Method for performing passenger flow OD calculation by combining bus and rail transit card swiping data

Abstract

The invention discloses a method for performing passenger flow OD calculation by combining bus and rail transit card swiping data. The method comprises the following steps: step 1, obtaining a getting-on station; step 2, reckoning a get-off station; step 3, inferring the get-off time of the passenger; step 4, generating a passenger flow OD matrix; and step 5, performing station passenger flow statistics. According to the method, the bus and the rail transit data are combined for passenger flow OD inference, the transfer behavior between the bus and the rail transit is considered, and the passenger flow OD inference accuracy is improved. Passenger flow OD inference is carried out by combining a travel chain method and two station attraction weight methods, so that the uncertainty of predicting passenger getting-off stations only according to the station attraction probability is effectively improved. The model is deployed on a Spark platform to operate, and the passenger flow OD of allbuses and rail transit lines in the Chongqing is deduced for only 3 hours, so that the calculation efficiency is greatly improved.

Description

technical field [0001] The invention relates to the field of public transportation travel planning, in particular to a method for calculating passenger flow OD by combining bus and rail transit card swiping data. Background technique [0002] Passenger's public transport travel data plays an important role in traffic planning, and the inference and analysis of passenger travel OD (origin and destination) is beneficial to the adjustment and optimization of public transport line sites. [0003] At present, an important method to calculate the OD matrix is ​​the calculation method based on the equilibrium theory. The OD estimation of bus passenger flow based on the equilibrium theory can be summarized as a linear or nonlinear optimization problem. It is based on the theory of mathematical statistics and is constrained by the number of passengers getting on and off the bus. At present, the main equilibrium methods include maximum likelihood model, generalized least squares model...

AI Technical Summary

Problems solved by technology

[0007] 2) The above model generally calculates the probability of passengers getting on and off at the bus station, and then obtains the OD passenger flow at the station, but this method can only estimate the OD passenger flow, and cannot obtain real-time and accurate passenger flow; on the other hand, The passenger flow data of the station is greatly affected by the environment, weather, and time (morning peak and evening peak), so the accuracy of the obtained probability of getting on and off needs to be verified;

[0008] 3) With the development of public transportation, rail transit occupies an increasingly important position in the daily travel of passengers, and passenger transfers between public transportation and rail transit are becoming more and more common. The current model generally applies to public transportation or The OD passenger flow of rail transit is calculated separately, ignoring the transfer behavior between bus and rail transit

[0009] 4) Passenger alighting time is an important attribute in passenger OD, but passengers only swipe their card when they get on the bus, and do not swipe their card when they get off, so it is necessary to infer the passenger’s alighting time, but the existing model only Infer the passenger's alighting site, and use the passenger's boarding time to approximate the passenger's alighting time when estimating the passenger flow, resulting in a low calculation accuracy

[0010] 5) The inference efficiency of the alighting station is low. For example, in Chongqing City, there are 600 bus and rail transit lines, and there are 6-7 million passenger card data per day. The above model is difficult to apply to all lines for OD calculation

Images