| 摘要: | 橋梁是道路的樞紐，扮演著運輸的重要角色。近年來多數橋梁的橋齡漸增，橋梁本身狀況也隨之劣化，有必要採取維修或補強措施，但橋梁劣化之時間點並不容易預估，這也是導入預防性維護觀念之重要議題之一。為了掌控橋梁劣化狀況，各國目前均建置橋梁管理系統，並訂定檢測方法與機制，使橋梁歷史之維護與檢測紀錄可以完整保存，但利用歷史資料庫進行後續劣化加值分析之研究仍十分有限。導致橋梁劣化之因素眾多，除了橋梁自身的材質、構造形式外，與外部環境也有相當程度的關聯，例如交通流量或是降雨量等環境因素。本文即利用美國自1992年起建立之National Bridge Inventory (NBI) 資料庫進行資料探勘，該資料庫除了蒐集各州橋梁基本資料外，也完整記錄每一年檢測的歷史紀錄，是目前資料最為豐富且可公開取得之橋梁歷史資料庫。本研究透過K-Means、Two -Steps分群方法與C5.0分類方法進行分析，可將橋梁歸納成數個不同橋群，並描繪橋梁劣化分群與其規則的決策樹，續以關聯分析Apiori進行構件間劣化關聯，未來橋梁養護人員可根據該決策樹結果，清楚了解轄內橋梁可能所屬之群組，並對應其可能之劣化，提早因應，導入預防性維護管理之概念，使災害防患於未然，除了可以保障用路人之安全，更可以延長橋梁使用壽命，樽節橋梁公共工程維護預算。;Bridge is the interjunction of the roads and plays a crucial role in transportation. In recent years, the age of most bridges in Taiwan has gotten old gradually; as a result, the bridges deteriorate and need to be fixed and maintained. Nevertheless, it is not easy to predict the timing of bridge deterioration, and such situation has become one of the important issues in introducing the concept of preventive maintenance. In fact, in order to control the condition of bridge deterioration, many countries have set up bridge management system and regulated methods and mechanism of inspection. The record of bridge maintenance and inspection history is therefore well preserved, but the number of researches utilizing value added analysis of bridge deterioration with database of history is limited. There are many factors that contribute to bridge deterioration. In addition to the materials and forms of elemention of the bridges, it is associated with external environments as well, including environmental factors as traffic flow or rainfall. The present paper adopts National Bridge Inventory (NBI) established by USA since 1992 to process data mining. Currently, the database is the most abundant and publicly accessible database of bridge history, not only collecting basic information of bridges in each state, but keeping a complete record of the inspection history every year. This research processes analysis by means of Two –Steps, C5.0, and Apriori categorizing bridges into several different clusters and depicting the decision-making tree of clustering and association rules of bridge deterioration. In the future, bridge maintenance personnel can gain a clear idea of the cluster of the bridges they are responsible for according to the results of the decision tree. They can also cope with possible deterioration in advance by introducing the concept of preventive maintenance management. Consequently, the disasters can be avoided, the safety of pedestrians can be ensured, and the bridge maintenance budget of public engineering can be economized. |
