View in EDS

Knowledge-guided deep reinforcement learning framework for feature machining step sequencing.

Saved in:
Bibliographic Details
Title: Knowledge-guided deep reinforcement learning framework for feature machining step sequencing.
Authors: Wang, Pengyu1 (AUTHOR), You, Youpeng2 (AUTHOR), Liu, Weichao3 (AUTHOR), Ding, Dawei1 (AUTHOR), Xu, Fengyu1 (AUTHOR) 13915964356@njupt.edu.cn
Source: International Journal of Advanced Manufacturing Technology. Oct2025, Vol. 140 Issue 9/10, p5617-5636. 20p.
Subjects: Computer-aided process planning, Machining, Feature selection, Expertise, Reinforcement learning, Economic efficiency
Abstract: Machining sequencing is a critical step in computer-aided process planning (CAPP) for numerical control (NC) machining, directly impacting efficiency and cost. Consequently, the machining sequencing method is crucial for practical manufacturing processes. However, existing sequencing methods face two main limitations: (1) they often treat machining features (MFs) as the smallest unit, neglecting that complex features require decomposition into sub-features and multiple feature machining steps (FMSs); (2) while exact methods are very time-consuming, prevailing approximate methods suffer from inconsistent performance, repeated iterative computations, and an inability to learn from experience. To address these challenges, this paper proposes a knowledge-guided deep reinforcement learning (DRL) framework for FMS-level sequencing. The method incorporates dynamic selection masks to enforce machining constraints and a process knowledge-based cost model to ensure efficient convergence. Leveraging the sequencing experience acquired during the learning process, this method can solve the machining sequencing problem within an acceptable computation time. Experiments demonstrate that the proposed method outperforms existing methods in both total cost and computation time, enabling more efficient machining of complex parts. [ABSTRACT FROM AUTHOR]
Copyright of International Journal of Advanced Manufacturing Technology is the property of Springer Nature and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database: Engineering Source
Description
Abstract:Machining sequencing is a critical step in computer-aided process planning (CAPP) for numerical control (NC) machining, directly impacting efficiency and cost. Consequently, the machining sequencing method is crucial for practical manufacturing processes. However, existing sequencing methods face two main limitations: (1) they often treat machining features (MFs) as the smallest unit, neglecting that complex features require decomposition into sub-features and multiple feature machining steps (FMSs); (2) while exact methods are very time-consuming, prevailing approximate methods suffer from inconsistent performance, repeated iterative computations, and an inability to learn from experience. To address these challenges, this paper proposes a knowledge-guided deep reinforcement learning (DRL) framework for FMS-level sequencing. The method incorporates dynamic selection masks to enforce machining constraints and a process knowledge-based cost model to ensure efficient convergence. Leveraging the sequencing experience acquired during the learning process, this method can solve the machining sequencing problem within an acceptable computation time. Experiments demonstrate that the proposed method outperforms existing methods in both total cost and computation time, enabling more efficient machining of complex parts. [ABSTRACT FROM AUTHOR]
ISSN:02683768
DOI:10.1007/s00170-025-16606-4