A novel method for intersecting machining feature segmentation via deep reinforcement learning

Machining feature segmentation is a primary task in machining feature recognition, as it directly impacts downstream activities such as feature type identification and process planning. However, the intersecting and overlapping of multiple machining features within intersecting machining features disrupt the original geometries and topologies, resulting in complex structures and varied forms. Therefore, segmenting intersecting machining features is highly challenging. Existing methods for intersecting machining feature segmentation often suffer from issues related to poor effectiveness and the difficulty of applying segmentation results to subsequent process planning. To address these concerns, this paper presents a novel framework for segmenting intersecting machining features using deep reinforcement learning. The framework takes intersecting machining features represented by attributed adjacency graphs as input and generates isolated machining features as output. Leveraging the robust feature representation, decision-making, and automatic learning capabilities of deep reinforcement learning, this framework enhances the effectiveness of intersecting machining feature segmentation. Moreover, the framework takes into account the machining process when designing the reward mechanism, thereby ensuring that the results are beneficial to subsequent process planning. In the experimental studies, plenty of intersecting machining features from 2.5D computer numerical control parts are used as examples to verify the feasibility and effectiveness of the proposed approach. The experimental results demonstrate that the proposed approach successfully addresses some existing challenges faced by several state-of-the-art methods in intersecting machining feature segmentation. The findings highlight the robustness of the proposed approach and its potential to address the challenges associated with intersecting machining feature segmentation. Consequently, this study contributes to the advancement of machining feature recognition and holds practical implications for process planning.