Urban crowd flow prediction method based on long-term and short-term memory network model

Abstract

The invention discloses an urban crowd flow prediction method based on a long-term and short-term memory network model, and the method comprises the steps: selecting an input parameter and an output parameter of urban crowd flow prediction, and considering the impact from crowd flow data of adjacent dates; marking holidays and festivals as relatively important crowd flow prediction influence factors; labeling the date of each day as a crowd flow prediction influence factor and date labeling information; defining a model among the input parameters, the cascade information and the output parameters; carrying out training modeling on the population flow of the major cities in the first three years by using a long-term and short-term network model, calculating an error between a predicted value and a real value during training by using a root-mean-square error, and carrying out error feedback; establishing a long-term and short-term memory network prediction model through network training;and loading the trained model, and carrying out multi-step prospecting to obtain a prediction result. According to the method, for urban population flow prediction, the precision of a prediction result is improved, and good robustness is achieved.

Description

technical field [0001] The present invention relates to the technical field of time series forecasting, in particular to a method for realizing a longer period of time series forecasting—long-short-term memory network Background technique [0002] For popular tourist cities such as Yunnan Province, especially popular tourist attractions such as Dali and Kunming, the phenomenon of crowd "blowout" often occurs, and this phenomenon is more obvious during holidays. In order to prevent various potential safety hazards in urban traffic gathering places, scientific and effective prediction methods are urgently needed to predict the flow of people, detect the characteristics of crowd flow early, and provide decision-making basis for the coordination of regional management resources. Most of the existing crowd forecasting methods use the traditional time series forecasting algorithm or its improved algorithm to predict the crowd flow in the future. Poor predictions and more. [000...

AI Technical Summary

Problems solved by technology

Most of the existing crowd forecasting methods use the traditional time series forecasting algorithm or its improved algorithm to predict the crowd flow in the future. Poor predictions, etc.

Images