Short-term Traffic Flow Analysis And Prediction Based On Deep Learning

In recent years,with the rapid development of the transportation industry,the urban traffic data is increasing day by day,and the traffic load is increasing.Traffic congestion has gradually become a major problem in traffic guidance.In order to solve the above problems,many scholars serve urban traffic efficiently by mining the potential information in traffic flow big data,so as to realize traffic scheduling and route optimization.Therefore,intelligent transportation system is bound to become an important research direction in the future transportation field,and the key technology in this direction is to realize efficient prediction of short-term traffic flow based on intelligent algorithm.Therefore,forecasting future short-term traffic flow based on traffic flow data will be a research hotspot in this direction.In this paper,taking the short-term traffic flow data as the research object,aiming at the spatio-temporal dependence of the data,the prediction models are constructed based on neural network to improve the accuracy of short-term traffic flow prediction,and a software system is designed and implemented based on the above algorithm.The main research contents are as follows:(1)Traffic flow characteristic analysis and data preprocessing: Firstly,the data set used is introduced,and the time and space dependence of short-time traffic flow time series is analyzed.Then,aiming at the missing and abnormal short-term traffic flow data,the average method and threshold method are used to preprocess the original data set to obtain high-quality data samples.(2)Short-term traffic flow prediction based on time dependence analysis: Aiming at the time dependence of short-term traffic flow data,a prediction model based on gated cycle unit is proposed.The gating cycle unit is used to extract the time dependence from the time series data of traffic flow to learn the changing trend of traffic flow data.Then it is compared with the classical traffic forecasting method.(3)Short-term traffic flow prediction based on spatio-temporal dependency analysis: Aiming at the spatial dependency in short-term traffic flow data under the constraint of spatial road network,in order to capture the spatio-temporal dependency at the same time,a prediction model based on graph convolution cyclic neural network is proposed.By combining graph convolution neural network and gating cycle unit,the spatial and temporal dependencies in the data are extracted respectively,and the traffic prediction is analyzed.(4)Short-term traffic flow prediction prototype system: A short-term traffic flow prediction prototype system is designed and implemented by Python3 language,Py Qt5 and other tools.The system realizes the functions of traffic bayonet search,current bayonet flow status display and current bayonet flow change trend.