以資料倉儲技術為基建構大型工程資產故障診斷系統-以大型電力變壓器為例

Large Engineering Asset Fault Diagnosis System Development Based on Data Warehouse Technology-Using Large Size Power Transformer as Example

張捷中(Chang, Jie Chung)

指導教授 ： 張瑞芬

國立清華大學/工學院/工業工程與工程管理學系/碩士(2015年)

摘要

對於許多組織而言，工程資產管理已是日常管理中最重要的一部分，將會直接影響組織的生產力與可持續發展性，乃是競爭能力的重要關鍵，而在電力供應系統中，大型電力變壓器一直是最受重視的工程資產之一，若變壓器在運轉期間由於故障造成電力中斷，對經濟與民生將造成極大的影響，因此著重於提前預防與即時診斷來維護大型變壓器，早已成為企業的營運要素。然而隨著用電量的俱增大型變壓器的設置量也隨之攀升，導致變壓器相關監控歷史資料量的快速增加，故本研究提出以資料倉儲技術建立故障診斷系統，期望使用資料倉儲技術管理如此大量、即時與多變的資料。
本研究以變壓器油中溶解氣體以及各項監控參數等資料建構故障診斷系統，該系統包含前端三種診斷模組與後端資料倉儲系統，於資料倉儲系統內建立多維度資料方塊使變壓器相關的歷史數據更能彈性的被利用，並且根據不同的資料類型結合多維度運算式(Multidimensional Expressions, MDX) 在資料倉儲中進行OLAP分析後送往前端模組分析，前端模組包含數據監控模組、故障診斷模組與智慧診斷模組，於後端利用MDX語法查詢多維度表格在參數監控模組提供多維度報表以及視覺化統計圖形，在故障診斷模組中整合Institute of Electrical and Electronics Engineers (IEEE) 、International Electrotechnical Commission (IEC) 還有日本電氣協同所制訂共五種國際油中氣體分析法，並且根據各法則診斷步驟搭配MDX語法計算之統計數據，快速限縮出有故障可能之時間點有助於資產管理者提升分析速率，最後則以MDX語法查詢之多維度資料表做為投入參數，結合倒傳遞類神經網路建立智慧診斷模組，提供交叉比對之診斷分析，前端模組結合MDX語法多維度特性取代傳統關聯式資料庫SQL語法二維度資料，讓變壓器相關歷史資料達到最佳之運用。
本研究著重於資料倉儲技術整合故障診斷系統之應用，該技術適用於處理大量以及多維度類型資料，並且強調從歷史資料挖掘出有價值之資訊以提供未來決策所需，適用領域如: 製造業、行銷業務以及能源相關領域。

關鍵字

工程資產管理 ； 故障診斷 ； 變壓器 ； 資料倉儲 ； 資料探勘 ； MDX

並列摘要

For many enterprises, engineering asset management (EAM) already had been an important part of daily management now. Especially in power supply chain. If there is a shutdown during operation, the stability and productivity will suffer a huge impact. It will directly affect the competitiveness of enterprises. Therefore, focusing on early prevention and instant diagnosis to maintain large transformers is the best important EAM of enterprises. However, with the significant increase of civil and industrial electricity consumption, the number of large transformers also increased. So, in order to solve the problem such as managing the big data generate by a large number of transformers. This paper integrates a Data Warehouse technology to establish fault diagnosis system.
The system using transformer related data to establish different data cubes in Data Warehouse. These data cubes are applied for OLAP analysis in various decision support modules by using Multidimensional Expressions (MDX) code. Decision support modules including condition monitoring module, failure diagnostic module and intelligent diagnose module. Condition monitoring module will use the MDX code to provide multidimensional reports and graphical visualization of statistics. Further failure diagnostic module using three international dissolved gas analysis (DGA) methods, i.e., Institute of Electrical and Electronics Engineers (IEEE), International Electrotechnical Commission (IEC), and Electric Technology Research Association Japan (ETRAJ). Using MDX code’s statistic strength to integrate in these DGA methods’ diagnose step can help transformer maintainers find high potential faulty period. Finally, Intelligent diagnose module using MDX code to query multidimensional data cube to be the inputting parameters of Back Propagation Artificial Neural Network (BPANN) algorithm. Asset managers can diagnose potential transformer malfunctions and provide maintenance suggestions by using this system. The research methodology and system modules are evaluated and verified with real data from a series of 161 kV transformers in operations.
This research focused on Data Warehouse technology application. The Data Warehouse technology is good at processing multidimensional data and can store great amount of data. It can help analyzers understand the data before making decision. The method used in this research can widely apply in many different fields, e.g., manufacturing, marketing and energy related fields.

參考文獻

[37]林建民及陳俊瑜，2007，「以本質較安全設計探討油浸式變壓器危害預防」，勞工安全衛生研究季刊，第15卷第4期，pp. 380-398。

[47]陳冠儒，2011，「以即時監控資訊為基之變壓器智慧維護保養建議平台-以油浸式變壓器為例」，碩士論文（指導教授：張瑞芬），國立台北科技大學工業工程與管理研究所。

[50]黃舒郁、謝銘智、蕭釧瑛以及林永青，2010，「淺談資料倉儲」，中興工程季刊，2010年10月，109期，pp.63-71。

[2] Abu-Elanien, A. E.B., Salama, M. M. A., and Ibrahim, M, 2012, “Calculation of a health index for oil-immersed transformers rated under 69 kv using fuzzy logic,” IEEE Transactions on Power Delivery, Volume: 27, Issue: 4, 2029-2036.

[3] Annapoorani,K.I., and Umamaheswari, D. B., 2012,” Fault prediction based on dissolved gas concentration from insulating oil in power transformer using Neural Network,” In Properties and Applications of Dielectric Materials (ICPADM), July 24-28, Bangalore, pp. 1-4.