Based on MLP-SVM Lane Change Conflict Prediction Model and Analysis of Influencing Factors

In order to study the characteristics of lane change conflict of vehicles on urban expressways and reduce the risk of lane change conflict, the significant factors causing traffic conflict in the characteristics of lane change behavior were explored. Firstly, the microscopic lane change trajectory of vehicles on urban expressway is extracted, and secondly, the judgment index of lane change traffic conflict is proposed, considering the traffic conflict metrics of time and space dimensions. Considering the information of vehicles in close proximity to lane change, focusing on the vector of vehicle lane change behavior, improving and expanding the traditional TTC theory, constructing a lane change conflict risk identification model, and subdividing the conflict risk level. Based on the MLP-SVM algorithm, the prediction of vehicle lane change conflict is compared with the performance of Support Vector Machine (SVM), Multilayer Perceptron (MLP), Random Forest (RF) and Extreme Gradient Boosting (XGBoost) algorithms. Finally, the model is optimized and verified based on the measured data. The results show that the MLP-SVM model has a better prediction effect, and the accuracy and F1-score of the model in predicting truck lane change collision are 74.5% and 85.3%, respectively. Moreover, the safety of lane change is greatly affected by the type of vehicle in front of the original lane and the target lane, the state of motion and the distance between the front of the vehicle. The research results are helpful to provide theoretical support for the traffic management department to formulate dynamic early warning plans and optimize traffic control plans.