Urban human traffic prediction method and system

Abstract

The present invention relates to the technical field of mobile computing, especially to an urban human traffic prediction method and system. The method comprises the steps of: performing regional division of a city according to an urban road network graph to obtain a plurality of preliminary divided regions; performing clustering of human traffic movement features of each preliminary divided region to obtain regions after clustering; taking human flow time features, human flow space features and human flow speed features as feature extraction standards to perform feature extraction of the regions after clustering; and performing fusion of the human flow time feature extraction data, the human flow space feature extraction data and the human flow speed feature extraction data, taking the fused data of the human flow time feature extraction data, the human flow space feature extraction data and the human flow speed feature extraction data as input date input into a graph convolutional neural network structure for training, and obtaining a training result, namely an urban human traffic prediction result. The present invention further discloses an urban human flow prediction system. The urban human traffic prediction method can effectively predict human flow and has high prediction precision.

Description

technical field [0001] The invention relates to the technical field of mobile computing, in particular to a method for predicting urban flow of people and a system for predicting urban flow of people. Background technique [0002] Urban crowd flow prediction is of great significance in urban planning, traffic management, public safety and other fields: the law of urban crowd movement can help city managers to carry out reasonable traffic control and energy supply, avoid environmental pollution and resource waste; it can help passenger transportation companies Plan bus routes and provide mini-buses in traffic hotspots to improve service quality and efficiency; urban crowd flow forecasting can detect large-scale crowd gathering activities in time to avoid stampede incidents. [0003] In the traditional method, people use the GPS data of taxis taken by users to analyze the flow of urban crowds. However, the number of users that can be covered by this method is very limited, and...

AI Technical Summary

Problems solved by technology

[0004] Nevertheless, accurate prediction of crowd flow data still faces great challenges: first, the movement of crowds has complex spatial and temporal correlations, and simple time series models and machine learning models are difficult to achieve high-precision predictions; People flow prediction needs to consider different regional division methods, such as division by administrative area, division by functional area, etc. People often pay attention to regional crowd flow information with semantic information, such as gymnasiums, squares, etc. These semantic areas are often irregular, This makes the traditional deep method represented by convolutional neural network (CNN) unable to be used directly

Images