Forecasting system and method for resident traffic flow

Abstract

The invention discloses a resident traffic flow forecasting system and a forecasting method thereof, which specifically adopt a full convolutional network for forecasting. Geographic grid division is used to divide the urban residents' travel area into multiple grid areas, and the traffic flow data of residents' travel is converted into a flow map, which is used as the input of the full convolutional neural network, and the data is increased by different channels of the expanded image. dimension, the features of the traffic flow map are extracted through the convolutional layer, and then multiple deconvolutions are obtained to obtain an image of the same size as the input traffic map, and the entire network is reversely adjusted by the error between the actual traffic map and the predicted traffic map of the next time period . The invention eliminates the operation of the full connection of the traditional convolutional neural network, and adopts deconvolution to predict the traffic flow map of the next time period, which can not only improve the prediction accuracy, but also greatly reduce model parameters and calculation consumption.

Description

technical field [0001] The invention belongs to the technical field of traffic flow forecasting, in particular to a resident traffic travel flow forecasting system and a forecasting method thereof. Background technique [0002] With the increase of urban residents, the scale of various commercialization gradually expands, which brings convenience to people's life, but also causes a certain degree of congestion, and even safety accidents caused by excessive crowd flow, such as stampede ACCIDENT. The prediction of residents' travel flow not only plays a positive role in alleviating traffic flow, but more importantly, it is of great significance to urban management and public safety, and is one of the important factors to prevent safety accidents. If we can predict the flow of people in an area, we can alleviate or prevent these tragedies by using emergency mechanisms, such as traffic control, sending warnings, and evacuating people, so as to bring a smooth and safe traffic en...

AI Technical Summary

Problems solved by technology

[0011] The purpose of the present invention is to provide a resident traffic travel flow forecasting system and its forecasting method, which solves the problem that in the prior art, the characteristics of traffic flow flow in interconnected areas and different areas are not well extracted, and the actual traffic conditions are not accurate. The controllability is too large, and the problem of the defects of the prediction model itself

Images