國內剛性鋪面數量雖然較柔性鋪面少，但大多分布於機場、高速公路收費站、隧道內等重要交通處所，故在交通運輸上扮演相當重要且關鍵的角色；但剛性鋪面因長期使用，難免會產生破損，這些破損若未能即時有效的予以修復，任其惡化，將增加後續維修的困難度、維修經費等，甚至影響交通安全。所以如何快速、有效的修復損壞鋪面，是鋪面維護人員相當重要的課題。 剛性鋪面的維修問題，牽涉的因素相當複雜，目前仍需依賴有經驗的鋪面維修專家作適當的決策和判斷，但專家在決定維修策略時，必須同時考慮眾多之不確定性因素，而這些因素可能直接或間接，單獨或混合的影響維修策略的選擇。為讓專家在選定維修策略之前，能更客觀、正確的先診斷出造成鋪面損壞的原因，進而針對鋪面現況及維修需求等因素，選定適當之維修策略(維修工法及維修材料)，本研究首先應用模糊邏輯推論，以專家的知識為基礎，建立機場剛性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，進而協助選定適當之維修策略。 維修策略的選擇是機場鋪面管理系統中重要的一環，適當的維修策略，是確保維修成效的保證。通常鋪面維護人員會依照專家的經驗與知識，並根據現地現況，找出造成鋪面破損的原因和建議維修策略；但在實務上常看到維修後再損壞的例子，顯示先前選擇的維修策略不能有效修補鋪面，而必須改用其他工法或材料維修，才能確保維修後鋪面服務績效。本研究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。 本研究所建立之智慧型機場剛性鋪面維修專家系統，除由專家實際操作驗證外，為求證系統分析結果與機場剛性鋪面專家解決鋪面維修問題的思考邏輯模式之差異比較，進行系統分析，專家驗證和系統分析結果，印證本系統在機場剛性鋪面維修策略之選擇時，能迅速的評選出適當之維修策略，達到提昇機場剛性鋪面之維修成效及維護飛航安全之目標。 Although the volume of rigid pavement in Taiwan is less than the flexible pavement, you will often find they are located at very important traffic spots such as airports, highway toll stations, interior of tunnels, etc. They often play an important role in pavements. After finishing construction and opening to public acess, it is easy to find some damage on rigid pavement. If we don’t deal with the problem properly, they will become serious and difficult to maintenance. Properly maintenance is an important subject to engineers. Maintenances of rigid pavement include a lot of complex factors; we often deal with them depending on experienced experts. Experts will often face to uncertain factors and hard to make a decision on the maintenance strategy. In this study, we have first applied a theory of fuzzy logic to establish a logic-reasoning mode of diagnosing expert system for rigid pavement damages assessments at the airport. We have used expert experiences and fuzzy reasoning to establish a diagnosing expert system. It is desired to help engineers make an accurate decision. To select repairing strategy is one of the most important tasks in the airport pavement management system. The performances of pavement are assured by suitable repairing strategy. We couldn’t efficiently repair damage of pavement by expert’s suggestions. The study also has taken advantage of a machine learning theory of neural network, expert questionnaire, and through special case study to develop suggestive expert system. The system offers us a properly and economic suggestions, it also have feedback learning function on repair materials. Experience experts have qualified the system. They have the same conclusion it will be useful for maintenance of rigid pavement.
