Research On Short-term Traffic Flow Prediction Method Based On Microwave Data

In recent years,China’s social economy has developed rapidly,people’s traffic demand has gradually increased,and the imbalance between the supply and demand of road infrastructure has become increasingly serious,which has brought a heavy burden to the transportation road network,and the problems of inefficient transportation operation and deterioration of the transportation environment have come along.Under this background,the intelligent transportation system came into being.Intelligent transportation is of great significance in improving road conditions,predicting emergencies,and planning travel routes.The advent of the era of big data in the transportation field has led to a surge in traffic data information.If you can use massive data reasonably and predict traffic flow in real time and efficiently,and then give corresponding theoretical and methodological guidance,it will certainly alleviate to some extent Traffic jam problem.Based on the characteristics of traffic flow,this paper proposes deep learning and integrated learning methods to achieve short-term prediction of traffic flow.Firstly,the domestic and foreign traffic flow prediction methods are sorted out,and the research background of this paper is given.On this basis,the technical route of this paper is proposed.Secondly,it briefly introduces four kinds of traffic flow data collection methods commonly used at home and abroad,and proposes traffic flow preprocessing methods for microwave data used in this paper.On this basis,the spatial and temporal characteristics of urban expressway traffic flow are analyzed.Thirdly,in order to make full use of the spatiotemporal characteristics of traffic flow,this paper proposes a deep learning model based on Convolutional Neural Networks,Gated Recurrent Unit and Attention mechanism,namely ACGRU model.The model uses the Convolutional Neural Networks model to mine the spatial characteristics of traffic flow,and the Gated Recurrent Unit model obtains the temporal characteristics of traffic flow data.At the same time,in order to capture the features more effectively,an attention mechanism is introduced.The results show that the ACGRU model can better fit the traffic flow change trend,and the prediction accuracy of this model is better than that of a single Convolutional Neural Networks or Gated Recurrent Unit model.At the same time,the results of ACGRU and its variant models AGRU model and CGRU model are compared,and the effectiveness of each module is evaluated.Finally,in order to further improve the accuracy of traffic flow prediction and make up for the shortcomings of a single model,this paper proposes a short-term traffic flow prediction method based on Stacking algorithm.A two-layer stacking algorithm is used to put multiple base learner algorithms such as ACGRU into the first layer,and linear regression is used as the second layer to construct a multi-model fusion framework.Experiments show that the Stacking algorithm improves the performance of the model,and its fusion result is better than a single algorithm.There are 40 pictures,22 tables and 69 references.