Research On Collision Warning Strategy Combined With Lane Change Intention Recognition In Internet Of Vehicles Environment

The rising populace of automobile proprietors each passing year has led to a surge in traffic congestion,vehicular accidents,and an array of transportation issues.Among these challenges,those arising from vehicle collisions stand out prominently.Consequently,the exploration of collision warning systems has garnered significant attention from professionals in relevant domains.Within the realm of vehicle intelligence and networking,enhancing the efficacy of vehicle collision warning systems has provided fresh inspiration to researchers.This paper focused on the following two aspects:(1)An improved lane change intention recognition model based on long short-term memory networks in Internet of Vehicles environment was proposed.First,the long short-term memory network was selected as the benchmark model for lane change intention recognition,and the effect of data packet loss and data input frequency on the effect of lane change intention recognition was investigated by the benchmark model.Then,in order to enhance the efficacy of lane change intention recognition,a bidirectional long short-term memory network architecture,coupled with an attention mechanism,was employed to augment the model’s generalization capabilities.The comparative experimental results demonstrate that the performance of the proposed model surpasses that of the improved models utilizing either the bidirectional structure alone or the attention mechanism alone.Finally,in order to investigate whether data packet loss had an impact on the recognition effect of the improved model,a set of comparison experiments was conducted with 30%packet loss,and by comparing the model effect without packet loss scenario.The findings indicate that data packet loss has a minimal impact on the recognition effectiveness of the improved model.(2)A hierarchical collision warning model incorporating lane change intention recognition was proposed.First,the minimum safe distance was calculated based on the vehicle braking process,and then the adaptive three-level critical safety distance was obtained based on the vehicle braking intensity analysis,to develop a graded collision warning model.To validate the safety of the graded collision warning model,a joint simulation platform was constructed using Python and SUMO.Verification experiments were conducted in three different scenarios: front vehicle acceleration,uniform speed,and deceleration.Then,the combination strategy of the lane change intention recognition model and graded collision warning model was designed.To validate the viability of combining the lane change intention recognition model and the graded collision warning model,a sample of 300 lane change information sets was randomly selected from the NGSIM(Next Generation Simulation)dataset.The vehicle data at the precise moment of lane change intention recognition was then fed into the graded collision warning model to conduct a thorough safety evaluation.The results demonstrate that the integration of the collision warning model with lane change intention recognition effectively alerts drivers at the suitable safety stage.This outcome substantiates the feasibility and practicality of the proposed model.