| In recent years,with the continuous improvement of urban construction,urban rail transit has become increasingly mature.People are getting better and better about the quality of travel.However,rail transit will also bring traffic congestion problems while people are traveling.Through the in-depth analysis of historical passenger flow data of each station in the rail transit network,the characteristics of rail transit passenger flow are excavated,and the congestion status of the station is predicted in a short time interval,thereby providing people with personalized travel guides and improving the service of the subway department.Quality.Rail transit passenger flow is very different from ordinary passenger flow.Rail transit stations are closely linked,forming a huge rail transit network.The passenger flow is formed by the continuous inbound and outbound of travelers in the whole network.Rail transit passenger flow not only has many characteristics such as periodicity and randomness,but also makes the passenger flow between stations have time-space relationship.On the basis of studying the characteristics of passenger flow in rail transit,this paper takes into account the spatial and temporal relationship of passenger flow at stations in the network,and predicts passenger flow congestion at stations,and achieves good results.Rail transit congestion prediction has become a hot spot in current researchIn this paper,CNN convolution neural network algorithm is used to predict the congestion of rail transit.By analyzing the characteristics of rail transit passenger flow and changing the structure of CNN convolution neural network,the influence on prediction performance is studied.The main research work of this paper is as follows:(1)In order to ensure the integrity of the data,the data preprocessing of the passenger flow data is carried out for the characteristics of the passenger flow of the rail transit,including the addition of missing data and the conversion of the congestion level of the passenger flow.(2)Constructing the structure and training model of CNN convolution neural network.On the basis of learning convolutional neural network theory,an algorithm model suitable for rail transit designed and trained.(3)Using the trained model to explore the predictive performance of the model through the real data of Shanghai rail transit passenger flow.The experiment is mainly divided into two parts: first,the data pattern is divided into the passenger flow data set of the rail transit;secondly,the influence of the network structure on the prediction performance is discussed for the size of the convolution kernel and the depth of the network.The traffic congestion prediction algorithm based on CNN presented in this paper has good performance for real data congestion prediction.When the time interval is less than or equal to 15 minutes,compared with the historical average method,the algorithm in this paper has higher accuracy.The research in this paper provides a new idea and direction for the further development of CNN convolution neural network algorithm and its practical application,and has practical significance. |
