Traffic Operation Status Research Based on Multi-source Data Fusion

At present, modern technologies such as big data, cloud computing and vehicle-road collaboration are developing rapidly, and the existing traffic detection data based on a single detector cannot meet the demand for high-precision data in the vehicle-road cloud environment. Facing the massive, heterogeneous and complex traffic data, this paper selects the data collected by LiDAR, HD camera and V2X communication unit as the data sources for this research, establishing the multi-source traffic data fusion model with wavelet neural network and genetic algorithm optimized BP neural network respectively, and builds the traffic status classification model based on fuzzy C-mean clustering algorithm. The fusion model and the traffic status classification model are verified for the single intersection road section in the vehicle-road cloud cooperative environment. The results show that the fusion accuracy of genetic algorithm optimized BP neural network reaches 93% and the status classification model has high feasibility, which can provide data support for intersection signal control and traffic guidance optimization.