Research on Dehumidification Efficiency Prediction of Transformer Breather Silica Gel Air-Drying and Stirring System Based on Improved Random Forest Algorithm

DOI: https://doi.org/10.62517/jbdc.202501427

Author(s)

Guangyao Wang*, Chaofei Yan, Zhikang Sun, Yanmei Cao, Kaixin Ding, Pinyi Zhao

Affiliation(s)

State Grid Jia Xian Electric Power Supply Company of Henan Electric Power Company, Pingdingshan, China *Corresponding Author

Abstract

Transformer breathers play a critical role in maintaining the insulation performance of transformers by preventing moisture ingress, and the dehumidification efficiency of silica gel in the breather directly affects the stable operation of the entire transformer system. However, the traditional method of evaluating dehumidification efficiency through repeated experiments is time-consuming and costly, making it difficult to meet the real-time monitoring needs of transformer operation. To address this problem, this study proposes an improved random forest (RF) algorithm for predicting the dehumidification efficiency of a transformer breather silica gel air-drying and stirring system. First, the mutual information method is introduced to screen the core influencing factors (including silica gel initial humidity, stirring speed, microwave heating temperature, and air-drying airflow intensity) from multiple operation parameters, eliminating redundant features and reducing model complexity. Then, the hyperparameters of the RF algorithm are optimized using the grid search method to enhance the model's prediction accuracy. Experimental results show that compared with the traditional RF algorithm and XG Boost algorithm, the improved RF algorithm proposed in this study reduces the prediction error by 8.2% and 5.6% respectively, and the prediction speed is increased by 12.5%. This model can quickly and accurately predict the dehumidification efficiency of the system, providing a reliable decision-making basis for the intelligent adjustment and optimal operation of the transformer breather silica gel air-drying and stirring system.

Keywords

Transformer Breather; Silica Gel Dehumidification; Random Forest Algorithm; Efficiency Prediction; Feature Selection

References

[1] B. Jiang, Z. Zheng, C. Pi, J. Long, and Z. Yue, "A lightweight three-phase induction motor fault detection method based on thermal imaging," IEEE Access, vol. 13, pp. 161909-161922, 2025. [2] T. Islam and S. Kumar, "Testing and validation of smart and intelligent sensors," in Switchgear Design, Operation, and Maintenance Using Industry Standards. Elsevier, 2025, pp. 113-133. [3] A. P. Dolin and M. M. Kipriyanova, "Major defects and diagnostics of solid insulation of power transformers," Power Technol. Eng., vol. 56, no. 4, pp. 612-618, 2022. [4] M. Nuruzzaman, G. Q. Limon, A. R. Chowdhury, M. S. Hossain, and S. I. Moin, "Predictive maintenance in power transformers: A systematic review of AI and IOT applications," ASRC Procedia: Global Perspect. Sci. Scholarsh., vol. 1, no. 1, pp. 34-47, 2025. [5] S. Veerappan, "Digital twin modeling for predictive maintenance in large-scale power transformers," Nat. J. Elect. Mach. Power Convers., pp. 39-44, 2025. [6] M. Mowbray, M. Vallerio, C. Perez-Galvan, I. D. L. Bogle, and F. Zhang, "Industrial data science–a review of machine learning applications for chemical and process industries," React. Chem. Eng., vol. 7, no. 7, pp. 1471-1509, 2022. [7] Y. Yang and H. Wang, "Random forest-based machine failure prediction: A performance comparison," Appl. Sci., vol. 15, no. 16, p. 8841, 2025. [8] B. G. Mejbel, S. A. Sarow, M. T. Al-Sharify, A. A. Al-Amiery, and K. A. Kadhim, "A data fusion analysis and random forest learning for enhanced control and failure diagnosis in rotating machinery," J. Fail. Anal. Prev., vol. 24, no. 6, pp. 2979-2989, 2024. [9] L. Liao, H. Li, W. Shang, Y. Lyu, and M. R. Lyu, "An empirical study of the impact of hyperparameter tuning and model optimization on the performance properties of deep neural networks," ACM Trans. Softw. Eng. Methodol., vol. 31, no. 3, pp. 1-40, Jul. 2022. [10] H. Zhang, R. Li, Q. Du, Y. Chen, and L. Wu, "Decision-focused learning for power system decision-making under uncertainty," IEEE Trans. Power Syst., early access, 2025. [11] K. Cao, T. Zhang, and J. Huang, "Advanced hybrid LSTM-transformer architecture for real-time multi-task prediction in engineering systems," Sci. Rep., vol. 14, no. 1, p. 4890, 2024. [12] M. Paniagua-Gómez and M. Fernandez-Carmona, "Trends and challenges in real-time stress detection and modulation: The role of the IoT and artificial intelligence," Electronics, vol. 14, no. 13, p. 2581, 2025. [13] H. A. Salman, A. Kalakech, and A. Steiti, "Random forest algorithm overview," Babylonian J. Mach. Learn., vol. 2024, pp. 69-79, 2024. [14] I. D. Mienye and Y. Sun, "A survey of ensemble learning: Concepts, algorithms, applications, and prospects," IEEE Access, vol. 10, pp. 99129-99149, 2022. [15] J. M. Kernbach and V. E. Staartjes, "Foundations of machine learning-based clinical prediction modeling: Part II-Generalization and overfitting," in Machine Learning in Clinical Neuroscience: Foundations and Applications. Cham, Switzerland: Springer, 2021, pp. 15-21. [16] C. F. G. D. Santos and J. P. Papa, "Avoiding overfitting: A survey on regularization methods for convolutional neural networks," ACM Comput. Surv., vol. 54, no. 10s, pp. 1-25, Sep. 2022. [17] R. Zemouri. "Power transformer prognostics and health management using machine learning: A review and future directions," Machines, vol. 13, no. 2, p. 125, 2025.