| 摘要: | 台灣近年來由於經濟蓬勃發展，交通運輸量日益繁重，機場鋪面之負荷亦愈形沈重，故如何延長及維持鋪面之服務績效，已成為交通工程單位重要課題之一。然以目前空運幾近飽和之狀況，已不容長時期關閉機場重新鋪築，因此對破損的鋪面，如何能即時、有效的予以修復，已是目前機場鋪面績效維護主要的議題。 目前國內對於機場鋪面損壞的維修，大多依賴維修專家經驗的累積或由材料廠商建議適合的修補材料，尚無一套完整實用的鋪面維修規範可供遵循；對於鋪面損壞原因未適當的診斷和瞭解即從事維修，或採用不當的維修方法和維修材料，是國內機場鋪面維修效率偏低之主要原因；另鋪面維修後缺乏完整的檔案記錄，致使維修經驗無法累積，進而無法評估維修方法及維修材料的成效，導致經常發生重複失敗的維修案例，亦是目前國內機場鋪面維修成效不彰之主要原因。 本研究旨在建立一套結合匯集機場鋪面維修專家知識和經驗及具有處理、儲存、展示機場鋪面的屬性資料及空間資料之「機場鋪面維修系統」。首先應用模糊邏輯推論，以專家的知識為基礎，建立機場鋪面損壞原因診斷專家系統模糊推理模式，使系統能處理專家對機場剛性鋪面損壞現象在程度上確認之不確定性的問題，以改善傳統專家系統的限制，診斷出較合理、正確的原因。另應用類神經網路機器學習理論，以專家問卷調查的方式，經由案例的學習與歸納，累積知識，以利日後針對現地破損進行建議維修時，更能選擇出適合的策略，確保鋪面服務績效。另考量維修材料具有創新性及適用性之問題，本研究同時建立維修材料回饋學習的功能，讓系統能隨時間不斷學習，確保維修材料的適用性。 本研究並因應機場營運作業特性，研發一套整合GIS及GPS之鋪面調查方法，可進行機場鋪面全面性及臨時性之鋪面調查工作。 本研究所建構之機場鋪面維修系統，初步已具有診斷鋪面損壞原因、建議維修策略及儲存、展示每一版塊歷次維修紀錄的功能；所研發之鋪面調查法，經現地實際使用，亦較傳統鋪面調查法便捷快速；期望本研究能對國內機場鋪面維修效益有正面提昇的作用。 Due to the economic boom in Taiwan in recent years, transportation and communication increases daily and the airport pavement load also becomes heavier. How to maintain and extend the life of pavement has now become one of the major topics in the transportation engineering department. However, since air cargo is almost at its saturation, it is not advisable to close the airport for a long period of time to repave the field. Therefore, how to repair the distressed pavement in time and effectively has now become the major issue of maintaining the pavement at the airport effectively. Repair of distressed pavement at the airport in Taiwan at present mostly relies on the accumulated experience of repair experts or on the repair material suggested by the material supplier(s). There is still no set of perfect and practical pavement repair criteria to be followed. The main reason for poor pavement repair at the airport in Taiwan was that repairs were done without understanding and diagnosing the cause of distress or adopting improper repair method and materials. Also, it lacks a complete file record after repairing the pavement, so there is no way of accumulating repair experience and unable to evaluate the results of repair methods and materials. As a result, there is frequent repetitious repair failure, which is also one of the main reason of poor pavement repair at the airport in Taiwan. The purpose of this study is to create an airport pavement repair system, that is able to process, store, and show the attribute data and spatial data of airport pavement by collecting and integrating the knowledge and experience of airport pavement repair experts. First, fuzzy logic was applied to do the inferring, based on the knowledge of experts, and create a fuzzy inferring mode of diagnostic expert system for the cause of pavement distress at the airport, to enable the system to process uncertain problems of experts on the degree of rigid pavement distress at the airport, in order to improve the restrictions on conventional expert system and thus diagnose a more rational and accurate cause. The machine learning of neural network theory was also applied in the experts’ questionnaire survey, and then study and classify the cases in order to accumulate the knowledge for the advantage of giving suggestions to repairs on field distress in the future and be able to choose an appropriate plan in order to ensure effect of pavement service. Innovative and appropriate repair materials were also taken in account. The function of feedback learning on repair materials was simultaneously created in this study to enable the system to keep on learning any time in order to ensure the suitability of repair materials. In this study a set of GIS & GPS pavement survey method was also developed to conduct a total and temporary pavement survey on the airport pavement in order to deal with the characteristic of business operation at the airport. Initially the airport pavement repair system that was constructed in this study already has the function of diagnosing the cause of pavement distress, suggesting repair plan, and storing and showing the subsequent repair records of each block. Upon practical field application, the pavement survey method developed is more convenient and faster than the conventional pavement survey method. |
