Abstract: Based on the early behavior of COVID-19 until February 1st, the SIR model is modified to solve the dynamic equation of virus evolution by using the number of susceptible persons, the probability of infection and the latent infection rate. The change trend of infected persons is studied, and the influence of government administrative actions on the trend is analyzed. The results showed that after January 24th, 2020, the administrative action of the government has effectively slowed down the spread virus and the number of infected people decreased obviously. The number of current infections had reduced more than a half compared with the previous trend forecast according the trend of January 24th, 2020. The number of susceptible persons, the probability of infection and the latent infection rate were all greatly reduced. Based on the current trend and the optimal parameters, it is predicted that the outbreak will peak around February 9th, 2020 and then decline.