Risk warning for special population gathering in border areas
Author:Chunxue Wu 1 , Hui Zhou, Yan Wu, Naixue Xiong Author Unit: School of Optical-Electrical and Computer Engineering, University of Shanghai for Science and Technology, 200093, China The School of Public and Environmental Affairs, Indiana University Bloomington, 47405, USA College of Intelligence and Computing, Tianjin University, 300350, China
Abstract：Due to the strong randomness of special populations, it is affected by various factors, so the accuracy
of population aggregation prediction is not high. Aiming at the above phenomenon, a short-term crowd
congestion prediction method based on LSTM (Long Short-Term Memory) deep neural network is proposed.
The historical data is used as input to train the network and establish a short-term crowd congestion
prediction model to realize a short-term crowd in advance. It is called crowded prediction. The experimental
results show that using LSTM deep neural network to predict short-term crowd congestion is basically
consistent with the actual short-term crowding, and the prediction effect is ideal. Comparing the proposed
method with the prediction results of other prediction methods, the prediction average absolute error is the
smallest, which indicates that the proposed method has higher prediction accuracy, and verifies the feasibility
and effectiveness of the above prediction method in short-term crowd congestion prediction.
Keywords：Long short-term memory; Recurrent neural network; Special population prediction; Deep neural network; Early warning
Download information [SIZE：2.11 MB Amount of downloads： second]
