EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

ML Framework for Predictive Hazard Monitoring in Smart Manufacturing Environments

Jahangir Shekh, Md Badsha Nuruzzaman Shahin, Md Jakaria Islam, Mehidi Hasan Suvo, Dipankar Nandy

International Journal of Scientific Research in Science, Engineering and Technology · 20.12.2024 · 11 (6) · 583-592

DOI: 10.32628/ijsrset2512553 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Journal of Scientific Research in Science, Engineering and Technology, online erschienen am 20.12.2024. Verfasst von Jahangir Shekh und 4 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Manufacturing facilities continue to face serious safety challenges despite regulations and training programs. Traditional approaches react to incidents after they occur rather than preventing them. This paper presents a machine learning framework designed to predict when hazardous conditions might emerge in manufacturing environments. The framework operates through five stages: data preprocessing, feature engineering, multi-model prediction, ensemble decision-making, and risk alerting. We engineered manufacturing-specific features including worker fatigue metrics, safety compliance scores, equipment maintenance indicators, and environmental risk factors. The system was evaluated on a synthetically generated dataset of 5,000 safety records modelled after seven manufacturing industries. Four machine learning algorithms were compared—Random Forest, XGBoost, Support Vector Machines, and Neural Networks—along with an ensemble approach. The Neural Network achieved the highest F1-score of 0.2717, while the ensemble achieved the best ROC-AUC of 0.5926. These modest results reflect the inherent difficulty of predicting rare, multifactorial safety events and indicate that further refinement of features, models, and data quality is needed before practical deployment. Composite risk scores, maintenance schedules, and facility characteristics were among the strongest predictive features.

Zitieren

APA

Shekh, J., Shahin, M. B. N., Islam, M. J., Suvo, M. H., & Nandy, D. (2024). ML Framework for Predictive Hazard Monitoring in Smart Manufacturing Environments. International Journal of Scientific Research in Science, Engineering and Technology, 11(6), 583-592. https://doi.org/10.32628/ijsrset2512553

Vancouver

Shekh J, Shahin MBN, Islam MJ, Suvo MH, Nandy D. ML Framework for Predictive Hazard Monitoring in Smart Manufacturing Environments. International Journal of Scientific Research in Science, Engineering and Technology. 2024;11(6):583-592. doi:10.32628/ijsrset2512553

BibTeX
@article{Shekh2024,
  author = {Shekh, Jahangir and Shahin, Md Badsha Nuruzzaman and Islam, Md Jakaria and Suvo, Mehidi Hasan and Nandy, Dipankar},
  title = {ML Framework for Predictive Hazard Monitoring in Smart Manufacturing Environments},
  journal = {International Journal of Scientific Research in Science, Engineering and Technology},
  year = {2024},
  volume = {11},
  number = {6},
  pages = {583--592},
  doi = {10.32628/ijsrset2512553}
}
Kostenlos starten →