Research On Short-term Traffic Flow Data Completion And Prediction Based On Machine Learning

In order to solve the problem of traffic congestion and improve the efficiency of road traffic,intelligent transportation system is widely used.Short-term traffic flow completion and short-term traffic flow prediction are important contents of intelligent transportation system.Short-time traffic flow completion can effectively solve the problem of incomplete traffic data caused by detector fault.In addition,accurate short-term traffic flow prediction can provide drivers with unimpeded road information and assist road control personnel to alleviate road congestion.Therefore,accurate completion and prediction of short-term traffic flow play an important role in guiding and controlling traffic flow.This paper mainly focuses on the completion and prediction of traffic flow,and the contents are as follows:(1)Research on short-time traffic flow completion model.The existing models mainly use the correlation between the data to repair,does not consider the impact of abnormal completion results on the correlation distance.And they do not fully explore the information contained in the missing position.This paper proposes a short-time traffic flow completion model based on Jaccard-Pearson correlation distance.The model calculates the Jaccard correlation coefficient between samples to reduce the influence of outliers,and obtains additional information from missing positions while using the first completion results.Combining the Jacquard correlation coefficient with the Pearson correlation coefficient,the Jaccard-Pearson correlation distance is constructed when the missing rate is considered.It can measure the correlation between samples more accurately and improve the accuracy of completion.The results of simulation experiments on a British highway show that the short-time traffic flow completion model proposed in this paper has good completion performance.(2)Research on short-term traffic flow prediction model.Existing short-term traffic flow prediction models are often combined with variational mode decomposition,but the modal component number and penalty factor in the variational mode decomposition are set by experience,so the optimal decomposition effect is not always achieved,and the prediction model does not fully consider the implicit information in the error.In view of these shortcomings,a new short-term traffic flow prediction model is proposed in this paper.The sparrow search algorithm is used to obtain the optimal modal component number and penalty factor of the variational mode decomposition.By adding the error correction module to extract the hidden information in the error,the prediction accuracy can be further improved.The comparison experiment with the existing short-time traffic flow prediction model on a British highway data set shows that this model has a better prediction performance.