Method and system for predicting rail traffic flow based on spatio-temporal information

Abstract

The invention relates to a method and system for predicting the flow of people on a track by integrating spatio-temporal information, which belongs to the field of data mining. First, preprocessing operations such as cleaning, integration, transformation, and reduction are performed on the original data of subway traffic flow, and the subway traffic data is transformed into a dual-channel flow matrix with time and space attributes; then, the temporal proximity, periodicity, and For traffic change sequences with changing trends, three residual unit branches are designed, and then convolutional neural networks are used to capture regional correlations for each branch, and a prediction model based on deep learning is built; finally, the established The model and real-time data set are used for feature extraction, and the prediction results are sent to the mobile terminal through real-time model prediction, so as to realize the real-time performance and lightweight of the subway passenger flow prediction system. The invention solves the problems of low prediction accuracy and poor real-time performance existing in the existing traditional subway passenger flow prediction system, thereby reducing the bearing pressure of urban traffic.

Description

technical field [0001] The invention belongs to the field of data mining, and relates to a method and a system for predicting the flow of people on a track by integrating spatio-temporal information. Background technique [0002] Traditional subway passenger flow forecasting systems often ignore the correlation between the change of subway passenger flow and time and space, or can only predict the flow change of a specific subway station. However, in the real world, the subway traffic flow will be affected by various spatio-temporal environmental factors. From a spatial point of view, the subway population is highly mobile, and the inflow and outflow of population in different regions influence each other, and the surrounding areas also have a huge impact on it; The influence of the time period, and the fixed time period will also be affected by social schedules and seasons. In addition, some external factors (weather, social events, etc.) may also greatly change the subwa...

AI Technical Summary

Problems solved by technology

Although the CNN model can capture the spatial dependence well, its prediction model does not consider the time dependence and external influence factors, and still faces the problem of low prediction accuracy; while in the RNN / LSTM cycle neural network modeling method, The RNN model can handle time series data well, and predict the trend and peak of human traffic in the future based on known data

However, the RNN / LSTM neural network prediction model can only deal with short-term time series and spatial attributes of adjacent areas, and it still faces problems such as low temporal and spatial dependence and low prediction accuracy.

Therefore, none of the existing methods can improve the accuracy of subway passenger flow prediction very well, and still face the problem of low accuracy and high delay of the constructed prediction model

Images