良好的運輸設施狀態不但能提高國民的生活品質，更能促進社會經濟之繁榮。為了讓設施狀態都能維持在最佳狀態，必須以生命週期管理的方法來做維護，並採用最佳化的規劃方式來產生維護策略。但是以最佳化的方式所產生的預擬計劃策略較難被實務單位接受，通常實務上的作業流程須先經由儀器測量設施的狀態值，再將設施的狀態值交由決策者決定維護方式，其所使用的維護方式訂定流程與一般使用最佳化所產生的預擬計畫策略有很大的落差。為了實現最佳化的規劃方式並避免與實務單位的施工流程發生衝突，本研究使用維修門檻值策略來達成此一理想。維修門檻值策略可提出一組維修門檻供實務單位作為決定設施維護動作的標準，如此將可接近一般實務單位的維護流程。
本研究使用動態混合模式來建立門檻值維修策略之模型，動態混合模式包括：事件產生器、有限狀態機、模式選擇器及切換系統，使用此一模式建立的模型可同時考量不同狀態下的設施劣化方式以及不同維護方式所得到的維護效果，但由於維修門檻值策略在求解大規模設施時難以在有限時間內利用混合整數規劃求解，因此本研究撰寫分支界限法、貪婪演算法與遺傳演算法之程式來輔助求解維修門檻值策略。其中遺傳演算法尤為重要，其全域搜尋特性可保障問題求得較好的可行解。
最後以桃園縣大溪鎮的路網作為實例驗證的目標，探討不同演算法用於求解不同資料規模之維修門檻值策略的效果與差異，並由測試所得的結果整理出結論與建議，以供實務單位做為決策時的參考。

The condition of transportation infrastructure is very important for the development of a country. If the condition of transportation infrastructure became better can not only improve people's quality of life but also can promote the socio-economic prosperity. In order to improve the functionality of transportation infrastructure, the concept of life-cycle management has been introduced to the infrastructure maintenance decision-making process. Actually, transportation agencies measure the conditions of transportation facilities by instrument or manual inspection, and then make maintenance decisions based on the measurements and maintenance thresholds. However, most of the optimization methods generate pre-determined maintenance plans, which are incompatible with the workflow of the transportation management agencies and difficult to interpret. As a result, the pre-determined maintenance plans are not well accepted by transportation agencies in practice. This study uses threshold-based maintenance method, which can generate a set of thresholds, to achieve the optimal methods and to help transportation management agencies determine the facility maintenance actions as standard.
Considered the different deterioration and maintenance effects, this study uses hybrid dynamic mode (HDM) to describe the transportation infrastructure. The dynamic hybrid mode includes event generator, finite state machine, mode selector and switched systems. However, HDM is difficult to solve for large problems. Branch and bound, greedy algorithm, genetic algorithm is developed to increase the efficiency of HDM.
The road network of Dasi Township, Taoyuan County, is used as an example to verify the proposed methodology. Various levels of data are tested to analyze the differences between greedy algorithm and genetic algorithm. Finally, the conclusions and future directions for research are provided.

1. American Association of State Highway and Transportation Officials (1993). AASHTO Guide for Design of Pavement Structures 1993. American Association of State Highway and Transportation Officials, Washington, DC.
2. Bemporad, A. and Giorgetti, N. (2006). Logic-based solution methods for optimal control of hybrid systems. Automatic Control, IEEE Transactions, 51(6), 963 - 976, Siena Univ., Italy.
3. Chu, J. C. and Y.-J. Chen (2012). Optimal threshold-based network-level transportation infrastructure life-cycle management with heterogeneous maintenance actions. Transportation Research Part B.
4. Johnson, E. L., Nemhauser, G. L. and Savelsbergh, M. W. P. (1997). Progress in Linear Programming-Based Algorithms for Integer Programming. INFORMS Journal on Computing, 12(1), 2 - 23.
5. Lam, K. M. and Yan, H. (1994). Fast greedy algorithm for active contours. Electronics Letters, 30(1), 21 - 23.
6. N.D. Lea International Ltd (1995). Modelling Road Deterioration and Maintenance Effects In HDM-4. RETA 5549-REG Highway Development and Management Research Final Report.
7. Paterson, W. D. O. (1990). Quantifying the effectiveness of pavement maintenance and rehabilitation. Proceedings of the 6th Road Engineering Association of Asia and Australia Conference.
8. Prozzi,J.A (2001).Modeling pavement performance by combining field and experimental data. Transportation Research Record 1699, Transportation Research Board, National Research Council, Washington, D.C.,87-94.
9. Ricardo, O.D.S., Silvrano D.N., and Marcio M. D. F.(2003). Improving Pavement With Long-Term Pavement Performance: Products for Today and Tomorrow. U.S. Department of Transportation, publication NO. HRT-06-109.
10. Sayers, M. W. and Karamihas, S. M. (1998).The Little Book of Profiling-Base Information about Measuring and Interpreting Road Profiles, University of Michigan, Ann Arbor.
11. Underhill, L. G. (1994). Optimal and suboptimal reserve selection algorithms. Biological Conservation, 70(1), 85 - 87.
12. Waxman, B. M. (1988). Routing of Multipoint Connections. IEEE Journal on selected areas in communications, 6(9), 1617 - 1622.
13. 李明憲(2012)，運輸設施生命週期管理維護策略訂定之研究，國立中央大學土木工程學系碩士論文，桃園縣中壢市。
14. 張其教(2001)，柔性路面網級養護管理維修系統建立之研究，國立中央大學土木工程學系博士論文，桃園縣中壢市。
15. 鼎漢國際工程顧問股份有限公司(2006)，桃園都會區整理運輸系統分析及需求模式之建立與應用期末報告書。