Knowledge-based expert support in an assembly-oriented CAD environment
Assembly-oriented design offers great potential for product rationalisation, increasing productivity and reducing lead time and cost. It results in simpler and more reliable products which are less expensive to assemble and manufacture. To facilitate assembly-oriented design, an assembly-oriented CAD environment is needed to incorporate Design for Assembly (DFA) evaluation from an early design stage. Assembly planning should also be integrated to support the DFA evaluation. This thesis reports the results of research towards supporting such an assembly-oriented CAD environment. A novel approach has been used to deploy an Expert Assembler to support proactive DFA evaluation and assembly sequence definition. This is particularly useful, as designers are rarely if ever assembly experts. Based on the fact that there are several areas needing expert support in this assembly-oriented CAD environment, but that different areas have very different requirements and different knowledge is involved, the Expert Assembler deployed contains several separated modules. Each module is an expert agent devised to tackle a problem area that uses a suitable problem solving strategy, knowledge representation and reasoning method. This brings a number of advantages that are detailed in the thesis. The thesis presents systematical ideas for support proactive DFA, with the focus on support for part count reduction and assembly sequence generation. This is realised by three elements of the expert agents: Part Count Advisor, Starting Part Advisor, and Next Part Advisor. Part count reduction is usually based on dialogue with the user. There is little computational support for this issue in any of the DFA methodologies and related literature. This research fills the gap: it brings computational support for part count reduction from the early design stage. The work has also made new progress in assembly sequence generation. The Starting Part Advisor and the Next Part Advisor cooperate with each other and with the user to provide suggestions dynamically and transparently regarding base part and the most suitable next part selection in assembly sequence definition. Case studies were used to test the effectiveness of the Advisors.