Abstract
The main concept of the authors is to place Reinforcement Learning (RL) into various fields of manufacturing. As one of the first implementations, RL for Statistical Process Control (SPC) in production is introduced in the paper; it is a promising approach owing to its adaptability and the continuous ability to perform. The widely used Q-Table method was applied for get more stable, predictable, and easy to overview results. Therefore, quantization of the values of the time series to stripes inside the control chart was introduced. Detailed elements of the production environment simulation are described and its interaction with the reinforcement learning agent are detailed. Beyond the working concept for adapting RL into SPC in manufacturing, some novel RL extensions are also described, like the epsilon self-control of exploration–exploitation ratio, Reusing Window (RW) and the Measurement Window (MW). In the production related transformation, the main aim of the agent is to optimize the production cost while keeping the ratio of good products on a high level as well. Finally, industrial testing and validation is described that proved the applicability of the proposed concept.
Highlights
Artificial Intelligence (AI) and Machine Learning (ML) approaches are spreading across all areas in our live, it is valid for technical fields, e.g., for the manufacturing sector as well
A significant advantage of this concept was that every component of the reinforcement learning, and its environment was connected to a real application, and real measurements were feed-back to the simulation model and updated the knowledge base
Beyond the rare and particular cases presented as process control in the given industrial assignments, the current paper proposes a generalized, widely applicable Reinforcement Learning for Statistical Process Control (RL for SPC) framework to keep the production between the prescribed Lower Tolerance Limit (LTL) and Upper Tolerance Limit (UTL) of the related Statistical Control Chart (SCC), important novel ex tensions are introduced as well
Summary
Artificial Intelligence (AI) and Machine Learning (ML) approaches are spreading across all areas in our live, it is valid for technical fields, e.g., for the manufacturing sector as well. The frequently arising, novel AI and ML techniques have to be continuously adopted to the given domain to achieve the best match This mission is valid to manufacturing, while the well-known Industry 4.0 global initiative (called as Industrial Internet or Cyber Physical Production Systems (CPPS)) supports, facilitates, incorporates these directions, the actual situation in this sector is quite promising. As described in the paragraph, the reviewing the scientific literature mirrors that the domain specific adaptation of reinforcement learning to various production fields concentrates mainly, only to production scheduling and robotics. This state-of-the-art status provoked the moti vation to extend and adapt reinforcement learning to further potential fields of manufacturing.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
