HIERARCHICAL KNOWLEDGE-BASED PROCESS PLANNING IN MANUFACTURING

摘要

Artificial intelligence planning methods haven't been used until recently to address the problem of computer-aided process planning (CAPP) in manufacturing in its entirety. They were simply not developed enough to tackle real-world problems of that complexity. In the paper we show that with so-called Hierarchical Task Networks, a recently matured general-purpose domain-independent planning method, we could model the planning process itself, represent and utilize different kinds of technological knowledge and keep in check the complexity of the plan generation process. To this aim the planner was extended with search methods for finding the best plans and supporting mixed-initiative, interactive planning. The proposed CAPP system deals with geometry analysis, setup planning, selection and ordering of machining operations and the assignment of resources. The first experiments with prismatic and rotational parts show considerable merit of the approach.