The aim of this research is to obtain in advance the short-term temperature change trend of concrete in the first stage of temperature control, and take corresponding temperature control measures in time to prevent from temperature cracks. With high arch dam concrete in construction period as the research object, we examined the comprehensive influences of initial temperature, water cooling, adiabatic temperature rise, environmental temperature and layer heat dissipation on the concrete temperature of pouring warehouse, and then established a dynamic prediction model of temperature change in the first stage temperature control stage of high arch dam concrete by timely updating the initial temperature. Furthermore, in view of the difference of concrete in different pouring warehouses, we adopted the nonlinear optimization method to optimize the important parameters of the model, and verified the accuracy of the model using such indicators as maximum absolute error (MAE), average absolute error (AAE) and relative error (RE). With engineering practice as case study, we updated the initial temperature with 2 days as the step and optimized the model parameters, and predicted the concrete temperature of the pouring warehouse with 12 days as the age of concrete. The maximum absolute error (MAE) between predicted value and measured value is within 0.6 ℃, the average absolute error (AAE) within 0.2 ℃, and the relative error (RE) within 0.9%. The prediction accuracy of the model meets the requirements of the construction site.