Data-Driven and Knowledge-Guided Approach for NC Machining Process Planning

Abstract

With the wide applications of computer-aided technologies in industries, plenty of process data composed of 3D CAD models and CAM models are increasingly generated, which are the direct and effective carriers of knowledge, intelligence, and experience of skilled designers. However, existing methods cannot effectively learn and mine the implicit and explicit knowledge embedded in process data, and the process scheme generated is difficult to satisfy the manufacturing logics of a part in an industry. In this paper, a data-driven and knowledge-guided approach for NC machining process planning is proposed. First, a multi-level structured NC machining process model based on sub-machining regions is introduced to represent the relations between the 3D CAD model and the CAM model in process data. Then, an attention-based working step sequence generation model is proposed to predict the candidate working step and its associated sub-machining regions. Moreover, a context-free grammar describing the semantic and temporal relationships between working steps is extracted to construct a process knowledge And-Or graph with hierarchy and compositionality (PK-AOG). Finally, guided by PK-AOG, an NC machining process scheme generation method based on grammar parser is proposed to find a joint optimal solution according to the probabilities of the candidate working steps iteratively. A prototype system based on CATIA has been developed to verify the effectiveness of the proposed method.