The research on indirect incident detection in expressway under high traffic density

Abstract

With the continuous construction of expressways in China, the timely process of traffic incidents has been increasingly valued to avoid heavy traffic and secondary incidents. To timely eliminate the terrible influence of incidents, the indirect incident detection method is designed and presented based on the actual data of the segment from the Lingdian interchange to the Tangwang interchange hub of the Jinan-Guangzhou expressway in this paper. Firstly, the existing data from diverse detection systems are collected and integrated into a unified format. Then, an actual-data-based simulation experiment is achieved to offer the required information to the indirect incident detection method through VISSIM. On this basis, a series of parameters are constructed to identify the mutation of traffic operation from multi-dimension. Finally, the indirect incident detection method is trained to recognize the mutation of traffic and classify the normal scene and incident scene with LightGBM. Compared to KNN, RF, and SVM, the LightGBM has an excellent performance on incident identification, with a 98.6% accuracy, a 100% precision, an 88.9% recall, and a 94.1% F1 score.