Research on Lane-Changing Model for Intelligent Vehicles Based on Sparrow Search Algorithm

Lane changing behavior is one of the typical driving behaviors in micro traffic flow. In this paper, lane change decision was studied in order to reveal the driver's decision-making behavior characteristics. Through the analysis of experimental data from the NGSIM project of the Federal Highway Administration of the United States, a substantial amount of traffic data and lane change trajectories were extracted. This paper proposes a lane change decision model based on XGBoost optimized by the Sparrow algorithm. After simulation comparison verification, the results show that the established lane change decision model has significantly improved in all performance indicators, with an evaluation accuracy of 93.20%, which is significantly better than the unoptimized XGBoost model (87.68%). This finding confirms that the SSA-XGBoost model has a higher level of accuracy, stability and effectivity compared to the traditional XGBoost model.