電子巡檢系統的改善方案－以A公司為例

A Study on Electronic Patrol System Improvement – The Case of Company “A”

梁羿翔(YI-HSIANG LIANG)

指導教授 ： 范錚強

國立中央大學/管理學院/資訊管理學系碩士在職專班/碩士(2014年)

摘要

電子巡檢系統的改善方案－以A公司為例
摘要
由於水泥產業幅員遼闊且工作環境充斥著粉塵，因此無法藉由廣設監視器，達到全面監控廠區的目的，同時為了瞭解設備目前的運轉狀態是否正常並預防環安問題，故巡檢作業成為管理廠區勢在必行的例行項目。傳統的巡檢作業，是以紙本記錄的方式進行，容易衍生巡檢人員未實際到達指定地點、隨意填寫內容以及主管事後審核不易等問題，因此，A公司決定導入電子巡檢系統。
電子巡檢系統的導入，改善傳統紙本巡檢未能落實的缺點，但隨著電子巡檢記錄資料量巨幅成長，系統開始出現效能上的瓶頸，舉凡報表查詢功能、PDA下載任務與上載紀錄等等，皆出現影響，為了確保巡檢作業能夠順利進行，廠端開始要求解決問題。另一方面，由於先前電子巡檢系統導入的有著顯著的效益，因此A公司決定全面導入電子巡檢系統，截至目前為止共計有11個分廠導入該系統。但因為欠缺一套整合各個分廠巡檢記錄的系統，A公司的總管理處無法立即掌握全場的巡檢狀況，所以必須逐一查閱每個分廠的巡檢資訊而花費了相當多的時間。此外，總管理處的高階主管，亦相當重視巡檢作業，但人工彙整各廠的巡檢記錄過程太過繁瑣，往往只能隔日提供，不符合主管對資訊取得即時性、透明化的要求。
本改善方案是以解決各個分廠的資料庫效能為首要目標，採行的作法為將現行資料庫分割為 OLTP與 OLAP 架構，一方面限制OLTP資料表增長，降低笛卡兒乘積的組合進而提昇整體電子巡檢系統的效率。另一方面使用ETL工具自動將各分廠的 OLAP資料彙整回總處，實現建置巡檢紀錄單一窗口的資料來源目標，並且進一步的完成商業智慧模型的開發。最後採用行動裝置方案，主動告知並提供主管即時之巡檢紀錄，大幅提升資訊獲得的效率。

關鍵字: OLAP、ETL、Cube、行動裝置

關鍵字

OLAP ； ETL ； Cube ； 行動裝置

並列摘要

A Study on Electronic Patrol System Improvement – The Case of Company “A”
Abstract
In the cement industry, it is necessary to monitor the operating status of facilities in order to ensure smooth operations and to prevent environment safety hazards. However, a typical cement factory occupies a vast territory, and is faced with severe level of dust. Therefore, it is impractical to deploy video surveillance systems to monitor the entire factory. As a result that manual patrol procedures have to be implemented. The traditional patrol procedure requires all records be written and kept on paper. There are many potential problems with the paper-based system: it is impossible to ensure that the assigned staff was at right place at the right time as expected, and filled in the requisite data; and it is difficult for supervisors to audit the patrol records. Hence, company “A” decided to deploy an electronic patrol system to solve the problem.
The initial success resulting in an all out deployment covering all 11 factories in Company “A”. While the electronic patrol system result in vast improvements, it generates a tremendous amount of patrol records, causing performance bottle necks. The performance problems are multifold, including data aggregation, report generation, response to inquiry, excessive time for server/PDA connections (downloading patrol assignments and uploading patrol records, etc. In addition, the patrol data from the 11 plants are not integrated, thus management in the headquarter has to access the data one factory at a time. Moreover, the data is so complicated that daily reports are only available on the next day, making realtime visibility a much desirable feature.
This study aims to provide a solution to integrate the independent databases across all factories, and to solve the database performance issue. First, we propose to establish a new OLAP structure, along with the current OLTP database, so as to limit OLTP table data growth, and to reduce Cartesian product in database joins. Second, we propose to use ETL tools to push OLAP table from each plant to headquarter for automatic data aggregation. Third, the development of business intelligence features are also proposed. Finally, we employ mobile devices to provide senior managers with realtime patrol records, so that realtime visibility can be achieved.
Keyword: OLAP, ETL, Cube, mobile device.

並列關鍵字

OLAP ； ETL ； Cube ； mobile device

參考文獻

1. Berry, M. & Linoff, G (1997), Data Mining Techniques. New York: Wiley.

2. Han, J. & Kamber, M. (2001), Data Mining: Concepts and Techniques. Morgan Kaufmann Publishers.

2. 林于新 (2009)，以企業流程模型導向實施資料庫重構之研究－以S公司為例，國立政治大學資訊管理研究所碩士論文。

4. 李卓翰 (2006)，資料倉儲理論與實務，台北：學貫行銷股份有限公司。

【英文文獻】