本研究針對船舶柴油機系統在運轉時，可能發生的常見故障進行研究，通過對其故障特點、故障模式部位、常見故障等方面的統一分析，歸納出發生該故障的異常現象並分析故障原因，為系統故障診斷與故障排除。本研究藉由某船提供之2800 TEU 貨櫃船 (Container Vessel) 實際運轉的方式進行各種故障分析包含燃油系統、空氣系統、滑油系統、冷卻水系統、排氣系統。首先利用故障樹分析(Fault Tree Analysis,FTA)方法結合專家知識與常見故障類型進行預處理，對其做故障樹分析，統整後繪製出樹狀圖，找出底事件即是故障原因，並將其設定標籤，作為故障的特徵參數，彙整成系統故障的數據集，再由K-means演算法進行集群分析，找出相關的特徵或模式，第一種方法使用倒傳遞類神經網路(Back Propagation Neural Network)對故障特徵參數進行訓練和識別診斷，另一種方法使用主成分分析(Principal Component Analysis,PCA)對資料進行降維和特徵提取，不僅降低了計算量同時也提高了分類器的診斷精度。接著使用支持向量機(Support Vector Machine,SVM)訓練模型去進行分類與故障診斷，能夠有效提高診斷的準確率，具有很好的理論和實用價值。最後透過徑向基函數(Radial Basis Function,RBF)進行參數優化產生懲罰因子(C)與核參數(gamma,σ)，實驗結果顯示確實提高支持向量機的準確率。

This study investigaes the common faults that may occur when the marine diesel engine system is running. By analyzing the fault characteristics, the mode fault positions, and common fault examples, giving the abnormal phenomenon that occurs with a fault and analysis for the cause of that fault, to diagnose and troubleshoot system failures. The various trouble shootings such as fuel oil system, air system, lubricating oil system, cooling water system, exhaust gas system performed by the actual operation of the 2800 TEU Container Vessel provided. First, we use the method of fault tree (Fault Tree Analysis; FTA) combines expert knowledge with common fault types for pre-treatment, perform fault tree analysis and draw a tree diagram after integration, and find out the bottom event which is the cause of failure, and setting the label as the characteristic parameter of the fault, and aggregate it into a data set of system faults. Furthermore, we perform cluster analysis by K-means algorithm to find out the relevant features or modes, and the first method uses Back Propagation Neural Network to train and identify fault characteristic parameters, and another method uses Principal Component Analysis to reduce the dimensionality and feature extraction of the data, which not only reduces the amount of computation but also improves the diagnostic accuracy of the classifier. Then we use the support vector machine training the model to classify and troubleshoot, which can effectively improve the accuracy of diagnosis and has good theoretical and practical value. Finally, through the radial basis function (Radial Basis Function, RBF) parameter optimization to generate penalty factor (C) and kernel parameters (gamma, σ), the experimental results show that the accuracy of the support vector machine is indeed improved.

摘要 i
Abstract ii
致謝 iv
符號表 xiii
一、 緒論 1
1.1 研究目的 1
1.2 研究動機與文獻回顧 4
1.3 論文架構 6
二、 實驗設備 8
2.1 燃油系統 8
2.1.1 燃油系統的組成結構與功能 8
2.1.2 燃油系統結合專家知識統整的故障原因 9
2.2 空氣系統 18
2.2.1 壓縮空氣系統的組成結構與功能 18
2.2.2 空氣系統結合專家知識統整的故障原因 22
2.3 滑油系統 28
2.3.1 滑油系統的組成結構與功能 28
2.3.2 滑油系統結合專家知識統整的故障原因 31
2.4 冷卻水系統 37
2.4.1 冷卻水系統的組成結構與功能 37
2.4.2 冷卻水系統結合專家知識統整的故障原因 40
2.5 排氣系統 48
2.5.1 排氣系統的組成結構與功能 48
2.5.2 排氣系統結合專家知識統整的故障原因 51
2.6 硬體系統 58
三、 實驗方法 59
3.1 故障樹分析(FTA) 59
3.1.1 引言 59
3.1.2 故障樹分析概述 59
3.2 K-means分群演算法 67
3.2.1集群分析概述 67
3.2.2 K-meams分群演算法概述 68
3.2.3 K-meams分群演算法的優缺點 71
3.3 倒傳遞類神經網路 72
3.3.1 倒傳遞類神經網路概述 72
3.3.2 類神經網路模型 73
3.3.3 BP神經網路的結構和學習算法 75
3.3.4 BP神經網路的優缺點 82
3.4 主成分分析 83
3.4.1 主成分分析的原理與推導 83
3.4.2 主成分分析的貢獻率 90
3.5 支持向量機 91
3.5.1 支持向量機的原理和推導 91
3.5.2 最優分類超平面與間隔最大化 92
3.5.3 對偶問題(Dual Problem) 95
3.5.4 核技巧(Kernel Trick)與核函數(Kernel Function) 98
3.5.5 常用的核函數(Kernel Function) 99
3.5.6 最佳化演算法 100
3.6 研究步驟 103
3.6.1 實驗架構 103
3.6.2 實驗流程 104
四、 結果與討論 105
4.1 K-meams演算法 105
4.2 BP神經網路的分析結果： 107
4.3 PCA+SVM的分析結果： 110
4.4 SVM參數優化(Tune Parameters)： 114
4.5 PCA+RBF_SVM 結果討論： 115
五、 結論及未來展望 120
5.1 結論 120
5.2 未來展望 121
參考文獻 122

[1]HU, Y. H. , (2013). Accident analysis and safety assessment of marine machinery system[M]. Beijing: China Communications Press, 12 (in Chinese).
[2]LI, B. , CHEN, M. Y. , WANG, R. R. , et al. (2015). Fault diagnosis for the ship electric propulsion system[C], Proceedings of 2015 11th International Conference on Natural Computation. Zhangjiajie, China: IEEE, pp. 714-718.
[3]Cao, L. , Shen, Y. , Shan, T. , Xia, Y. , Wang, J. , Lin, Z. , (2018). Bearing Fault Diagnosis Method Based on GMM and Coupled Hidden Markov Model, 2018 Prognostics and System Health Management Conference (PHM-Chongqing), IEEE.
[4]I. Morgan, B. Tormos, A. Salai, et al. ,(2009). Detection and Diagnosis of Incipient Faults in Heavy Duty Diesel Engines [J]. Industrial Electronics, 57(10): pp. 3522-3532.
[5]SINHA, J. K. , ELBHBAH K. , (2013). A future possibility of vibration based condition monitoring of rotating machines[J]. Mechanical Systems and Signal Processing, 34(1/2): pp. 231–240.
[6]LAMARIS, V. T. , HOUNTALAS. D. T. , (2009). Possibility to determine diesel engine condition and tuning from the application of a diagnostic technique at a single operating point[J]. SAE International Journal of Engines, 2(1): pp. 585–605
[7]YU, Y. H. , (2007). Research on monitoring and diagnosing for marine diesel engine based on instantaneous angular speed and thermal parameters[D]. Wuhan: Wuhan University of Technology, (in Chinese).
[8]ARVESON, P. T. , (2000). VENDITTIS, D. J. , Radiated noise characteristics of a modern cargo ship[J]. The Journal of the Acoustical Society of America, 107(1)： pp. 118–129.
[9]LIN, T. , HORNE, B. G. , TINO, P. ,et al. ,(1996). Learning longterm dependencies in NARX recurrent neural networks[J]. IEEE Transactions on Neural Networks, 7(6): pp. 1329–1338.
[10]Liu, B. , Pan, H. , Li, X. , (2010). An Expert System for Fault Diagnosis in Diesel Engine Based on Wavelet Packet Analysis and Hybrid PSO-DV Based Neural Network. 2010 International Conference on Intelligent Computing and Cognitive Informatics, IEEE.
[11]Liu, C. , (2017). Research and Application of Computer Aided Analysis and Optimization Algorithm for Fault Tree. 2017 International Conference on Computer Technology, Electronics and Communication (ICCTEC), IEEE.
[12]Alsabti, K. , Ranka, S. , Singh, V. , (1998). An Efficient Parallel Algorithm for High Dimensional Similarity Join[C], Parallel Processing Symposium, Ipps/spdp 1998. Proceedings of the First Merged International. and Symposium on Parallel and Distributed Processing. IEEE, pp. 556-560.
[13]Ni, F. Y. , Cheng, H. Z. , (2005). Improved neural network BP algorithm based on prediction of short-term power load. Distribution & Utilization, 2(25): pp. 16-19.
[14]Sehgal, S. , Singh, H. , Agarwal, M. , Bhasker, V. , & Shantanu, (2014). Data analysis using principal component analysis. 2014 International Conference on Medical Imaging, m-Health and Emerging Communication Systems (MedCom), pp. 45-48, doi: 10.1109/MedCom.2014.7005973.
[15]Chu, D. , Lu, R. , Li, J. , (2018). Optimizing Top- k Multiclass SVM via Semismooth Newton Algorithm. IEEE Transactions on Neural Networks and Learning Systems ( Volume: 29 , pp. 6264 - 6275 ), IEEE.
[16]Chen, K. ,(2019). Indirect PCA Dimensionality Reduction Based Machine Learning Algorithms for Power System Transient Stability Assessment. 2019 IEEE Innovative Smart Grid Technologies - Asia (ISGT Asia), IEEE.
[17]Björn Waske, Sebastian van der Linden, Jón Atli Benediktsson, Andreas Rabe, Patrick Hostert, (2010). Sensitivity of Support Vector Machines to Random Feature Selection in Classification of Hyperspectral Data. IEEE Transactions on Geoscience and Remote Sensing ( Volume: 48 , pp. 2880 - 2889 ), IEEE.
[18]S. Randive Vishal, Katkar Omkar Prataprao, Nashte Adarsh Pravin, A. Rammohan, (2017). Investigation of effect of air filter clogging on performance and emissions from engine. 2017 International conference on Microelectronic Devices, Circuits and Systems (ICMDCS), IEEE.
[19]田文國、葉榮華、哈子銘 （2009）。輪機當值與管理（第十二章，頁11-22）。 基隆市：田文國。
[20]Csbc corporation taiwan department of design, (2019). FOR WORKING.
[21]Csbc corporation taiwan department of design, (2020). FOR WORKING.
[22]翁澤民（2000）。船舶柴油機故障診斷仿真係統研究。計算機仿真。
[23]呂傳曾、鄭書修、吳基榮、黃道祥、郭錦榮、方福樑（2006）。船舶柴油機。臺北市：呂傳曾。
[24]ABS RULES FOR BUILDING AND CLASSING MARINE VESSELS, (2021). pp. 613
[25]Tri Oktaviana Putri, Rahmadwati, Bambang Siswojo, (2014). Temperature control of liquid egg pasteurization system using PLC (programmable logic controller) siemens simatic S7-200 and HMI (human machine interface) simatic HMI panel. 2014 Electrical Power, Electronics, Communicatons, Control and Informatics Seminar (EECCIS)，2014，IEEE
[26]Sarfraz Ali Quadri, Swapneel R. Zende, Dhananjay R. Dolas. , (2014). Reliability Estimation using Fault Tree Analysis Method. International Journal of Engineering Research Volume No.3 Issue No: Special 1.
[27]Mohammad Sadegh Javadi, Azim Nobakht, Ali Meskarbashee, (2011). Fault Tree Analysis Approach in Reliability Assessment of Power System, International Journal of Multidisciplinary Sciences and Engineering.
[28]Bollen M. H. J. , (2001). Effects of adverse weather and aging on power system reliability, IEEE Trans. Ind. Appl.
[29]Wold, S. , Esbensen, K. , Gelasi, P. , (1987). Principle Components Analysis. Chem. Intell. Lab. Syst,pp. 37-46.
[30]Skoundrianos, E. N. , Tzafestas, S. G. ,(2002). Fault diagnosis in via local neural networks [J]. Mathematics and Computers in Simulation, pp. 169-180.
[31]韓曉靜（2008）。基於小波變換與神經網絡的電力電子電路故障診斷研究。中國。
[32]Wang, J. P. ,(2002). Fault Diagnosis Technology Based on Neural Network Multisensor Data Fusion. Journal of Mechanical Science and Technology. 21(1), pp. 127-130.
[33]Bishop, Ch. M. , (1995). Neural networks for pattern recognition. Oxford University Press, Oxford, New York, pp. 164-177.
[34]石靈丹、槐博超、馬修真（2009）。RBF神經網絡在柴油機燃油系統故障診斷中的應用研究。船電技術，18-22。
[35]Chaudhari, N. S. , Tiwari, A. ,Thomas, J. ,(2008). Performance evaluation of SVM based semi-supervised classification algorithm. 2008 10th International Conference on Control, Automation, Robotics and Vision, IEEE.
[36]Hu, Y. Z. , Wang, W. X. , Liu, H. , Liu, L. Q. , (2019). Robotic Tracking Control with Kernel Trick-based Reinforcement Learning. 2019 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), IEEE.
[37]M‘hamed Bilal Abidine, Nawel Yala;Belkacem Fergani, Laurent Clavier, (2014). Soft margin SVM modeling for handling imbalanced human activity datasets in multiple homes. 2014 International Conference on Multimedia Computing and Systems (ICMCS), IEEE.
[38]Barabino, N. , Pallavicini, M. , Petrolini, A. , et al. ,(1999). Support vector machines vs multi-layer perceptrons in particle identification. Verleysen M. Processdings ESANN, Brussels: Dfactro, pp. 257-262.
[39]Bartlett, P. L. , Shawe-Teylor, J. , (1999). Generalization performance of support vector machines and other pattern classifiers. Scholkopf B, Burges C.J.C., Smola A.J., Advances in Kernel Methods-Support Vector Learning, Cambridge: MIT Press , pp. 43-54.
[40]Scholkof, B. , Mika, S. , Burges, C.J.C. , et al. ,(1999). Input space versus feature space in kernel -based methods. IEEE Trans Neural Networks, 10(5), pp.1000-1017.
[41]Drucker, H. , Wu, D. , Vipnik, V. N. ,(1999). Support vector machines for spam categorization. IEEE Trans Neural Networks, 10(5), pp. 1048-1054.
[42]Smola, A. , Scholkopf, B. ,(1998). A tutorial on support vector regression. Neuro COLT: Technical Report 19.
[43]Vapnik, V. N. , (1987). Support vector method for function approximation. regression, and signal processing. In: Proc. NIPS’9, San.
[44]Watanabe, H. , Katagiri, S. , Yamada, K. , McDermott, E. , Nakamura , A. , Watanabe, S. ,(2010). Minimum Error Classification with geometric margin control. 2010 IEEE International Conference on Acoustics, Speech and Signal Processing, IEEE.
[45]Chen, S. , Harris, C. J. ,(2000). Design of the optimal separating hyperplane for the decision feedback equalizer using support vector machines. 2000 IEEE International Conference on Acoustics, Speech, and Signal Processing. Proceedings (Cat. No.00CH37100), IEEE.
[46]李航，（2012）。統計學習方法。清華大學出版社，中國。
[47]Han, X. M. , Xu X. Y. , Xie K. M. , (2010). An approach of Support Vector Machine to improve the training speed. 2010 International Conference On Computer Design and Applications, IEEE.
[48]Li, H. , Zhang, Y.X. , (2009). An algorithm of soft fault diagnosis for analog circuit based on the optimized SVM by GA. 2009 9th International Conference on Electronic Measurement & Instruments, IEEE.
[49]Wu, C. M. , Wang, X. D. , Bai, D. Y. , Zhang, H. D. , (2009). Fast Incremental Learning Algorithm of SVM on KKT Conditions. 2009 Sixth International Conference on Fuzzy Systems and Knowledge Discovery, IEEE.
[50]Lupu, D. , Necoara, I. ,(2018). Primal and dual first order methods for SVM： applications to driver monitoring. 2018 22nd International Conference on System Theory, Control and Computing (ICSTCC), IEEE.
[51]Wu, D. , Cao, F. L. , (2009). Learning rates for SVM classifiers with polynomial kernels. 2009 International Conference on Machine Learning and Cybernetics, IEEE.
[52]Bong, K. , Kim, G. , Yoo, H. J. , (2014). Energy-efficient Mixed-mode support vector machine processor with analog Gaussian kernel. Proceedings of the IEEE 2014 Custom Integrated Circuits Conference, IEEE.
[53]Qu, L. P. , Zhou, H. H. , Liu, C. J. , Lu, Z. , (2018). Study on Multi-RBF-SVM for Transformer Fault Diagnosis. 2018 17th International Symposium on Distributed Computing and Applications for Business Engineering and Science (DCABES), IEEE.
[54]Hsu, C. W. , Chang, C. C. , Lin, C. J. , (2003). A practical guide to support vector classification. Technical Report，University of National Taiwan，Department of Computer Science and Information Engineering.
[55]Li, C. H. , Lin, C. T. , Kuo, B. C. , Chu, H. S. , (2010). An automatic method for selecting the parameter of the RBF kernel function to support vector machines. International Geoscience and Remote Sensing Symposium(IGARSS), IEEE, pp. 836-839.
[56]謝佩鈞，（2015）。Full Bandwidth RBF核函數參數自動挑選法與其在特徵選取之應用。國立臺中教育大學教育資訊與測驗統計研究所碩士論文，12-14。
[57]曾琨勝，（2020）。船用柴油機之故障診斷系統。國立高雄科技大學電機工程系碩士班碩士論文。