📄 中文摘要
次生碰撞概率预测是主动交通管理系统的重要组成部分，旨在缓解因次生碰撞造成的拥堵和不利影响。然而，现有方法主要依赖于后碰撞特征（如碰撞类型和严重性），这些特征在实时情况下很少可用，限制了其实际应用。为了解决这一限制，提出了一种不依赖于后碰撞特征的混合次生碰撞概率预测框架。该框架设计了一个动态时空窗口，从初次碰撞位置及其上游路段提取实时交通流和环境特征。框架包括三个模型：一个初次碰撞模型，用于估计次生碰撞的可能性。
实时次生碰撞概率预测不依赖于初次碰撞特征
📄 English Summary
Real-time Secondary Crash Likelihood Prediction Excluding Post Primary Crash Features
The prediction of secondary crash likelihood is a crucial component of an active traffic management system aimed at mitigating congestion and adverse effects caused by secondary crashes. Existing methods primarily rely on post-crash features, such as crash type and severity, which are rarely available in real-time, thus limiting their practical applicability. To address this limitation, a hybrid secondary crash likelihood prediction framework is proposed that does not depend on post-crash features. A dynamic spatiotemporal window is designed to extract real-time traffic flow and environmental features from primary crash locations and their upstream segments. The framework consists of three models: a primary crash model to estimate the likelihood of secondary crashes.