高速公路高風險情境之關鍵影響因素分析──以國道客運駕駛輔助警示及異常事件為基礎

INVESTIGATING THE KEY FACTORS CONTRIBUTING TO FREEWAY ACCIDENT-PRONE SCENARIOS BASED ON ADAS WARNINGS AND RISK EVENTS OF FREEWAY BUSES

以往研究在指認道路高風險情境之關鍵影響因素，大多以事故資料為基礎。惟道路事故具隨機性與稀少性，須長期觀察，故較難發揮杜漸防微之效。鑑於駕駛輔助警示系統（ADAS）日益普及，如何用以檢視高風險情境之關鍵影響因素，為一重要課題。基此，本文以H客運公司200趟次國道客運資料為基礎，利用負二項模式指出影響各路段及趟次所發生警示及異常事件頻次之關鍵因素，再利用次序普羅比模式分析影響警示事件危險程度之關鍵因素。結果顯示（以未保安距為例），影響發生頻次的關鍵影響因素為：總交通量高、小車比例高、行車趟次多。影響危險程度則為上坡、時速高、南下（返程）、排班趟次，以及晨峰。相關結果可供道路主管機關提升道路交通安全之參考。

Most of previous studies identify the high-risk scenarios of road networks based on traffic accidents. However, due to the scarcity and randomness of accidents, the studies require a long-time observation and is nearly impossible to investigate the newly formed risk scenarios proactively. With the rapid growth of ADAS adoption, how to use of the crash surrogate data to investigate the risk factors is essential. Accordingly, based on the data 200 freeway trips of the bus company A, this study uses of a negative binomial model to investigate the key factors forming the high-risk freeway segments and bus trips based on the frequency of warnings and risk events. Additionally, an ordered probit model is used to examine the key factors causing the risk levels of a warning events. The estimation results (a case of unsafe distance events) show that high traffic volume, high ratio of passenger cars and high number of trips are key risk factors. Meanwhile, for the risk level of unsafe distance warnings, the warnings activated at the site with upgrade, higher speed, southbound (return trip), high number of trips and in the morning peak hours would have higher risk level. The identified risk factors helpful for road authorities to improve traffic safety.