Traffic Flow Prediction of Urban Intersection Based on Environmental Impact Factors and Markov Chains

The widespread promotion and application of big data and artificial intelligence technology has brought new challenges and strong supports to traffic management. The traffic state estimation and traffic flow prediction of urban road network are important issues for intelligent transportation systems. This paper proposes a data-driven model to classify environmental impact factors through cluster analysis of large number of historical data, and then found that environmental impact factors correspond to traffic states of relevant intersections in different traffic conditions. On the basis of obtaining the corresponding traffic state, high-order multivariate Markov chain is used to predict traffic flow under different traffic states, and the accuracy of the traffic flow prediction is effectively improved. In addition, this paper also analyzes the Markov property of traffic flow time series after clustering, and use the index of Markov property to predict prediction performance. The analysis of several practical cases proves the effectiveness of the proposed method.