臺灣博碩士論文加值系統

電腦化維修管理系統(CMMS)提供工廠更彈性的維修工作規劃及更詳細的維修工作紀錄，然而在推動CMMS執行的過程中，許多因素導致系統推動效果不佳，導致維修的紀錄品質逐年下降。由於CMMS的紀錄品質變化屬於漸進式的轉變，專家診斷系統時多為結果導向，難以找到轉變過程的關鍵點。因此，本研究結合數據量化指標及資料探勘技術，並輔以診斷專家，期以更快速、更有效地提供CMMS的異常原因診斷及系統改善的優先順序。
本研究以石化廠CMMS的故障維修履歷為對象，透過指標量化分析以找出CMMS的系統規劃、紀錄品質問題，並輔以關聯規則、群聚分析，找出各項技術無法評估的關鍵因素、造成記錄品質下降的程式面、執行面等問題。
本研究結果於良好的CMMS功能建置中為達成工廠的日常維修及設備管理需求系統應紀錄欄位共79個，其中工單紀錄欄為共52個。以良好的CMMS欄位為基礎執行指標量化評估，在架構完整性指標中範例廠得分為48.1分，透過無誤性指標找出多數欄位記錄品質逐漸下降，其中以故障發生時間最為明顯。透過關聯規則分析找出造成可靠度數據無法計算及計算失準得主要原因來自於設備資訊及時間資訊的填寫不確實。藉由群聚分析得出維修履歷中故障現象、故障原因、故障部位、處理方法等失效代碼有這共同錯誤，其原因來自於使用者的紀錄不確實。設備失效過程中等時間資訊主要來自於使用者紀錄不確實及系統設定為做好卡控。
由於每家企業CMMS的規劃與使用管理上略有不同，只要有該企業的維修履歷資料，則該方法能適用於各家CMMS運行狀況的診斷。CMMS的使用狀況改善後，未來可針對欄位與欄位之間的紀錄資料進行診斷，找出資料紀錄的正確性與否。

The implementation of computerized maintenance management system (CMMS) can give factories more flexible maintenance work planning and more detailed maintenance work records. However, in the past, in the process of promoting the implementation of CMMS, various factors have led to poor system promotion effects, or the quality of maintenance records has decreased year by year. However, the quality of the CMMS system is a gradual change, and experts are less able to find the error message. Therefore, this study combines machine learning and data exploration techniques, with artificial intelligence supplemented by diagnostic experts, to provide faster and more effective CMMS abnormal cause diagnosis and system improvement.
This study takes the fault maintenance history of the CMMS of the petrochemical plant as an object to establish a systematic diagnostic method. It is expected that through quantitative analysis of indicators, problems of CMMS system planning and recording quality can be identified. It is supplemented by association rule analysis and k-means cluster analysis to find out the key factors that cannot be evaluated by various technologies and the management level that causes the quality of records to decline.
The quantitative analysis of indicators can be based on the establishment of good CMMS functions. Review the repair history record fields and repair processes of the petrochemical plant one by one, and find out the bad characteristics of the repair history record of each field. Finally, the system planning status, system field usage status, and past record quality rise and fall trends are presented in a quantitative way as a result of the system ’s preliminary assessment. The results of the association rule analysis show the key factors that cannot be implemented in the equipment evaluation technology of the petrochemical plant, and k-means cluster analysis is used to find out the main sources of these key factors. Provide owners as the focus of improvement.
Because the planning and use management of CMMS of each enterprise is slightly different, as long as there is maintenance history data of that enterprise, this method can be applied to the diagnosis of the operating status of each CMMS. After the usage condition of CMMS is improved, in order to diagnose the record data between the fields and find out the correctness of the data records.

摘要 I
ABSTRACT II
致謝 III
目錄 IV
表目錄 VI
圖目錄 VII
第一章 緒論 1
1.1 研究背景 1
1.2 研究動機 3
1.3 研究目的 5
1.4 論文架構 7
第二章 文獻回顧 10
2.1 電腦化維修管理系統 10
2.1.1 CMMS使用問題 11
2.2 資料品質評估 13
2.2.1 完整性(completeness) 13
2.2.2 無誤性比率(free of error) 14
2.3 監督與非監督式演算法 15
2.3.1 關聯規則 15
2.3.2 k-means群聚演算法 16
第三章 系統優劣化分析方法建立 20
3.1 完善的CMMS建立流程 20
3.1.1 文獻彙整 21
3.1.2 CMMS功能定義 23
3.1.3 CMMS所需欄位建置 25
3.1.4 欄位存在原因 28
3.2 案例數據說明 32
3.3 維修履歷指標量化分析 33
3.3.1 評估流程與實驗數據 34
3.3.2 架構完整性 35
3.3.3 實體完整性 36
3.3.4 欄位完整性 37
3.3.5 無誤性比率 38
3.3.6 資料輸出 42
3.4 基於數據驅動的維修履歷缺失解析 43
3.4.1 資料擷取及轉換 44
3.4.2 關聯規則分析 48
3.4.3 k-means 群聚分析 52
第四章 結果與討論 56
4.1 維修履歷指標量化分析 56
4.1.1 架構完整性 56
4.1.2 欄位完整性及實體完整性結果 58
4.1.3 無誤性比率 59
4.2 基於數據驅動的維修履歷缺失解析 62
4.2.1 關聯規則結果 62
4.2.2 k-means群聚分析結果 67
第五章 結論與建議 72
5.1 結論 72
5.2 建議 74
第六章 參考文獻 75

[1]Crowl, Daniel A, and Joseph F Louvar. , 2001, Chemical Process Safety: Fundamentals with Applications. Pearson Education.
[2]Wang Qingfeng, Liu Wenbin, Zhong Xin, Yang Jianfeng, Yuan Qingbin,2011,Development and application of equipment maintenance and safety integrity management system
[3]黃文聰，2005年，石化廠維修效益主管決策系統建立之研究，中華大學科技管理研究所。
[4]林時東，2012年，完善的資產管理：實現以CMMS為中心的維護管理
[5]Labib A. ,2008,Computerised Maintenance Management Systems. In: Complex System Maintenance Handbook. Springer Series in Reliability Engineering. Springer, London.
[6]Lawrence Kau,Arie Wessels,Jan-Harm Pretorius,2019, Predictive maintenance strategy to increase theavailability of positive displacement pumps: a case study based in Ekurhuleni base metals in South Africa.
[7]Bagadia, K. ,2006, Computerized Maintenance Management Systems Made Easy,McGraw-Hill.
[8]Cato, W., and Mobley, K. , 2001 , Computer-Managed Maintenance Systems:A Step-by-Step Guide to Effective Management of Maintenance.
[9]Society of Petroleum Engineers,2015,Building an In-House CMMS to Simplify Maintenance Management in an Oil and Gas Company.
[10]Sartawi, A. 2013. Effective Implementation and Utilization of CMMS System: Challenges and Solutions. In Cases on Enterprise Information Systems and Implementation Stages: Learning from the Gulf Region (pp. 161-180). IGI Global
[11]Gitzel, R., Subbiah, S., & Ganz, C. 2018. A Data Quality Dashboard for CMMS Data. In ICORES (pp. 170-177).
[12]Mather, D, 2002, CMMS: A Timesaving Implementation Process, CRC PRESS, New York.
[13]Wang, R. Y., Storey, V. C., & Firth, C. P. 1995. A framework for analysis of data quality research. IEEE transactions on knowledge and data engineering, 7(4), 623-640.
[14]孫造杰，高麗君，2003年，CMMS成功實施的分析及策略研究，天津大學管理學院，河北工業大學管理學院。
[15]沈育樹，宮大川，2003年，整合性維護管理系統設計，中原大學工業工程系。
[16]曹常成，王振華，2013年，設備維護管理結合設備完整性與風險基礎檢查技術研究，勞動部勞動及職業安全衛生研究所。
[17]William W. Cato R. Keith Mobley,2001, Computer-Managed Maintenance Systems: A Step-by-Step Guide to Effective Management of Maintenance, Labor, and Inventory. Butterworth-Heinemann, Oxford .
[18]Hodkiewicz, M. R., Batsioudis, Z., Radomiljac, T., & Ho, M. T. 2017. Why autonomous assets are good for reliability–the impact of ‘operator-related component’failures on heavy mobile equipment reliability. In Annual Conference of the Prognostics and Health Management Society (pp. 2-5).
[19]Righini, R., Bottazzi, A., Fichera, C., Papastathi, M., & Perasso, L. 1998. An Innovative Software Tool for the Reliability Centered Maintenance Integrated with the Existing CMMS. IFAC Proceedings Volumes, 31(15), 987-992.
[20]O’Donoghue, C.D., Prendergast,2004, J.G.: Implementation and benefits of introducing a computerised maintenance management system into a textile manufacturing company.J. Mater. Process. Technol. 153–154, 226–232 .
[21]Zhang, Z., Li, Z., Huo,2006, Z.: CMMS and its application in power system. Int. J. Power Energy Syst. 26, 75–82 .
[22]Zbigniew Wisniewski, Artur Blaszczyk,2019, Conditioning of Computerized Maintenance Management Systems Implementation, Faculty of Management and Production Engineering Lodz University of Technology.
[23]Codd , 1970, A Relational Model of data for Large Shared Data Banks.
[24]Pipino L, Y.Lee , and R. Wang , 2002, Data Quality Assessment .
[25]董芳妘，2019，應用演算法與專家系統於往復式壓縮機的診斷，國立高雄科技大學。
[26]Cios, K. J., Swiniarski, R. W., Pedrycz, W., & Kurgan, L. A. 2007. Unsupervised learning: association rules. In Data Mining (pp. 289-306). Springer, Boston, MA.
[27]Giudici , Paolo, and Silvia Figini. , 2009, Applied Data Mining for Business and Industry. Wiley Online Library.
[28]劉凡平，2017年，大數據時代的演算法，松崗圖書。
[29]S.Aranganayagi , K.Thangavel , 2007, Clustering Categorical Data using Silhouette Coefficient as a Relocating Measure .
[30]Coates, A., & Ng, A. Y. 2012. Learning feature representations with k-means. In Neural networks: Tricks of the trade (pp. 561-580). Springer, Berlin, Heidelberg.
[31]Yang W. Lee, Leo L. Pipino, James D. Funk, Richard Y. Wang , 2009, Journey to Data Quality-The MIT Press.