Abstract: In the case of emergencies and big data, the importance decision algorithm of emergency material demand based on data stream fuzzy C-means clustering algorithm is established to identify the importance of emergency material demand for the cluster supply chain subsystem. For the coupling cha- racteristics between the cluster supply chain subsystems and fast changing data stream characteristics of prediction indexes, the dynamic forecasting algorithm of emergency material demand based on the long short-term memory network is proposed for cluster supply chain to extract the time series characteristics of emergency material demand parameters from multiple subsystems of cluster supply chain. The system identification and estimation for the uncertainties of emergency material demand of the interconnected large-scale system are carried out dynamically. The simulation results prove the feasibility and accuracy of the dynamic prediction algorithm for emergency material demand based on the long short-term memory network in cluster supply chain.

Key words: demand forecasting, material demand importance, long short-term memory network algorithm, cluster supply chain；emergency material