Performance evaluation of multisource data detection system for highway tunnel traffic incidents

The traffic incident detection system is a vital tool for highway tunnel managers to monitor real-time traffic conditions inside the tunnel. Traditional traffic incident detection systems rely on a single data source, making it difficult to comprehensively reflect the real-time status of vehicles and tunnel environments. A multi-source data detection system for highway tunnel traffic incidents is proposed in this study. It utilizes the deep fusion of two technologies: image-based machine vision analysis and millimeter-wave radar perception. The traffic flow and traffic incident data of the Dayaoshan Tunnel No. 1 over the past year are analyzed. An evaluation index system for highway tunnel event detection systems and a performance evaluation method are proposed. The detection performance of the event detection system is evaluated using this method. The evaluation results show that this evaluation method has good applicability for evaluating the detection performance of highway traffic incident detection systems. It offers essential evaluation tools for enhancing highway tunnel traffic incident detection systems and holds valuable insights for elevating the operational and management standards of highways.