Research on the traffic flow measurement method of intelligent connected vehicles based on series and cascade technology

Abstract

At present, the major cities in our country have entered the development of smart cities, the utilization rate and prevalence rate of intelligent connected cars in smart cities have reached a new peak,to manage the real-time traffic flow of intelligent connected vehicles effectively in smart cities. This paper proposes to calculate the flow of intelligent connected vehicles based on series cascade technology. Firstly, this paper collects multi-source heterogeneous data based on multiple collectors; Secondly, a generative adversarial network is employed for data fusion, synthetic data, closely mirroring the original, is generated for handling unknown variables during model training. Next, an autoencoder model generates input data for feature representation. Lastly, CatBoost, a regression model that leverages a cascaded machine learning algorithm, is utilized to forecast the traffic flow of intelligent connected vehicles in smart cities. The results show that the method based on CASCaded machine learning has remarkable effect in predicting smart city traffic flow and alleviating urban road congestion.