Short-term Traffic Flow Prediction Based On Hybrid BP Recurrent Neural Network

In recent years,the increasing number of private cars has made the traffic congestion worse.In order to better solve this problem,it is of great significance to develop a software application related to intelligent transportation.Obviously,the implementation of this measure can not only provide convenience for people’s travel,but also lay the foundation for the systematic management of the transportation sector.At present,the functional design of related software mainly involves traffic guidance and control to complete the implementation of traffic flow.This method does not enable accurate prediction of short-term traffic flow.Based on this,this article has two different prediction models and prediction methods.Dimensional cut,in-depth analysis around traffic flow prediction,the main research content can be summarized into the following three aspects,as follows:(1)Select the BP recurrent neural network prediction model,and directly use it as the basic prediction model according to the analysis needs,and then start the analysis based on this.Because the traffic flow data sequence itself exhibits a high degree of non-linearity,the BP recurrent neural network can effectively identify and predict the sequence with this characteristic.So use this network to carry out the research and analysis of this article.(2)Optimize the traffic flow prediction method.When predicting short-term traffic flow,the error formed has certain rules.In order to collect error data,this article introduces the error compensation extreme value in the process of research and analysis,which can effectively improve the accuracy of prediction and analysis;There is still a certain degree of linear correlation in data with highly nonlinear characteristics.In order to ensure the accuracy of the final result,this paper uses BP recurrent neural network to predict the nonlinear part of the sequence,while the linear part uses Kalman Filter for prediction,and build a prediction model based on the hybrid BP recurrent neural network.(3)Use the optimized ant colony algorithm to improve the hybrid BP recurrent neural network model.When the neural network is used for prediction,it has high sensitivity in terms of threshold and weight changes,so the ant colony algorithm is used to further improve it.At the same time,in order to further strengthen the optimization performance of the ant colony algorithm,the heuristic equation and the pheromone concentration of the ant colony algorithm are also improved in a rational way to build a hybrid BP recurrent neural network based on the optimized ant colony algorithm Forecast model.In the course of the experiment,this article uses traffic flow data published by MinnesotaDuluth to combine BP cyclic neural network(BPRNN),hybrid BP cyclic neural network(KFBPRNN-EC),and optimized BP cyclic neural network(LIACO-BPRNN)and the hybrid BPRNN(LIACO-KF-BPRNN-EC)of optimized ACO were all integrated and analyzed,and short-term traffic flow prediction simulations were carried out respectively,and the experimental results were obtained.The research and analysis conclusions show that compared with other models,LIACO-KF-BPRNN-EC has been greatly improved in terms of fit and prediction accuracy.