Research On Expressway Traffic Congestion Discrimination Based On Traffic Flow Parameter Prediction

As an important transportation channel and transportation hub,expressway plays a key role in the convenient travel of the people,the safe transportation of goods and the stable development of the economy.In recent years,although the construction volume of expressways has been gradually increased according to the upper policy,the growth rate of car ownership is still much higher than that of expressways mileage,resulting in the normal traffic congestion problem of expressways.At present,the construction of intelligent transportation system is regarded as the best way to alleviate traffic congestion problems,and accurate parameter prediction and congestion discrimination are important prerequisites for it to play the functions of traffic control and vehicle guidance,so this paper decided to carry out the research on expressway traffic congestion discrimination based on traffic flow parameter prediction.Firstly,this paper expounds the research background and significance of highway traffic parameter prediction and congestion discrimination,and analyzes the research status of traffic parameter prediction and congestion discrimination technology at home and abroad,conclude that the existing prediction technology has problems such as insufficient spatiotemporal feature extraction and large training calculation amount,while the existing discrimination technology has the problem of low computational efficiency and lacks technical research on the lane level.Based on this,this paper formulated the overall research content and technical route.Secondly,this paper introduces the working principle,advantages and disadvantages of various traffic acquisition technologies on expressways,and completes the collection work of traffic flow data on American I-5 expressways.Then,it carries out missing check,redundant check,threshold check and mechanism check in order to mark outliers in the original data.Thirdly,this paper introduces the working principles and advantages and disadvantages of various outlier repair techniques,and chooses linear regression,ridge regression and lasso regression as the basic model for repair,and then uses the combined form to build the repair data model based on regression.Verified by examples,the repair accuracy of the model for traffic flow outliers is significantly higher than that of the mean method.Then,use the correlation index to carry out spatiotemporal correlation analysis on the repaired data.Fourthly,this paper analyzes the current typical deep learning theory,and proposes a traffic parameter prediction model based on deep learning to solve the existing prediction problems.The model takes 3D spatiotemporal data as input,uses convolutional neural network to mine the spatial features of data,and uses bidirectional long short-term memory neural network and bidirectional gated recurrent unit neural network to mine the temporal features of data,while adding a soft attention mechanism to optimize the calculation process,and then predict the short-term traffic flow parameters of each lane section.Verified by the example,the prediction accuracy of this model is significantly higher than that of the traditional traffic prediction model.Finally,this paper analyzes the current typical clustering algorithm theory,and decides to choose particle swarm optimization algorithm to improve the iterative process of K-means clustering algorithm,so as to construct a traffic congestion discrimination model based on the improved clustering algorithm.The model takes the predicted traffic parameters as input,and then predicts the traffic congestion status of each lane and each section.Verified by an example,the discriminative effect of the model is obviously better than that of the traditional clustering algorithm,and the obtained discriminant conditions are basically consistent with the discriminant conditions in the Road Traffic Congestion Evaluation Method.There are 59 figures,28 tables and 75 references in this paper.