Research On COVID-19 Epidemic Prediction Based On Neural Network Multi-algorithm Combination Model

Since the outbreak of COVID-19 at the end of 2019,China has invested tremendous human and material resources to support the comprehensive fight against the epidemic,and the novel coronavirus(COVID-19),the cause of the epidemic,has rapidly spread under the global "pandemic" trend.The novel coronavirus disease 2019(COVID-19)virus is mainly transmitted through respiratory tract and contact.As of July 10,2022,the latest data from the World Health Organization(WHO)shows that there are still many new confirmed cases and asymptomatic infections in China,with a total of 4.32 million confirmed cases,550 million confirmed cases and 6.35 million deaths worldwide.The recurrence of the epidemic had a serious impact on the normal development between regions and people’s lives.Therefore,the establishment of statistical models to effectively predict the trend of the epidemic has important practical significance for people to adhere to epidemic prevention.This topic first to the domestic epidemic data and descriptive statistics analysis,according to the characteristics of the outbreak of domestic actual condition and data,selection of epidemic data in Europe,as the experimental data,control data at the same time.Secondly,constructs a model of the combined algorithm based on cyclic neural network as the experimental model,and set up the comparison model,using the experimental data and control data to do empirical analysis,The purpose is to verify the superiority of experimental data prediction.Finally,the experimental data are decomposed into multiple stable components based on EMD,and the actual simulation of each component by different models is studied.The results show that the prediction effect of the experimental model is the best for both experimental data and control data.the prediction accuracy is the highest only for the low-frequency data of each component of the experimental data.