- 學位論文
都會地區無號誌交叉口事故因子分析與安全提升對策
Crashes at Urban Unsignalized Intersections – Factor Analysis and Safety Improvement Treatments
摘要
近十年臺灣無號誌交叉口事故涉入人數佔總事故約21%，緩解此類事故係重要的交通安全工作項目之一。儘管在相關人士努力下，我國道路交通事故仍呈成長之勢，且尚有部分未完善之處，如目前道路事故網站所呈現事故熱點，未能大量識別曾發生事故的無號誌交叉口位置，遑論對這類路口道路環境有整體性認識。而目前無號誌交叉口安全研究，(i)未系統性探討路口的優先權管制設施，以及未設號誌故無法以時間分隔路口交岔衝突情況下，更顯重要的視距議題；以及(ii) 欲研擬緩解對策前，應先了解造成事故之因子，然而對此仍未臻完全。 本研究目標為「提升無號誌交叉口安全」，並著重於都會地區。首先完成將零散事故資料有效歸類──判別何者歸屬於無號誌控制、且同一路口之事故識別為同一路口。本研究以DBSCAN聚類演算法透過經緯度分類後，再以文字紀錄校正，發展「以事故資料識別交叉口位置」的方法，解決目前未能呈現的「無號誌交叉口事故熱點」。另一方面，為了解無號誌交叉口控制型態與環境因素情形，並探討其與事故頻次的關係與分析事故因子，依路口平均一年交岔撞件數，以及未發生交岔撞的無號誌交叉口，抽樣取得四岔路口樣本，並採「路口層級」與「交岔流動層級」兩種樣本尺度，蒐集現有交通事故調查表中未記錄的控制型態之標誌、標線設置，以及有關視距的環境因素。 採古典統計方法後之數值結果顯示「行車優先權標誌(停標誌)之設置」與「行車方向視距無建築截角」均可能是都會地區無號誌交叉口交岔撞的事故因子。因此，本研究提出對應之 「路口優先權標誌設立準則」與「建築截角不足改善對策」措，包含法規與實務面探討。同時提出考量路網問題的無號誌交叉口安全改善流程，惟對路網問題僅於初探，本研究結果難以完全涵蓋尚需更多研究。
並列摘要
In the past decade, the occurrence of crashes at unsignalized intersections, accounting for approximately 21% of all accidents in Taiwan, stands out as a critical issue in traffic safety improvement. However, the current crash hotspot websites fall short in accurately pinpointing the precise locations of accidents at unsignalized intersections, thus limiting our comprehensive understanding of the road environment in these areas. Research on unsignalized intersection safety lacks a systematic exploration of (i) prioritizing control facilities at these intersections and the intersection sight distance (ISD) issues, especially in the absence of traffic signals to separate crossing conflicts, and (ii) understanding the factors leading to crash before devising mitigation strategies, which remains incomplete. This study aims to enhance safety at unsignalized intersections, with a specific focus on urban areas. To achieve this, the study begins by categorizing scattered accident data, distinguishing incidents related to unsignalized control, and identifying multiple accidents occurring at the same intersection using the clustering algorithm DBSCAN. Subsequently, text records of accident locations are utilized for further refinement and modifications. Moreover, to understand control types and environmental factors at unsignalized intersections and analyze their relationship with crash frequency, the study samples four-way intersection cases based on average number of right-angle collisions per year. Both intersections with and without collision crash are sampled at two scales: “intersection-level” and “crossing movement pair-level”. Data on control types such as signs, markings, and ISD-related environmental factors, not recorded in existing traffic accident investigation forms, are collected. Both “Stop sign placement” and “The absence of setback and truncation for buildings within the sight triangle at intersections” are potential accident factors at unsignalized intersections in urban areas. These factors are modeled using classical statistical methods. Consequently, the study proposes specific measures: “criteria for establishing priority signs at intersections” and “strategies to address inadequate building corner visibility.” Additionally, the study suggests a safety improvement process that considers network-level issues at unsignalized intersections. However, further research is needed to comprehensively address these network-level issues.