運輸設施(如：鋪面與橋梁)的維護，對於行車品質與安全而言都是一個重要的課題，因此為了要科學化分析鋪面狀態的維修與養護，鋪面生命週期的最適控制是最新趨勢，然而傳統的施工流程往往與鋪面狀態的最佳化分析相互衝突，其中有三個明顯的限制，第一：工程單位通常採用人工量測以及維修門檻作為維修策略維修，使得最佳化之維修方案常常難以被接受，導致最佳化模式產生的策略與實際養護單位工作流程造成不相容的情形；第二：通常最佳化數學模式不易考量不同的劣化模式，不同的劣化趨勢維修效果，卻都用相同的模式進行分析，這會導致決策上的錯誤，造成生命財產的損失；第三：設施與設施之間的關聯性往往缺乏考量，因此設施之間車流的轉換往往被忽略，造成更多預測的誤差。 因此本研究採用了動態混合模式，以解決上述的三項缺失，首先針對動態混合模式 之各個系統進行介紹，之後再將邏輯運算式轉換成數學限制式，接著加入交通指派，以建構一雙層問題。由於整個探討分析的問題無法直接求解且規模過大，因此利用啟發式解法中的禁忌搜尋演算法將問題分層考慮，並且將問題依據時間切割以加速求解。最後，使用實例分析不同成本下對於目標鋪面的最佳維修門檻，並比較分析不同成本下的最佳門檻，來測試本研究提出之方法。 本研究綜合最適控制與最佳養護門檻策略，為鋪面維護提供有效的決策方法，使得 在生命週期成本能達到最小化，並且可以應用同樣的方法於交通運輸設施，以幫助施工人員在養護設施時進行合理的決策與評估。 Transportation infrastructure management deals with maintenance decision-making of transportation facilities such as pavements and bridges. Various methodologies have been adopted to determine the optimal allocation of limited funds over multiple-periods. Optimal control is the state-of-the-art methodology for maintenance decision-making. However, the methodology has three major limitations. First, most of the models adopting optimal control generate highly detailed plans that contain maintenance policy for every facility at every time period. These plans are not well-accepted by transportation agencies because they are incompatible with their workflows for pavement management, i.e., threshold-based maintenance. Second, these models are difficult to solve so simple deterioration and maintenance effectiveness models are generally assumed. Third, the interdependencies between facilities are often ignored. Therefore, to address the above limitations, this paper formulates the problem as a dynamic hybrid model to find the optimal thresholds considering available budgets. Dynamic hybrid systems are easy to formulate with logical statements but logical statements are difficult to optimize. The conversion between the logical statements and mathematical programming is thus discussed. The user equilibrium for traffic flow is also introduced in the problem to consider facility interdependency. As a result, a bi-level programming problem has to be formulated. The problem is solved with Tabu search algorithm. Further, the technique of decomposition is used to speed up the solution process. Finally, to test the methodology, a numerical example is conducted to find the optimal thresholds under different budgets.
