Fault Automatic Identification and Personnel Allocation in the Production Line

DOI: https://doi.org/10.62517/jbdc.202501221

Author(s)

Baxian Chen, Yurong Wu, Shan He, Yisha Liu

Affiliation(s)

School of Artificial Intelligence, Guangzhou Huashang College, Zengcheng, Guangzhou, China

Abstract

This paper focuses on the issues of fault automatic identification and personnel allocation in industrial production lines. First, comprehensive data preprocessing is conducted, involving meticulous data cleaning to remove noise, strategic handling of 3 missing values through interpolation or deletion, and rigorous feature selection based on correlation analysis. To tackle the problem of sample imbalance, both up-sampling (duplicating minority samples) and down-sampling (reducing majority samples) are applied, significantly improving the training effect of the model. A fault - alarm model is then built using key characteristics like equipment operation states and process parameters, enabling timely prediction of potential faults. Meanwhile, advanced machine - learning models, such as random forest and decision trees, are utilized to analyze the intricate relationship between workers' years of service and production efficiency, helping formulate an optimal shift-scheduling plan that balances operator experience and shift workload. The proposed methods have been proven to effectively enhance production line stability and operational efficiency, offering practical and innovative solutions for industrial manufacturing and solid scientific decision - making support for production management and human resource allocation.

Keywords

Difference; Linear Regression; Up-sampling and Down-sampling; Decision Tree; RF Model; SHAP Model

References

[1] Fu, L. Y., Li, H., & Yu, D. (2012). A review of the research on random forests in machine learning. Journal of Frontiers of Computer Science and Technology, 6(5), 457-473. [2] Li, J. H., Wang, Y., & Hu, Y. (2017). A review of the random forest algorithm. Computer & Digital Engineering, 45(11), 2432-2435. [3] Lundberg, S. M., Erion, G. G., & Lee, S. I. (2020). Consistent individualized feature attribution for tree ensembles. arXiv preprint arXiv:1802.03888. [4] Lundberg, S. M., & Lee, S. I. (2020). Explainable AI for trees: From local explanations to global understanding. Nature Machine Intelligence, 2(1), 56-67. [5] Lin, X. T. Machine Learning Foundations. Tsinghua University Press, 2015. [6] Wu, X. Z. SPSS Modeling and Application: Multivariate Statistical Analysis Practice. Tsinghua University Press, 2017. [7] Zhou, X., et al. (2024). Ensemble learning for industrial fault diagnosis: A comparative study. Journal of Manufacturing Systems, 68, 456-468. [8] Molnar, C. (2021). Interpretable Machine Learning. Springer. [9] Gupta, R., & Sharma, S. (2023). Optimal shift scheduling in manufacturing using linear programming: A case study. International Journal of Production Economics, 258, 108291. [10] Li, W., et al. (2022). Data-driven production optimization: Trends and challenges. IEEE Transactions on Industrial Informatics, 18(9), 6321-6330. [11] Wang, Y., et al. (2024). Reinforcement learning for dynamic production scheduling: A survey. Computers & Operations Research, 159, 105