臺灣博碩士論文加值系統

本研究將資料探勘（Data mining）運用在離心式冰水主機故障診斷，透過ASHRAE RP-1043[1]離心式冰水主機運轉數據，驗證本研究之方法，藉由資料探勘的方式將數據進行有效的分類。運用R軟體做資料探勘，當資料量大且變數繁多時，為了提升分類效率，進行資料分析前會將資料進行降維（Dimension Reduction），故本研究使用相關特徵快速過濾法（FCBF）的方式提取特徵最佳子集，接著透過梯度提升決策樹（GBDT）發展離心式冰水主機故障偵測與診斷（FDD），針對七種常見離心式冰水主機故障進行偵測與診斷。研究結果發現本研究使用相關特徵快速過濾法（FCBF）結合梯度提升決策樹（GBDT）診斷準確率較線性判別分析（LDA）結合支撐向量機（SVM）高出20.89%。

In this research, Data mining is applied to the fault diagnosis of Centrifugal Chiller. Then verify the research method through ASHRAE RP-1043[1] Centrifugal Chiller operational data, and make effective data classification by data mining. When the R software is applied for data mining, to imporve classification efficieny of large data amount and multiple variables, Dimension Reduction is applied to data before data analysis. Therefore, the Fast Correlation Based Filter for Feature Selection（FCBF） is used in this research to retrieve optimal feature subset, and develop Fault Detection and Diagnosis（FDD） of Centrifugal Chiller by Gradient Boosting Decision Tree（GBDT） to undertake the Fault Detection and Diagnosis of seven common Centrifugal Chiller faults. The results show that the diagnostic rate using the Fast Correlation Based Filter for Feature Selection（FCBF） combined with the Gradient Boosting Decision Tree （GBDT) is 20.89% higher than the Linear Discriminant Analysis（LDA） combined with the Support Vector Machine（SVM）.

摘要ii
Abstractiii
誌謝iv
目錄v
表目錄vii
圖目錄viii
第一章 緒論1
1.1 前言1
1.2 研究動機與目的2
1.3 文獻回顧4
1.3.1 冰水主機故障診斷文獻回顧4
1.3.2 梯度提升決策樹（GBDT）文獻回顧6
1.3.3 相關特徵快速過濾法（FCBF）文獻回顧7
1.4研究流程8
第二章 理論分析10
2.1 中央空調系統10
2.2 離心式冰水主機故障12
2.3 故障偵測與診斷理論14
2.3.1 故障偵測（Fault Detection）14
2.3.2 故障診斷（Fault Diagnosis）15
2.3.3 故障評估（Fault Evaluation）15
2.3.4 系統反應（Reaction）15
第三章 研究方法17
3.1 梯度提升決策樹（GBDT）17
3.1.1 集成學習（ensemble learning）概述17
3.1.2 決策樹（Decision Tree）20
3.1.3 梯度提升決策樹（GBDT）分類23
3.1.4 梯度提升決策樹（GBDT）正則化24
3.1.5 梯度提升決策樹（GBDT）範例25
3.2 相關特徵快速過濾法（FCBF）28
3.3 標準分數（Standard Score）29
3.4 GBDT與FCBF結合數據資料庫之離心式冰水主機故障偵測與診斷策略30
第四章 實驗結果與分析31
4.1 GBDT-FCBF模型建置31
4.1.1 案場介紹31
4.1.2 相關特徵快速過濾法（FCBF）篩選最佳子集38
4.1.3 梯度提升決策樹（GBDT）預測模型資料確定39
4.2 GBDT-FCBF故障偵測與診斷模行驗證40
4.2.1 故障偵測與診斷系統測試40
4.2.2 測試結果分析42
第五章 結論與建議45
5.1 結論45
5.2 建議46
參考文獻 47

[1]M.C. Comstock , J.E. Braun, “Experimental data from fault detection and diagnostic studies on a centrifugal chiller,” ASHRAE RP-1043 Report, 4036-1,1999.
[2]台灣電力公司，過去電力供需資訊，2018。
[3]經濟部能源局，非生產性質行業能源查核年報，2018。
[4]江長學，「應用模糊推論與類神經網路理論於離心式冰水機故障偵測與診斷策略之發展」，碩士論文，國立臺北科技大學能源與冷凍空調工程學系，2010，取自https://hdl.handle.net/11296/8ucgku
[5]S.Katipamula , M.R. Brambley, “Methods for Fault Detection, Diagnostics, and Prognostics for Building Systems-A Review, part II,” HVAC&R Research,11（2）2005, pp.169-187.
[6]陳家榮，「冰水主機節能系統效率提升」，能源服務業成功案例示範觀摩研討會，新竹，2006。
[7]呂金翰，「冰水主機自我故障診斷之研究」，碩士論文，國立臺北科技大學能源與冷凍空調工程學系，2009，取自https://hdl.handle.net/11296/k6g9bv 。
[8]G.Li , Y.Hu , H.Chen ,et al. “An improved fault detection method for incipient centrifugal chiller faults using the PCA-R-SVDD algorithm,” Energy and Buildings, 2016,pp.104-113.
[9]R.Huang , J.Liu , H.Chen ,et al. “An effective fault diagnosis method for centrifugal chillers using associative classification,” Applied Thermal Engineering , 2018 , pp.633-642.
[10]Ｙ.Zhao , F.Xiao , S.Wang. “An intelligent chiller fault detection and diagnosis methodology using Bayesian belief network,” Energy and Buildings, 2013 , pp.278-288.
[11]D.A.T.Tran , Y.Chen , M.Q.Chau ,et al. “A robust online fault detection and diagnosis strategy of centrifugal chiller systems for building energy efficiency,” Energy and Buildings, 2015.pp.441-453.
[12]D.Li , G.Hu , C.J.Spanos, “ A data-driven strategy for detection and diagnosis of building chiller faults using linear discriminant analysis,” Energy and Buildings, 2016 ,pp.519-529.
[13]G.N.Li , Y.P.Hu , Q.J.Mao , et al. “A deep neural network based fault diagnosis method for centrifugal chillers,” IOP Conference Series： Earth and Environmental Science, 2019.
[14]P.Wang , J.Xin , X.G , et al. “ Chiller gradual fault detection based on Independent Component Analysis,” The 27th Chinese Control and Decision Conference（2015 CCDC,2015.
[15]X.Liu , Y.Li , X.Liu , et al. “Fault diagnosis of chillers using very deep convolutional network,” 2018 Chinese Automation Congress （CAC） ,2018.
[16]T.Yang , W.Chen , G.Gao, “Automated classification of neonatal amplitude-integrated EEG based on gradient boosting method,” Biomedical Signal Processing and Control, 2016 ,pp.50-57.
[17]D.Tian , G.He , J.Wu , et al. “An accurate eye pupil localization approach based on adaptive gradient boosting decision tree,” 2016 Visual Communications and Image Processing （VCIP） , 2016.
[18]M.Zhang , Z.Liu , X.Dang . “Fault Diagnosis on Train Brake System Based on Multi-dimensional Feature Fusion and GBDT Enhanced Classification,” 2018 International Conference on Intelligent Rail Transportation （ICIRT） , 2018.
[19]X.Fu , D.W , H.Z . “Identifying Transportation Modes Using Gradient Boosting Decision Tree,” ICIC 2018： Intelligent Computing Theories and Application , 2018 , pp.543-549.
[20]H.Duan , Q.Zhang , M.Zhang. “FCBF algorithm based on normalized mutual information for feature selection,” Huazhong University of Science and Technology , 2017 , pp.52-56.
[21]K.R.Kavitha , A.Gopinath , M.Gopi. “Applying improved svm classifier for leukemia cancer classification using FCBF,” 2017International Conference on Advances in Computing, Communications and Informatics （ICACCI）,2017.
[22]Y.Liu , J.Zhang , L.Ma. “A fault diagnosis approach for diesel engines based on self-adaptive WVD, improved FCBF and PECOC-RVM,” Neurocomputing , 2016 , pp.600-611.
[23]Y.Khourdifi , M.Bahaj. “Feature Selection with Fast Correlation-Based Filter for Breast Cancer Prediction and Classification Using Machine Learning Algorithms,” 2018 International Symposium on Advanced Electrical and Communication Technologies （ISAECT） , 2018.
[24]L.Gao , M.Ye , C.Wu . “Cancer Classification Based on Support Vector Machine Optimized by Particle Swarm Optimization and Artificial Bee Colony,” molecules ,2017.
[25]經濟部能源局，空調系統能源查核與節約能源案例手冊。
[26]“Applications Engineering Manual,” TRANE, 2011 , pp.21.
[27]International Energy Agency （IEA）, “Demonstrating automation fault detection and diagnosis methods in real buildings,” FINLAND： Technology Research Center of FINLAND, 2001.
[28]凌士剛，「性能指標迴歸模式準確性對離心式冰水機故障偵測與診斷策略影響之研究」，碩士論文，國立臺北科技大學能源與冷凍空調工程學系，2009，取自https://hdl.handle.net/11296/49af97 。
[29]R. Isermann, “Process fault detection based on modeling and estimation a survey,” Automatica, 1984 , pp. 387-404.
[30]Alberto Torres-Barrán, Álvaro Alonso , JoséR. Dorronsoro , “Regression tree ensembles for wind energy and solar radiation prediction”, Neurocomputing, 2017.
[31]L.Yu , H.Liu. “Feature Selection for High-Dimensional Data： A Fast Correlation-Based Filter Solution,” ICML03 Proceedings of the Twentieth International Conference on International Conference on Machine Learning , 2003 , pp.856-863.