臺灣博碩士論文加值系統

對於發電業者而言，氣渦輪機於運轉期間之穩定運作對於其是相當重要的，倘若氣渦輪機於運轉期間內發生異常跳機之情形，則可能產生連同汽輪機減少之發電量造成的電力缺口、機組重新啟動之燃料成本、機組材料耗損與其他無法當下評估之隱藏成本等有形與無形之成本，因此異常檢測對於運轉是相當重要的。本研究將某民營電廠之歷史運轉記錄使用機器學習方法CART、SVM、RF以及深度學習CNN進行模型建置，期以能夠有效提前預測出GT運轉過程中發生之異常狀況供相關人員可以及早在異常發生前先進行防範措施，降低異常發生時之損失。本研究實驗結果顯示CART方法在異常狀況預測中表現最好，其F1-measure平均結果為98.94%以及Recall平均結果為99.1%對預測異常狀況之發生較有幫助，其中以異常前10分鐘之時間點之預警效果最好。

It is important for Power Producer to maintain the stability of Gas Turbine operation. If malfunction happen during Gas Turbine operation, the power generation of Steam Turbine will be interrupted and lead the gap of power generation. Besides, the fuel costs of restarting machine, consumption costs of machine materials and other implicit costs which are hard to evaluate, are also the substantial consequences of Gas Turbine malfunction. Thus, Anomaly Detection is important to the stability of Gas Turbine operation. In order to predict the malfunction for engineer to maintain the machine in advance, and further reduce the damage cost cause by GT malfunction. In this study, the historical operation records from a private power plant are modeled using machine learning methods CART, SVM, RF and deep learning method CNN. The experiment results show that CART has the best prediction performance. The average score of F1-measure is 98.94% and the average score of Recall is 99.1%, which is helpful for prediction the occurrence of Anomaly Detection, among which the warning effect is the best at the time point 10 minutes before the anomaly.

目 錄
書名頁 i
論文口試委員審定書 ii
中文摘要 iii
英文摘要 iv
誌 謝 v
目 錄 vi
表目錄 vii
圖目錄 xi
第一章、 緒論 1
第二章、 文獻探討 4
2.1 氣渦輪機 4
2.2 失效模式與影響分析 5
2.3 機器學習 7
2.1.1 分類與迴歸樹 7
2.1.2 深度學習 7
2.1.3 隨機森林 8
2.1.4 支持向量機 8
第三章、 研究方法 9
3.1 硏究架構 9
3.2 資料前處理 10
3.2.1 特徵擴充 10
3.2.2 資料平衡模式 11
3.3 CNN流程 12
3.4 預測 14
第四章、 實驗結果 15
4.1 資料描述 15
4.2 預測方式之結果 15
4.2.1 異常當下 16
4.2.2 異常前10分鐘 17
4.2.3 異常前20分鐘 18
4.2.4 異常前30分鐘 19
第五章、 討論 20
第六章、 結論與未來展望 30
參考文獻 31
附錄 34

中文文獻
[1]. 王韋翔(2018)，深度學習結合基因演算法在遊戲App的生命週期預測，元智大學，碩士論文
英文文獻
[1.] Arabian-Hoseynabadi, H., Oraee, H., & Tavner, P. J. (2010). Failure modes and effects analysis (FMEA) for wind turbines. International Journal of Electrical Power & Energy Systems, 32(7), 817-824.
[2.] Aung, Y. Y., & Min, M. M. (2018, June). Hybrid Intrusion Detection System Using K-Means and Classification and Regression Trees Algorithms. In 2018 IEEE 16th International Conference on Software Engineering Research, Management and Applications (SERA) (pp. 195-199). IEEE.
[3.] Banik, P. P., Saha, R., & Kim, K. D. (2019, February). Fused Convolutional Neural Network for White Blood Cell Image Classification. In 2019 International Conference on Artificial Intelligence in Information and Communication (ICAIIC) (pp. 238-240). IEEE.
[4.] Batayev, N. Gas Turbine Fault Classification Based on Machine Learning Supervised TechniQues. In 2018 14th International Conference on Electronics Computer and Computation (ICECCO) (pp. 206-212). IEEE.
[5.] Bâlc, C., Cretu, A., Iudean, D., Laslo, H., Enyedi, S., & Munteanu, R. (2018, May). Failure modes and effects analysis for an automatic level control system. In 2018 IEEE International Conference on Automation, Quality and Testing, Robotics (AQTR) (pp. 1-6). IEEE.
[6.] Bowles, J. B., & Peláez, C. E. (1995). Fuzzy logic prioritization of failures in a system failure mode, effects and criticality analysis. Reliability Engineering & System Safety, 50(2), 203-213.
[7.] Breiman, L. (2001). Random forests. Machine learning, 45(1), 5-32.
[8.] Carazas, F., & De Souza, G. (2009). Availability analysis of gas turbines used in power plants. International journal of Thermodynamics, 12(1), 28-37.
[9.] Carazas, F. J. G., Salazar, C. H., & Souza, G. (2011). Availability analysis of heat recovery steam generators used in thermal power plants. Energy, 36(6), 3855-3870.
[10.] Chiu, C., Ku, Y., Lie, T., & Chen, Y. (2011). Internet auction fraud detection using social network analysis and classification tree approaches. International Journal of Electronic Commerce, 15(3), 123-147.
[11.] Choubin, B., Moradi, E., Golshan, M., Adamowski, J., Sajedi-Hosseini, F., & Mosavi, A. (2019). An Ensemble prediction of flood susceptibility using multivariate discriminant analysis, classification and regression trees, and support vector machines. Science of the Total Environment, 651, 2087-2096.
[12.] Feili, H. R., Akar, N., Lotfizadeh, H., Bairampour, M., & Nasiri, S. (2013). Risk analysis of geothermal power plants using Failure Modes and Effects Analysis (FMEA) techniQue. Energy Conversion and Management, 72, 69-76.
[13.] Ho, T. K. (1995, August). Random decision forests. In Proceedings of 3rd international conference on document analysis and recognition (Vol. 1, pp. 278-282). IEEE.
[14.] Jearaphun, P., & Tangjitsitcharoen, S. (2018). Reduction of Breakdown for Gas Turbine in Combined Cycle Power Plant. International Journal of Mechanical Engineering and Robotics Research, 7(6).
[15.] Karimi, F., Sultana, S., Babakan, A. S., & Suthaharan, S. (2019). An enhanced support vector machine model for urban expansion prediction. Computers, Environment and Urban Systems, 75, 61-75.
[16.] Li, X., Xi, H., Zhou, C., Gu, W., & Gao, T. (2018, October). Damage Degree Identification of Crane Girder Based on the Support Vector Machine. In 2018 Prognostics and System Health Management Conference (PHM-ChongQing) (pp. 920-924). IEEE.
[17.] Moazzen, Y., Çapar, A., Albayrak, A., Çalık, N., & Töreyin, B. U. (2019). Metaphase finding with deep convolutional neural networks. Biomedical Signal Processing and Control, 52, 353-361.
[18.] Noshad, Z., Javaid, N., Saba, T., Wadud, Z., Saleem, M. Q., Alzahrani, M. E., & Sheta, O. E. (2019). Fault Detection in Wireless Sensor Networks through the Random Forest Classifier. Sensors, 19(7), 1568.
[19.] Sarkar, A., Panja, S. C., & Das, D. (2015). Fault tree analysis of Rukhia gas turbine power plant. HKIE Transactions, 22(1), 32-56.
[20.] Sharma, R. K., Kumar, D., & Kumar, P. (2005). Systematic failure mode effect analysis (FMEA) using fuzzy linguistic modelling. International Journal of Quality & Reliability Management, 22(9), 986-1004.
[21.] Shipway, N. J., Barden, T. J., Huthwaite, P., & Lowe, M. J. S. (2019). Automated defect detection for fluorescent penetrant inspection using random forest. NDT & E International, 101, 113-123.
[22.] Xu, K., Tang, L. C., Xie, M., Ho, S. L., & Zhu, M. L. (2002). Fuzzy assessment of FMEA for engine systems. Reliability Engineering & System Safety, 75(1), 17-29.

[23.] Yang, H. D., & Xu, H. (2011, March). Reliability analysis of Gas turbine based on the failure mode and effect analysis. In 2011 Asia-Pacific Power and Energy Engineering Conference (pp. 1-4). IEEE.
[24.] Vigueras Zuniga, M. O. (2007). Analysis of gas turbine compressor fouling and washing on line.
[25.] Martin Leduc. (2001). The Gas Turbine. Retrieved from http://www.dieselduck.info/machine/01%20prime%20movers/gas_turbine/gas_turbine.htm (May 15,2019)