臺灣博碩士論文加值系統

本論文針對電力轉換器提出及研製具有故障預測功能的即時遠端監測和數據採集的管理系統。由於電子設備的維修保養，通常採取故障報修或零件定期更換。故障報修可能導致不可預期的故障發生，除了影響維修保養的執行，也可能危及生命安全。而定期更換則可能造成不必要的浪費，增加設備維修成本。本論文所開發的系統平台能提前預測可能即將發生的故障，以便提早更換元件，避免造成系統更大的損壞，以降低系統的保養與維修成本。本研究進行關鍵參數的特徵變化的分析，通過機器學習方法預測重要關鍵部件的健康狀況。由於電力轉換器中最易發生故障的部件是電容器，電容器被分配為檢測電子設備健康診斷和決策故障點的關鍵元件。但目前針對電容的壽命分析仍依賴可靠度分析與電容廠所提供的壽命規格，無法進行即時的預測。因此，有別於以往的分析電容ESR的變化，本論文提出的預測方式透過分析電容溫度曲線搭配遞迴神經網路，結合溫度感測器量測電容溫度與遠端即時監測與預測系統，達到即時的預測，以評估電容器是否將要損壞，進而提前對電子設備進行維修或更換元件。

In this paper, a system platform of real-time remote monitoring and data acquisition with capacitor-fault prognostic function are proposed and developed for power converters. In general, the maintenance and repair of electronic equipment mainly takes the methods of fault repair or regular replacement of components. The fault repair can easily lead to unpredictable failures, which may affect the execution of maintenance and may endanger the safety of personnel. However, the regular component replacement can cause unnecessary waste and increase equipment maintenance costs. In this paper, the system platform is developed to predict the possible failures in advance, so that components can be replaced early to avoid greater damage to the circuit system, thereby reducing the maintenance and repair costs of the complete system. The characteristic variations of key parameters have been analyzed to predict the health of some critical components through machine learning approaches. Since the most easily faulty component of power converters is the capacitor, the capacitor is assigned as a key component to detect the fault point for health diagnosis and decision making of electronic device. At present, the life analysis of capacitors mainly relies on the component specifications provided by the capacitor factory for reliability analysis, so it is difficult to make real-time predictions. Therefore, unlike the previous analysis based on the changes of capacitance ESR, the temperature sensing circuits are developed and the remote real-time monitoring and prediction system is integrated to implement the real-time measurement of capacitance temperature in this paper. Furthermore, the recursive neural network algorithm is applied to analyze the characteristics of capacitance temperature curve, and to achieve the real-time health prediction of the capacitor. It is found that the critical time for failure point of device circuits is close to the abrupt sharp slope change in slope variations of capacitor temperature. Therefore, whether the electronic components will be about to be damaged may be evaluated to predict the health status of device circuit in real time. In conclusion, the reliability and stability of electronic equipment will be expected to greatly improve and substantially reduce unnecessary expenditure and manpower waste in this study.

摘要...i
Abstract... ii
誌謝... iv
目錄... v
表目錄 ...ix
圖目錄 ...x
第一章 緒論 ...1
1.1 研究背景及目的 ...1
1.2 文獻探討... 1
1.3系統應用情境 ...2
1.4 論文架構... 3
第二章 智慧聯網整合技術探討... 4
2.1 虛實整合系統概述 ...4
2.2 實體區域網路架構 ...6
2.2.1控制器區域網路(CAN) ...6
2.2.1.1 CAN通訊原理 ...6
2.2.1.2 CAN實體層 ...7
2.2.1.3 CAN資料連結層 ...7
2.3 物聯網與智慧診斷系統架構 ...8
2.4 小結 ...9
第三章 電力轉換器故障探討與深度學習架構 ...10
3.1 交換式電源供應器故障探討 ...10
3.1.1 返馳式(Flyback)轉換器原理與架構 ...10
3.1.3 電解電容失效分析 ...12
3.2 遞迴神經網路 ...12
3.2.1 類神經網路基本架構與原理 ...12
3.2.2 遞迴神經網路(RNN) ...13
3.2.3 長短期記憶神經網路(LSTM) ...13
3.2.4 Gate Recurrent Unit(GRU) ...14
3.3小結 ...16
第四章 系統平台規劃與硬體設計 ...17
4.1 感測器硬體設計 ...18
4.1.1 感測器硬體架構 ...18
4.1.2 微處理器簡介 ...19
4.1.3 溫度感測電路 ...21
4.1.4 溫溼度感測器 ...23
4.1.5 電壓箝位電路 ...24
4.1.6 電壓偵測電路 ...25
4.2 感測節點韌體與系統平台程式設計 ...25
4.2.1 感測節點韌體規劃 ...25
4.2.2 SCADA系統設計 ...29
4.2.3 網頁與資料庫伺服器架設... 31
4.2.4 網頁遠端監測介面架構 ...31
4.2.5 資料庫規劃 ...31
4.2.6 雲端預測系統... 32
第五章 系統實測與討論... 34
5.1感測節點實體聯網 ...34
5.2 SCADA介面與網頁監測介面實作... 36
5.3測試環境 ...39
5.4預測方法分析與結果討論 ...42
5.5雲端資料庫 ...49
第六章 結論與未來展望... 51
6.1 結論 ...51
6.2 未來展望 ...51
參考文獻 ...52
Extended Abstract... 54
Abstract ...54
1. Design and configuration ...55
2. Conclusion... 58

[1]Military Handbook Reliability Prediction of Electronic Equipment ,MIL-HDBK-217, July 1992.
[2]Sony Mathew, Mohammed Alam, Michael Pecht,Identification of Failure Mechanisms to Enhance Prognostic Outcomes,Journal of Failure Analysis and Prevention, Vol. 12, No. 1, pp 66-73,Feb. 2012.
[3]Dan N. Bennett, “Off-Line SMPS Failure Modes PWM Switchers and DC-DC Converters”, Texas Instruments Application Report SLVA085, March 2000.
[4]EPARC，電力電子學綜論，第二版，全華科技圖書股份有限公司，新北市，民國一百年六月。
[5]Amine Lahyani, Pascal Venet, Guy Grellet, and Pierre-Jean Viverge,” Failure Prediction of Electrolytic Capacitors During Operation of a Switchmode Power Supply”, IEEE Transactions on Power Electronics, Vol: 13, No:6, pp:1199-1207,August 2002.
[6]Douglas Leonard Goodman, James Hofmeister, “Electronic prognostics for switched mode power supplies”,Elsevier Ltd, pp:1902-1906, Received 19 December 2006.
[7]Rolf Orsagh, Douglas Brown, and Michael Roemer, “Prognostic Health Management for Avionics System Power Supplies” ,2005 IEEE Aerospace Conference, 19 December 2005.
[8]Liu Xiao-Hua, Yuan Da, Li Jin-Jiang, “Image Contour Extraction Based on CNN and Active Contour Model”, IEEE,2009 Fifth International Conference on Natural Computation,2009.
[9]“Google’s Neural Machine Translation System: Bridging the Gap between Human and Machine Translation”,Google,CoRR,Vol. abs/1609.08144,2016.
[10]Qin Zhang , Hui Wang , Junyu Dong , Guoqiang Zhong , Xin Sun ,“Prediction of Sea Surface Temperature Using Long Short-Term Memory,”IEEE Geoscience and Remote Sensing Letters, Oct. 2017.
[11]工業4.0熱潮延燒 驅使製造業搶建智慧工廠，DIGTIMES物聯網， https://www.digitimes.com.tw/iot/article.asp?cat=158&id=0000465476_tju5y5px677its7uoew56，2016/05。
[12]“行政院生產力 4.0 發展方案”，行政院，民國一百零四年。
[13]Jay Lee, Behrad Bagheri, Hung-An Kao,“A Cyber-Physical Systems architecture for Industry 4.0-based manufacturing systems”, Manufacturing Letters, Vol.3, pp.18–23, 10 December 2014.
[14]“Controller Area Network (CAN BUS) 通訊協定原理概述”，國家儀器，http://www.ni.com/white-paper/2732/zht/.
[15]CAN in Automation (CiA): CAN physical layer,https://www.can-cia.org/.
[16]王晨原，CAN Bus的原理與應用，Leaptronix，2017.
[17]李傑，倪軍，王安正，從大數據到智慧生產與服務創新，前程文化，2017.
[18]Sebastian Raschka著，Python機器學習，劉立民，陳曉南譯，博碩文化，新北，民國一○五年八月。
[19]McCulloch, Warren S.; Pitts, Walter. A logical calculus of the ideas immanent in nervous activity. The bulletin of mathematical biophysics , 5 (4): 115–133. ISSN 0007-4985. doi:10.1007/BF02478259,1943.
[20]DE Rumelhart, GE Hinton, RJ Williams , “Learning representations by back-propagating errors”.
[21]Understanding LSTM Networks ,(http://colah.github.io/posts/2015-08-nderstanding-LSTMs/)
[22]Hochreiter and J. Schmidhuber. Long short-term memory. Neural Computation, 9(8):1735–1780, 1997.
[23]Junyoung Chung, Caglar Gulcehre, KyungHyun Cho, Yoshua Bengio ,“Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling”, NIPS 2014 Deep Learning and Representation Learning Workshop, 11 Dec 2014.
[24]RENESAS ,“RENESAS RX62T Group, RX62T Group User’s Manual: Hardware”, Rev.2.00, Dec. 2013 .
[25]Moshe Gerstenhaber , Michael O'Sullivan,高分辨率温度测量, Analog Dialogue, April 2011.
[26]“Sensirion Humidity Sensors SHT25 Datasheet,” Sensirion, Version 3, May 2014.
[27]PCAN-USB : CAN Interface for USB，https://www.peak-system.com/PCAN-USB.199.0.html?L=1.
[28]“65w單組輸出開關電源規格書”，明緯企業股份有限公司，2017。