Research on Multi-scale Traffic Flow Prediction of Urban Road Network Based on Multi-dimensional Data Mining
The urban traffic flow is usually large,data is complicated and there are many characteristic parameters in road net-work traffic prediction,resulting in low prediction speed and accuracy.Therefore,a multi-scale traffic flow prediction method of urban road network based on multi-dimensional data mining is proposed.The data are preprocessed by data cleaning method,and the characteristic parameters of traffic flow are determined according to the changes of road traffic.Fourier transform and convolution algorithm are used to excavate the spatial-temporal characteristics of traffic flow,attention function and weight ma-trix are used to excavate the temporal characteristics of traffic flow.The traffic flow prediction model is established by the spa-tial-temporal correlation and the topological structure characteristics of the graph to obtain the road traffic change in the next moment and complete the traffic flow prediction.The experimental results show that this method can effectively predict the traffic flow state of each period of road traffic,and the prediction accuracy is always higher than 97％,and the traffic flow pre-diction of 8 road sections can be completed within 15 s.
multi-dimensional data miningurban road networktraffic flowspatial-temporal characteristics
万方数据
维普
