Cargo volume prediction of logistics sorting center based on GCN-BiLSTM

Abstract

Taking into account the transport network and average cargo volume of each sorting center, a directed weighted graph is constructed in this paper. Next, the GCN model is used to extract the spatial characteristics of the transport connection information of the sorting center and input it into the BiLSTM network. The BiLSTM network uses the two-way information flow to learn the temporal characteristics, and then uses the GCN-BILSTM model combined with the spatio-temporal characteristics of the sorting center to predict the daily cargo volume in the next 30 days. The integrated learning model based on ARIMA and BiLSTM is then used to predict the next 30 days of hourly cargo volume, and adjust and optimize. The results show that GCN-BiLSTM model and BiLSTM model improve the prediction performance.