Urban people flow prediction method based on space-time dynamic neural network

Abstract

The invention discloses an urban people flow prediction method based on a space-time dynamic neural network, and the method comprises the steps: obtaining the historical movement track data of urban people flow, abstracting the urban people flow data into an image frame, dividing the urban people flow data into a training data set and a test data set according to the time, abstracting the urban people flow data into an image frame, and obtaining the urban people flow prediction result; converting a processing method into an image processing method, inputting an image frame into a three-dimensional convolutional neural network, extracting time characteristics and space characteristics, and capturing mobility characteristics of urban pedestrian flow; inputting the spatial features into a residual convolution block, and capturing the mutual influence of people streams in a region with a relatively long distance in space; and obtaining an urban area people flow prediction result through the training model. According to the method, the spatial-temporal dynamic graph and the residual convolution block are constructed, the urban area population flow characteristics and the spatial globalcorrelation characteristics are combined, the urban area people flow in a period of time in the future is predicted, and the method has the advantages of being high in convergence speed and accuracy.

Description

technical field [0001] The invention belongs to the technical field of flow forecasting, and in particular relates to a method for predicting urban flow of people based on a spatio-temporal dynamic neural network. Background technique [0002] The latest report of the United Nations shows that the "urbanization" process of the global rural population is accelerating. Data show that in 1950, the proportion of the global urban population was only 30%, but by 2018, the proportion of the population living in cities reached 55%. A large number of people are pouring into the city, and the traffic in the city becomes congested. To solve these problems, in modern intelligent transportation systems (ITS), people flow prediction is an integral part of providing accurate and reliable traffic information for travelers and transportation agencies. Knowing traffic information (such as traffic congestion, traffic volume, and people flow) in advance, cities can implement better traffic ma...

AI Technical Summary

Problems solved by technology

Zhao et al. used the cascaded LSTM model for short-term traffic flow prediction, but the model cannot learn regional dependencies, and the time to train the recurrent neural network is too long

Images