Improved Driving Risk Field – based identification of vehicle interaction risks in the upstream transition area of expressway work zones

In order to solve the problem that it is difficult to identify the vehicle interaction risk in the upstream transition area of expressway work zones, this paper proposes a vehicle interaction risk identification model based on the Improved Driving Risk Field. In this paper, the traffic flow data in the upstream transition area is collected through the fixed-point camera, and the vehicle trajectory data is output by simulation using VISSIM software to screen the pairs of following vehicles and lane-changing vehicles. The IDRF model in this paper comprehensively considers the characteristics of the upstream transition area, driver risk factors, vehicle dynamic risk differences and vehicle virtual risk bands, and calibrates the parameters using genetic algorithms and Poliakov models. The performance of the IDRF model is validated by comparing the performance of IDRF, DRF (Driving Risk Field), THWI (Time Headway Inverse) and TTCI (Time To Collision Inverse) models in vehicle interaction risk identification. The results show that the IDRF model has the best identification performance. Among them, the IDRF model shows higher consistency with the THWI model in the follow-through interaction risk identification. In lane change interaction risk identification, the IDRF model improved the risk identification rate by 5.21% compared to the DRF model, 49.66% compared to the THWI model, and 81.04% compared to the TTCI model. Based on the risk visualisation results of the model in this paper, the traffic management department can improve the safety supervision level of expressway maintenance and construction operations.