{"title":"ML2MAS: a multi-agent reinforcement learning and BNNs-GAN integration framework for smart manufacturing optimization","authors":"Shadia Yahya Baroud, Nor Adnan Yahaya","doi":"10.1016/j.susoc.2025.07.003","DOIUrl":null,"url":null,"abstract":"<div><div>The rise of smart manufacturing, driven by advanced technological enablers, has transformed traditional production processes, creating new opportunities for operational efficiency and predictive maintenance (PdM). This study introduces ML2MAS, an innovative framework that integrates machine learning (ML) models within a multi-agent reinforcement learning (MAS-RL) system to enhance PdM capabilities in manufacturing environments. ML2MAS combines Bayesian Neural Networks (BNNs) for handling high-dimensional data and quantifying prediction uncertainty alongside Generative Adversarial Networks (GANs) for synthetic data generation, addressing the challenge of limited labelled datasets and improving model robustness. Integrating these components within MAS enables PdM to make decentralized, real-time decisions. Empirical results from a case study demonstrate substantial improvements, achieving a 99 % F1-score in predictive accuracy and notable reductions in maintenance costs. The proposed ML2MAS framework ensures a cohesive, adaptive PdM solution and contributes to more sustainable and efficient manufacturing operations.</div></div>","PeriodicalId":101201,"journal":{"name":"Sustainable Operations and Computers","volume":"6 ","pages":"Pages 217-228"},"PeriodicalIF":0.0000,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Sustainable Operations and Computers","FirstCategoryId":"1085","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S2666412725000133","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0
Abstract
The rise of smart manufacturing, driven by advanced technological enablers, has transformed traditional production processes, creating new opportunities for operational efficiency and predictive maintenance (PdM). This study introduces ML2MAS, an innovative framework that integrates machine learning (ML) models within a multi-agent reinforcement learning (MAS-RL) system to enhance PdM capabilities in manufacturing environments. ML2MAS combines Bayesian Neural Networks (BNNs) for handling high-dimensional data and quantifying prediction uncertainty alongside Generative Adversarial Networks (GANs) for synthetic data generation, addressing the challenge of limited labelled datasets and improving model robustness. Integrating these components within MAS enables PdM to make decentralized, real-time decisions. Empirical results from a case study demonstrate substantial improvements, achieving a 99 % F1-score in predictive accuracy and notable reductions in maintenance costs. The proposed ML2MAS framework ensures a cohesive, adaptive PdM solution and contributes to more sustainable and efficient manufacturing operations.