Traffic Congestion Improvement Based on K-Means Clustering Algorithm and Yolov3 Modeling

Based on K-means clustering algorithm and Yolov3 model, this paper proposes a traffic improvement scheme combining congestion warning and emergency lane optimization. By collecting the key indicators of each monitoring point on the highway, this paper constructs various models to realize real-time warning of traffic congestion and effective management of emergency lanes. Firstly, this paper utilizes the Yolov3 model for target detection of video data captured by four cameras to extract traffic parameters such as traffic flow. Secondly, the study also proposes a congestion warning model based on logistic regression and time series analysis to further improve the efficiency of traffic management on highways. In addition, data degradation and feature extraction are carried out by self-encoder, combined with K-means clustering algorithm to classify the traffic state, and then realize the automatic identification of traffic congestion signals. Based on the clustering results, this paper provides clear guidance for the opening of emergency lanes and shows strong differentiation ability.