acquisition will be applied. Inductive learning algorithms are able to learn from facts represented in the form of cases and automatically extract rules from the cases representation. From "Workflow Cases" consisting of product structure characteristics on the one hand, and workflow definitions on the other, the algorithms are able to extract rules which establish the mentioned dynamic link and help to manage recurrent engineering tasks.


Technical University Clausthal-Zellerfel, D

EU Partners

Technical University Clausthal-Zellerfel, D
University of Cardiff, UWCC, UK
Mummert, D

Non-EU Partners

University of Rousse, BG
Technical University, Budapest, HU


Mr. Peter Kickartz
Tel: +49 53 23 720
Fax: +49 53 23 723 500

INCO synopses home page INCO cooperation index INCO keyword index
INCO acronym index INCO number index INCO Projects index
synopses home page all keywords index all acronyms index all numbers index

CONFLOW - CP96-243, May 1997

please address enquiries to the ESPRIT Information Desk

html version of synopsis by Nick Cook