Prediction method of od passenger flow of urban rail transit based on od attraction

Abstract

The invention provides a method for predicting OD passenger flow of urban rail transit based on OD attractiveness. The method includes: according to the historical data statistics of the attractiveness value between OD pairs in different periods in the rail transit network and the corresponding passenger flow between OD pairs, which are respectively expressed as an OD attractiveness matrix and a first OD matrix, wherein the attractiveness value reflects the OD pair The degree of attracting passenger flow between the rooms; based on the attractiveness value of the OD pair, it is divided into a plurality of attractiveness levels and selects a reference level; according to the OD attractiveness matrix, the attractiveness value extracted from the first OD matrix exceeds the specified The OD pairs of the above reference levels constitute a second OD matrix; the second OD matrix is ​​input to the deep learning model, and the predicted passenger flow between the OD pairs is obtained through training. The method of the invention can improve the accuracy, validity and real-time performance of passenger flow prediction.

Description

technical field [0001] The invention relates to the technical field of rail transit operation management, in particular to a method for predicting OD passenger flow of urban rail transit based on OD attractiveness. Background technique [0002] With its advantages of large capacity, high speed, high punctuality, low pollution, and high safety, urban rail transit has ushered in explosive development in China in recent years and has become an indispensable public transportation tool in large and medium-sized cities. As of the end of 2017, 165 urban rail transit operating lines have been opened in 34 cities in my country, with a total mileage of 5032.7 kilometers, of which 3883.6 kilometers are subways, accounting for 77.2%. In addition, a total of 254 lines in 56 cities are under construction, with a total mileage of 6,246.3 kilometers. The number of cities under construction, the number of lines under construction, and the number of kilometers under construction have all exce...

AI Technical Summary

Problems solved by technology

[0006] However, the OD short-term passenger flow prediction in the prior art mainly has the following problems: 1), there is an extremely complex spatio-temporal relationship between the OD passenger flow, and a single mathematical optimization method such as least squares method, state space model, etc. cannot be very good The description of complex spatio-temporal relationships, the "accuracy" needs to be improved

2) Most of the existing research is applied to small-scale subway networks, and with the expansion of the network, the computational complexity of the above model will be greatly increased, which cannot meet the "real-time" requirements

3) The number of OD pairs is far greater than the number of stations, which is n-1 times the number of stations (n ​​is the number of stations). However, the passenger flow between many OD pairs is very small or even non-existent. Therefore, existing research uses Treating all OD pairs equally will not only increase the complexity and calculation of the model, but also have a negative impact on the prediction results

4) For short-term passenger flow forecasting, the selection of time granularity is a basic task. Although the passenger flow sequence extracted at a smaller time granularity can describe the refined information of passenger flow, its regularity is poor, and the prediction accuracy is low. Although the passenger flow sequence extracted at a large time granularity will lose detailed passenger flow information, it has strong regularity and high prediction accuracy. How to grasp the selection of time granularity is also one of the existing problems

Images