Analysis Of Urban Road Traffic Flow Characteristics And Short-term Traffic Flow Prediction Model

With the acceleration of urbanization,the traffic problems that accompany it have become more serious.Intelligent transportation system plays an important role in mitigating traffic problems in the construction of smart cities.Mastering the traffic flow characteristics and accurately predicting them is the key to realize the efficient induction and management control of traffic flow by intelligent transportation systems.Therefore,it is always a hotspot in the field of transportation to construct a short-term traffic flow prediction model based on traffic flow characteristics and improve prediction accuracy.This paper summarizes the research status of traffic flow prediction models at home and abroad,and studies the application of Gaussian process regression in wind speed prediction and charge prediction.The optimization of hyperparameters in the traditional Gaussian process regression model is proposed.The disadvantages of strong value dependence and easy to fall into local optimum,in order to adopt the particle swarm optimization algorithm with strong global optimization ability,comprehensively consider the time series chaotic characteristics and spatio-temporal correlation characteristics of traffic flow data to construct an improved Gaussian process regression prediction model,and obtain high accurate future traffic flow information.The result can provide travellers with reliable travel information,improve road traffic efficiency and people’s travel quality,and provide theoretical support for mitigating traffic problems.Firstly,the characteristics of urban road traffic flow are analyzed,and the chaotic characteristics of traffic flow time series are analyzed by using the quantitative judgment index of chaotic characteristics.The temporal correlation,weekly similarity and sequence autocorrelation of traffic flow are analyzed based on correlation theory from time dimension.From the spatial dimension,the horizontal correlation between the one-way lanes and the longitudinal correlation between the adjacent upstream and downstream sections of the study section are analyzed,which lays a foundation for establishing a short-term traffic flow prediction model based on the characteristics of traffic flow.Then,according to the chaotic characteristics of the traffic flow time series,the C-C method is used to determine the optimal delay time and minimum embedding dimension of the original traffic flow time series,and phase space reconstruction is performed.The particle swarm optimization algorithm is used to optimize the hyperparameters of the Gaussian process regression model,and the phase space reconstruction data is used as the training set and test set of the prediction model.The short-term traffic flow prediction model based on phase space reconstruction and particle swarm optimization Gaussian process regression is constructed.The measured data of the Beijing East Fourth Ring Road Detector was used to compare the model prediction effects with multiple indicators.Finally,according to the analysis of the temporal and spatial correlation of traffic flow,explore the variation law and evolution process of space-time dimension,consider the similarity of traffic flow sequence in time dimension,and study the correlation between section and adjacent upstream and downstream sections in spatial dimension.A model input data set integrating different spatio-temporal information is established,and an improved Gaussian process regression prediction model based on spatio-temporal data fusion is constructed.Based on the data collected by Beijing Four-Circle Road Layout Detector,the comparison of the prediction results of different time-space dimension input sets through the model is carried out to verify the validity and accuracy of the model.It shows that the model constructed by considering the spatio-temporal correlation characteristics of traffic flow can further improve the short-term traffic flow prediction accuracy.