本研究的目標有三，分別為：1) 研究鋪面績效指標，2)陳述兩種鋪面平坦度預測模式，3)發展一個台灣區國道等級的鋪面績效管理資料庫。 第一部分，本文定義了浮動平坦度及表面破壞維護門檻值的設定流程。國道縮減鋪面狀況指標調查項目的合理性被探討。鑑於台灣區對於複合性鋪面績效指標發展的不足，本文提出了使用表面破損、平坦度、抗滑能力整合成反應鋪面整體績效為單一指標。特別的是，以往作為方案優選的理想解趨近法被改良為績效評量方法，其方法是藉著定義一個可行的解集合作為比較基礎。本研究所提出之複合性鋪面績效指標也與其他指標進行比較。 第二部分，本文探討兩種鋪面平坦度預測模式。其一為時間序列方法(單變數迴歸法)，該方法改良自鄧聚龍所發展的灰預測方法，並比較移動平均法、指數平滑法、及指數函數迴歸法。由誤差分析可得到所提出的方法在案例中能有較佳的模擬誤差。其二是基於因素-結果分析法(多變數線性迴歸方法)，藉著將平坦度為自變數，面層厚度、面層及路基彈性模數的比值、及累積等值單軸軸重為自變數的方法。此兩種方法並與HDM-4所建議的平坦度劣化模式相比較。 第三部分，我們發展了低成本的國際平坦度指標檢測車。架構於個人數位助理(PDA)上的鋪面狀況指標輔助檢測程式也於本文提出。最後本文採用極致軟體設計法發展一個新的鋪面績效管理資料庫，稱為高速公路績效管理資料庫(HPPMD)。HPPMD是一個目標導向的資料庫，能夠兼顧本土需求及與國際接軌之需求，並提供高速公路主管機關一個增進鋪面管理效率及品質的決策輔助工具。 The objective of this research is to (1) investigate pavement performance indexes, (2) present two pavement roughness prediction models, and (3) develop a pavement performance management database for the requirement of highway in Taiwan. First, we make a flowchart for defining floated threshold of international roughness index (IRI) and pavement condition index (PCI). The rationality of reduce of PCI survey items is discuss. Studies on the asphalt comprehensive pavement performance index method remain limited in Taiwan. In order to characterize pavement performance with a single parameter, we propose to integrate surface distress, roughness, and skid-resistance into a comprehensive pavement performance index. More specifically, we modify the TOPSIS (Technique for Order Preference by Similarity to Ideal Solution) method by defining a specific feasible solution sets as the benchmark for the aggregate pavement performance index. Comparisons between different performance indexes are then presented. Second, two models are proposed in this research to forecast the roughness of pavement. The first approach is a time-series model (univariate regression method) modified the grey model developed by Deng. Compare this model with the moving average method, the exponential smoothing method, the traditional grey model, and the exponential regression model, it can be shown that this proposed approach generates minimum prediction errors. The second approach is based on the cause-result analysis method (multiple variables linear regression method), where the prediction model is derived by regressing pavement roughness on the thickness of surface layer, the ratio of asphalt layer over sub-grade resilient modulus, and the accumulative equivalent single axle load. These two models are compared with the roughness-deterioration model suggested by HDM-4. Finally, we propose a low-cost vehicle for surveying IRI directly. Auxiliary software based on personal digital assistant is developed for surveying PCI quickly. We also adopt the extreme programming method in the development of a new database called Highway Pavement Performance Management Database (HPPMD). HPPMD is an object-oriented database whose data structure fits into both domestic requirements and international standards. The objective of HPPMS is to provide the highway authorities a decision assistant tool to improve the efficiency and quality of pavement maintenance.
