THE KEY TECHNOLOGIES FOR RISK IDENTIFICATION OF "TWO PASSENGER AND ONE HAZARDOUS" VEHICLES BASED ON MULTI-SOURCE DATA FUSION
Keywords:
Two Passenger and One Hazardous(TPOH), Multi-source data fusion, Risk identification, Seasonal frozen region, XGBoost-LSTMAbstract
"Two passenger and one hazardous" (TPOH) vehicles are primary targets of road traffic safety supervision. Jilin Province, located in the northeastern seasonal frozen region of China, experiences severe winter conditions, including icy and snow-covered roads and extremely low temperatures, which substantially increase the operational safety risks of these vehicles.This study proposes a risk identification framework based on multi-source data fusion. The framework integrates GPS positioning data, on-board OBD data, meteorological information, and road infrastructure data to construct a multidimensional risk identification model. The model captures five categories of abnormal driving behaviors: speeding, harsh acceleration, harsh braking, vehicle vibration, and abrupt lane changes.An improved sliding-window feature extraction method and a hybrid XGBoost–LSTM classification model are employed to achieve accurate vehicle risk identification under complex seasonal frozen conditions. In addition, a system management and control platform is developed to provide decision-support tools for the transportation authorities of Jilin Province.References
[1] Li C, Yang Y, Cao G. Quadruple-U Auxiliary Structure-Based Receiving Coil Positioning System for Electric Vehicle Wireless Charger. World Electric Vehicle Journal, 2023, 14(5): 115.
[2] Ma W, Yuan J, An K, et al. Route flow estimation based on the fusion of probe vehicle trajectory and automated vehicle identification data.Transportation Research Part C:Emerging Technologies, 2022, 144: 103907.
[3] Qi X, Ji Y, Li W, et al. Vehicle trajectory reconstruction on urban traffic network using automatic license plate recognition data. IEEE Access, 2021, 9: 49110-20. DOI: 10.1109/ACCESS.2021.3068866.
[4] Yao Z, Liu M, JiangY, et al. Trajectory reconstruction for mixed traffic flow with regular,connected, and connected automated vehicles on freeway. IET Intelligent Transport Systems, 2024, 18(3): 450-66. DOI: 10.1049/itr2.12457.
[5] Huang SE, Feng Y, Liu HX. A data-driven method for falsified vehicle trajectory identification by anomaly detection. Transportation research part C: emerging technologies, 2021, 128: 103196.
[6] He L, Niu X, Chen T, et al. Spatio-temporal trajectory anomaly detection based on common sub-sequence. Applied Intelligence, 2022: 1-23. DOI: 10.1007/s10489-022-04330-3.
[7] Li C, Feng G, Li Y, et al. DiffTAD: Denoising diffusion probabilistic models for vehicle trajectory anomaly detection. Knowledge-Based Systems, 2024, 286: 111387. DOI: 10.1016/j.knosys.2024.111387.
[8] Zhang Y, He Y, Zhang L. Recognition method of abnormal driving behavior using the bidirectional gated recurrent unit and convolutional neural network. Physica A: Statistical Mechanics and its Applications. 2023, 609: 128317. DOI: 10.1016/j.physa.2022.128317.
[9] Ma Y, Xie Z, Chen S, et al. Real-time detection of abnormal driving behavior based on long short-term memory network and regression residuals. Transportation research part C:emerging technologies, 2023, 146: 103983. DOI: 10.1016/j.trc.2022.103983.
[10] Shi Y, Wang D, Tang J B, et al. Detecting spatiotemporal extents of traffic congestion: A density-based moving object clustering approach. International Journal of Geographical Information Science, 2021, 35(7): 1449-1473. DOI: 10.1080/13658816.2020.1862853.