機車探針車行駛資料與事故資料之相關性分析

Abstract

傳統交通安全評估與易肇事地點判定皆依據歷史事故資料作為研究的基礎，且歷史事故資料蒐集期間至少為一年，而歷史事故就意味著憾事往往已經發生。相較於事故資料，替代安全指標提出以衝突的層級取代碰撞資料進行安全分析或評估，可以在短期間蒐集到足夠資料，替代安全指標已然能取代或補充事故資料來對交通事件建立危險程度評估。 在資料獲取上，本研究首先提出以聯網機車作為交交通探針車輛。將配備有傳輸感應裝置的機車投入研究範圍內交通系統中，透過行駛過程以探查交通系統的運作狀況，最終將其實時回傳的行駛資料用作為交通安全評估的可靠數據。 本研究旨在藉由機車探針車資料建立替代安全指標，以評估道路環境安全性。先提取適用機車探針車之替代性安全指標，並建立複合替代安全指標，再則是以各替代性安全指標對比於歷史事故發生的相關性驗證。於研究中提出的替代安全指標有：第85百分位速率、加速擾度、急減速事件、異常加減速事件、異常偏移事件，而複合替代安全指標便是以主成分分析的結果得出，能涵蓋上述五者的單一指標。 在空間中，對比急減速事件與事故的空間相關性結果顯示，兩者具高度相關，表示兩者在空間分布上具有明顯的重疊性。此外，在驗證替代安全指標與事故資料的相關性，本研究以事故數前高的道路交叉為目標，並以本研究提出的各替代安全指標、複合替代安全指標對照事故件數做相關分析與檢定。結果顯示，複合替代安全指標有最高相關性，與過去事故的相關程度也達到中度相關，表示該指標能夠以機車行駛行為反映道路交叉環境的安全性。

Traditional traffic safety assessment and accident-prone location determination are based on historical accident data, and the period of collecting historical accident data is at least one year, and historical accidents mean that regrettable incidents have often occurred. Compared with accident data, the surrogate safety measure (SSM) was proposed to replace crash data with conflict levels for safety analysis or evaluation. Sufficient data can be collected in a short period so that the alternative safety indicators can replace or supplement the accident data to create a safety assessment of the hazard level of a traffic event. In terms of data acquisition, the study first proposes to use the motorcycle as the traffic probe vehicle. Motorcycles equipped with a transmission sensor are put into the traffic system in the study area, and the traffic system operation is investigated through the riding process, and the real-time data is finally used as reliable data for traffic safety assessment. The purpose of the study is to establish SSMs to assess the safety of the road environment by using motorcycle probe vehicle data. The SSMs for motorcycles were first extracted, and the composite surrogate safety measure (CSSM) was established, and then we validated the relevance of each measure against the historical accidents. The SSMs proposed in the study are: 85th percentile velocity, acceleration noise, hard deceleration event, abnormal acceleration/deceleration event, and abnormal offset event, and the CSSM is a single indicator that can cover the above five by principal component analysis. In the spatial context, the spatial correlation results of comparing hard deceleration events and accidents showed a high correlation, indicating a significant overlap in spatial distribution between the two. In addition, to verify the correlation between the SSMs and the accident data, the road intersections with high accident numbers were targeted, and each SSM and the composite alternative safety indicator proposed in this study were used to correlate with the number of accidents. The results showed that the CSSM had the highest correlation and a moderate correlation with accident numbers, indicating that the measurement can reflect the safety of road intersection environment from the probe motorcycle data.