Adaptive Freeway Incident Detection Algorithm Using the Hilbert-Huang Transform

Automated detection of incidents is still an integral component of advanced traffic management systems (ATMS) and its role for the effective management of freeway operations should not be ignored. This paper presents a novel incident detection algorithm with adaptive thresholding capabilities based on the Hilbert-Huang transform (HHT). The HHT is a powerful tool for processing nonstationary data, and it employs the concept of instantaneous frequency; hence it is suitable for local analysis of traffic measurements. In particular, this paper demonstrates how the empirical mode decomposition (EMD) and Hilbert spectral analysis (HSA) components of HHT could be used to detect traffic incidents on freeways. Evaluation of the proposed algorithm was conducted using real-world traffic data with the California and low-pass filter (Minnesota) as baseline. The test results indicate that the HHT-based incident detection algorithm remarkably outperforms the benchmark algorithms with the highest detection rate (95.8%) and the lowest false alarm rate (0.001). This demonstrates the potential for practical application of the proposed algorithm in reality.