A prototype system has been implemented using the object- oriented expert system shell CLIPS (C Language Integrated Pro- duction System) [1], which is interfaced with a parametric- and feature-based CAD system, Solid Edge through C++. An ex- ample is provided to demonstrate our approach and to show its effectiveness in stamping process planning.

tage achieved by the system is the rapid generation of the most efficient strip layouts. Researchers at the University of Liverpool have worked on design automation for progressive piercing and blanking dies [10, 11]. Their work is based on applying a coding technique to characterize the stamped part geometric features, which is subsequently used to generate the type and layout of the die punches, and then develop the strip layout automatically. Researchers at Huazhong University of Science and Technol- ogy, China, have developed an intelligent progressive die design system, HPRODIE [12]. With feature mapping, rule-based rea- soning and case-based reasoning techniques, most of the design processes including strip layout design can be carried out  auto-matically. Researchers at Pusan National University, Korea, have

2 Related work

Research in the computer-aided stamping process planning has been widely reported since the 1970s. The advantages of auto- mated process planning are productivity improvements, cost re- ductions, and design automation.

From the mid  1970s  to  mid  1980s,  the  first  generation of CAD/CAM systems for progressive die design were de- veloped [2–5], though few of them are based on AI techniques. These early systems are characterized by basic computer graph- ics facilities, standardization of die components, and standard- ization of design procedures. They reduced design and drafting lead time. However, as these systems represent design know-how in the form of conventional procedural programming languages, only generation of the die part list and drafting of the assembly and part drawings are executed using computers. The designer still needs to decide most of the important decisions interactively, including strip and die layouts.

Since the late 1980s, significant efforts have been made by worldwide researchers to integrate a wide variety of AI and traditional CAD approaches to develop dedicated progressive die design automation systems, including strip layout design automation.

Knowledge-based approach is a popular AI technique that has been used in intelligent stamping process planning and die design system. For example, researchers at the University of Massachusetts, USA have described a knowledge-based sys- tem for design of progressive stamping dies for a simple hinge part [6]. The system generates the flat pattern geometry and de- velops a strip layout automatically. Researchers at the National University of Singapore have been developing an intelligent pro- gressive die (IPD) design system since the late 1980s. They used feature modeling and rule-based approach to realize automatic punch shape selection, strip layout development, and 3-D die configuration [7, 8]. Based on a feature-relationship tree that de- scribes the stamped metal part and its topological information, model-based reasoning and spatial reasoning techniques have been employed to reason out certain stamping processes and guide the overall planning process to develop the strip layout automatically. Researchers at the Indian Institute of Technology have developed a computer-aided die design system, CADDS, for sheet-metal blanks [9], based on heuristic rule-based reason- ing and parametric programming techniques. The greatest advan-

developed a compact computer-aided process planning (CAPP) system for progressive die design [13]. Based on production rules, the work is capable of carrying out an intelligent stamp- ing process planning work with automatic development of blank layout, strip layout and die layout.

Though knowledge-based systems have achieved a lot of suc- cess in stamping process planning, most of the intelligent pro- gressive die design automation prototypes reviewed above are rather restricted to specific application domains, or still need considerable interactive input from experienced designers to de- velop strip layouts. This is because they still inherit the disadvan- tages of the conventional architecture of knowledge-based expert systems, which are incapable of managing heterogeneous KSs effectively.