Modeling And Forecasting Of Traffic Flow In Guiyang Based On Multi-factor Analysis

As the data center of Southwest China,Guiyang City has lagged behind in urban traffic supply and serious traffic congestion,which hinders economic development and at the same time threatens the safety of citizens with various traffic problems.In order to alleviate traffic congestion,traffic flow prediction,as an important part of intelligent transportation system,has become a research hotspot at home and abroad.Taking a typical road section in Guiyang as the experimental object,this dissertation analyzes the characteristics of traffic flow data,and discusses three external factors affecting traffic flow: A hybrid neural network model is built to predict short-time traffic flow under the factors of road facilities,weather and holidays,and the model is used to forecast short-time traffic flow at target intersections in local road network.The main contents include:Firstly,the comparative experimental sections are determined to complete the statistics and preprocessing of cross-section traffic flow data to obtain the short-time traffic flow data set,and analyze the influence of road facilities factors,weather factors and holiday factors on traffic flow.Secondly,based on the basic theory of neural network,the structure and characteristics of back propagation(BP)neural network optimized by modified adaptive moment estimation(RAdam)and Bi-directional long short term memory(Bi LSTM)neural network are discussed and combined,A Bi LSTM-BP optimized model is proposed for short-term traffic flow prediction under various factors.Finally,a local trunk road network in Guiyang was established,and Principal Component Analysis(PCA)was used to explore the correlation between traffic flow data at intersections.The traffic flow data at Principal Component intersections were selected as the input of the Bi LSTM-BP optimized model.Short-term traffic flow prediction is carried out on target intersections in the local road network under multiple factors.From the comparison diagram of the listed evaluation indexes and prediction results,it can be concluded that the Bi LSTM-BP optimized neural network model proposed in this dissertation has good prediction effect under all factors and has practical application significance.It can be used to forecast the traffic flow data of target intersections in the local road network.