本研究以台灣電力公司年度事故災害維修資料進行資料探勘,運用Apriori演算法對事故災害狀態訊號及其他屬性加以分析,篩選出項目屬性間關聯性,且對Apriori演算法進行執行效能改善,建構新演算法資料切割排序法(Data Cutting Sorting Method, DCSM)以布林矩陣、切割資料庫及項目排序概念進行Apriori演算法效率改良,並利用台電事故維修紀錄資料庫,驗證本研究提出之資料切割排序演算法效能優於Apriori演算法。台電公司可以依據本研究成果預先瞭解事故發生之狀態訊號,並且建構事故災害預測系統來快速的解決事故,達到災害即時應變及事故維修之有效對策。
This research progressing the data mining based on the annual accident and calamity repairing data of Taiwan Power Company, and with the use of Apriori algorithm to analyze the accident and calamity’s status signals and other attributes, thus to sift out the relations between the power outage accident and its surrounding environment or status attributes, and then improving the algorithm. Furthers In this study we structure a new Data Cutting Sorting Method to use concepts of the matrix and segment the database to improve the efficiency of the algorithm, and to examine the effectiveness by case database. The Taiwan Power Company can make use of the reault of this study to construct the accident and calamity forecast system, consequently the accidents can be quickly solved, and the repairing goal of making response at once and carrying out the effective strategy can be achieved.