在電力變壓器中，絕緣油主要用來絕緣和冷卻繞組與鐵心。當變
壓器發生故障時，壓力或溫度的變化會導致油中產生氣體。透過分析
油中的氣體，可以有效地檢測變壓器內部的狀態，並診斷可能的故障
。然而，傳統的油中氣體分析方法存在許多局限性，例如需要大量的
時間和成本、需要專業的技能和設備、受到環境因素的影響等等。因
此，本論文提出了一種基於集成學習的變壓器油中氣體狀態評估方法
，以提高檢測的效率和準確性。本研究運用了集成學習的方法，利用
多個模型進行診斷，以減少模型產生誤差。其中，隨機森林是最為常
用的一種模型。在數據蒐集和預處理方面，本研究選用了包括遺漏值
填充、樣本平衡等在內的一系列技術來提高模型的穩定性和準確性。
本論文還將有載分接頭切換器(OLTC)對變壓器的影響納入了考
慮，針對變壓器油中氣體檢測結果進行了進一步的分析。透過將具
OLTC 之變壓器與一般變壓器分開討論，從而更加準確地診斷可能的
故障。本論文使用了多個實驗來試驗該方法的有效性和可行性，結果
顯示其具有良好的油中氣體的狀態評估能力。最終本論文提出了一種
基於集成學習的變壓器油中氣體狀態評估方法，針對現有方法的局限
性進行了改進和優化，並將OLTC 的影響納入了考量，從而提高了檢
測的準確性和可靠性。

In power transformers, insulating oil is one of the important elements
used for insulation and cooling. When a transformer fails, changes in
pressure or temperature can cause Gases to form in the oil. By analyzing
the Gas in the oil, the internal health conditions of the transformer can be
effectively detected, and possible faults can be predicted. However,
traditional methods of analyzing Gas in oil have many limitations, such as
requiring a lot of time and cost, requiring professional skills and
equipment, and being influenced by environmental factors. Therefore, this
paper proposes a transformer oil Gas state evaluation method based on
ensemble learning to improve detection efficiency and accuracy. This study
uses ensemble learning to reduce model bias and variance by using
multiple models for prediction, with random forest being the most
commonly used model. In terms of data collection and preprocessing, a
series of techniques including missing value filling and sample balancing
were used to improve model stability and accuracy.
This paper also considers the impact of OLTC (on-load tap changer)
on transformers and conducts further analysis of the results of Gas
detection in transformer oil. By discussing OLTC separately from general
transformers, possible faults can be predicted more accurately. This paper
uses multiple experiments to verify the effectiveness and feasibility of the
proposed method, which shows good results. This paper proposes a
transformer oil Gas state evaluation method based on ensemble learning,
which improves and optimizes the limitations of existing methods and
considers the impact of OLTC, thereby improving the accuracy and
reliability of detection.

[1] IEEE Guide for the Interpretation of Gases Generated in Mineral Oil-Immersed Transformers, IEEE Std C57.104-2019, Nov. 2019.
[2] J. Li, J. Zhao, R. Zheng, and Y. Liu, “Fault diagnosis system of transformer based on GAS chromatography,” in Proc. 2006 International Conference on Machine Learning and Cybernetics, Dalian, China, Aug. 2006.
[3] T. C. Garg and E. Huber, “Online monitoring prevented catastrophic failure of a GSU transformer,” in Proc. Electrical Insulation Conference and Electrical Manufacturing and Coil Winding Technology Conference, Indianapolis, IN, USA, Sept. 2003.
[4] T. Yang and W. Chen, “Detection of dissolved gas in transformer oil based on all-optical photoacoustic spectroscopy,” in Proc. 2020 IEEE International Conference on High Voltage Engineering and Application (ICHVE), Beijing, China, Sept. 2020.
[5] D. Bhalla, R. K. Bansal, and H. O. Gupta, “Application of artificial intelligence techniques for dissolved gas analysis of transformers-a review,” International Scholarly and Scientific Research & Innovation, vol. 4, no. 2, pp. 261–269, 2010.
[6] Y. -C. Huang and H. -C. Sun, “Dissolved gas analysis of mineral oil for power transformer fault diagnosis using fuzzy logic,” IEEE Transactions on Dielectrics and Electrical Insulation, vol. 20, no. 3, pp. 974–981, Jun. 2013.
[7] N. A. Fauzi et al., "Fault prediction for power transformer using optical spectrum of transformer oil and data mining analysis,” IEEE Access, vol. 8, pp. 136374–136381, Jul. 2020.
[8] D. Saravanan, A. Hasan, A. Singh, H. Mansoor, and R. N. Shaw, “Fault prediction of transformer using machine learning and DGA,” in Proc. 2020 IEEE International Conference on Computing, Power and Communication Technologies (GUCON), Greater Noida, India, Oct. 2020.
[9] M. -T. Yang and L. -S. Hu, “Intelligent fault types diagnostic system for dissolved gas analysis of oil-immersed power transformer,” IEEE Transactions on Dielectrics and Electrical Insulation, vol. 20, no. 6, pp. 2317–2324, Dec. 2013.
[10] N. Poonnoy, C. Suwanasri, and T. Suwanasri, “Neural network approach to dissolved gas analysis for fault analysis in power transformers,” in Proc. 2022 International Electrical Engineering Congress (iEECON), Khon Kaen, Thailand, Mar. 2022.
[11] A. Lakehal, Z. Ghemari, and S. Saad, “Transformer fault diagnosis using dissolved gas analysis technology and Bayesian networks,” in Proc. 2015 4th International Conference on Systems and Control (ICSC), Sousse, Tunisia, Apr. 2015.
[12] S. A. Wani, S. A. Khan, Y. Vikram, C. Bhasker, S. Perneen, and D. Gupta, “An improved dissolved gas analysis technique for incipient fault detection,” in Proc. 2017 3rd International Conference on Condition Assessment Techniques in Electrical Systems (CATCON), Rupnagar, India, Nov. 2017.
[13] S. M. Sreelakshmi, L. Tharamal, and P. Preetha, “Maximum likelihood classification for transformer fault diagnosis using dissolved gas analysis,” in Proc. 2021 IEEE Electrical Insulation Conference (EIC), Denver, CO, USA, Jun. 2021.
[14] M. S. A. Khiar, S. A. Ghani, I. S. Chairul, Y. H. M. Thayoob, and Y. Z. Y. Ghazali, “On-site OLTC monitoring using duval triangle and DWRM,” in Proc. 2014 2nd International Conference on Technology, Informatics, Management, Engineering & Environment, Bandung, Indonesia, Aug. 2014.
[15] M. S. A. Khiar, Y. H. Md Thayoob, Y. Z. Y. Ghazali, S. Ab Ghani, and I. S. Chairul, “Condition assessment of OLTC using duval triangle and static winding resistance test,” in Proc. 2012 IEEE International Power Engineering and Optimization Conference, Melaka, Malaysia, Jun. 2012.
[16] P. Taecharoen, P. Kunagonniyomrattana, and S. Chotigo, “Development of dissolved gas analysis analyzing program using visual studio program,” in Proc. 2019 IEEE PES GTD Grand International Conference and Exposition Asia (GTD Asia), Bangkok, Thailand, Mar. 2019.
[17] 蔡明勳，「電力變壓器故障偵測與諧波分析」，國立中山大學電機工程學系碩士論文，2003年。
[18] D. L. Pavia, G. M. Lampman, G. S. Kriz, and R. G. Engel, Introduction to Organic Laboratory Techniques. Boston, Massachusetts, USA: Thomson Learning, 1999.
[19] D. W. Ball, “Photoacoustic spectroscopy,” Chemistry Faculty Publications, vol. 21, no. 9, Sept. 2006.
[20] 王浩庭，應用類神經網路於油浸式變壓器之故障診斷系統，國立清華大學工業工程與工程管理學系專題報告，2009年
[21] A. Srivastava. (2022). A Simple Guide to Data Preprocessing in Machine Learning. [Online]. Available: https://www.v7labs.com/blog/data-preprocessing-guide
[22] Bureau of Reclamation, Transformers: Basics, Maintenance and Diagnostics. Denver, Colorado, USA: US Department of the Interior, Apr. 2005.
[23] Mineral Oil-Filled Electrical Equipment in Service - Guidance on the Interpretation of Dissolved and Free Gases Analysis, IEC Std 60599:2022, May 2022.