Passenger flow collection and distribution mode discovery and prediction method

Abstract

The invention discloses a passenger flow collection and distribution mode discovery and prediction method, and the method comprises the following steps: (1) collecting the information of the number of persons participating in a competition activity, grading and marking the competition activity, collecting passenger flow data, and carrying out trip chain extraction on the passenger flow data; (2) on the basis of a trip chain extraction result and competition activity grading and marking results, calculating a polygonal region net passenger flow volume to obtain a hot spot region and a site; (3) extracting passenger flow collecting and distributing characteristics of the hot bus stations; (4) carrying out clustering training on the constructed passenger flow collection and distribution characteristics based on a Gaussian mixture model to obtain a clustering result, and analyzing the clustering result; (5) passenger flow prediction features are extracted from the hot bus stations; (6) carrying out one-hot coding and standardization on discrete features in the passenger flow prediction features; (7) training and storing the prediction model based on an LSTM method; (8) inputting the prediction set into the stored model to obtain a hot bus station passenger flow prediction result during the competition activity, thereby realizing hot bus station passenger flow prediction during the competition activity.

Description

technical field [0001] The invention belongs to the fields of traffic big data and intelligent transportation, and relates to a method for discovering and predicting passenger flow collection and distribution patterns. Background technique [0002] During the event, the passenger flow is concentrated in the area, mainly at the main bus stations around the event venue. In terms of time, the passenger flow changes significantly before the event starts (passenger flow gathers) and after the event (passenger flow dissipates). Therefore, from the perspective of space and time Above all, grasping the pattern of passenger flow distribution and making accurate predictions on hot bus stops can provide reference for bus dispatching and traffic planning. [0003] Rakha et al. used statistical theoretical correlation analysis methods to study the spatial and temporal extent of traffic conditions in typical non-accident situations, and by analyzing the traffic index data between differen...

AI Technical Summary

Problems solved by technology

[0006] According to relevant research, there are differences in passenger flow change models in different scenarios. At the same time, bus passenger flow is related to weather and date. At present, there are few studies on passenger flow collection and distribution modes during event activities, and passenger flow prediction based on passenger flow data only considers passenger flow trajectory. Mining of Passenger Flow Distribution Mode

Images