Research and application of urban real-time traffic flow prediction based on STARIMA

Abstract

As one of the most basic and important functions in the intelligent transportation system, the related research of intelligent transportation management and control has attracted extensive attention of researchers from all over the world. Good traffic control and guidance are inseparable from accurate and real-time traffic flow prediction, and at the same time, real-time and reliable traffic flow prediction is also the key to the transition of the traffic system from “passive control” to “active control The extraction and analysis of traffic flow features are inseparable from the support of complete and accurate data sets. This paper considers the correlation between missing data and available spatiotemporal data, as well as the correlation between road network topology and traffic flow, in order to minimize the filling of missing data points In this paper, an optimization method for missing data filling based on spatiotemporal topological map is proposed. Firstly, use the defined “relevant road sections” and “traffic flow correlation levels” to analyze the temporal and spatial correlation of the traffic flow of the road network, and generate a traffic flow prediction map. Traffic flow prediction is performed using an improved space-time auto-regressive moving average model (Space-time Auto-regressive Integrated Moving Average STARIMA). The prediction results show that the algorithm proposed in this paper reduces the prediction time by reducing the number of filling points. At the same time, because the algorithm selects the most relevant traffic section for traffic flow prediction, the prediction accuracy is further improved, and a real-time and effective traffic flow is finally realized. predict.