Abstract: Simulation of COVID-19 epidemic situation has certain guiding significance for judgment of epidemic situation in China. The daily data released by the National Health Commission from January 20 to March 29, 2020 have been used to build ECM model and AR model to simulate the epidemic data trend. In ECM model, R² is 0.9449 and t-test model associated probability is less than 0.05, showing significant difference. The order numbers of two AR models are 6 and 7 respectively. The R² of two AR models are 0.9998 and 0.9978 respectively. F-test and t-test models have significant associated probability of less than 0.05, and the residual is white noise. ECM model reveals that the number of newly confirmed cases on a certain day can be estimated quantitatively according to the number of newly added suspected cases and newly confirmed cases of the previous day. Over time, the number of new cases and suspected new cases has decreased simultaneously. AR model estimates show that the cumulative number of confirmed cases will slow down and stabilize after March 1, 2020. By the 60th day, or March 19, 2020, the cure rate is 87.8278% and the corresponding sample value is 87.8753% with a relative error of 0.0514%. By the 60th day, the national cumulative simulation value of confirmed cases is 81796.0989 with a relative error of 1.0240%. The ECM model and the two AR models established in this study based on the time series data of Covid-19 epidemic in China have high simulation degree, low error and high accuracy of the calculation results. The simulation results can be used for the estimation of other relevant epidemic indicators without causing error amplification.

Key words: COVID-19, time series data, ECM model, AR model, simulation