Abstract: In order to construct a software system for dynamic correction of thermospheric model with density error principle-level correction, firstly, based on the vertical temperature profile calculation method, the calculation function of inflection point and exosphere boundary temperature was established according to the spherical harmonic coefficient of latitude and longitude. A method of estimating the correction coefficient based on the model and the measured density was studied. From the aspect of numerical calculation, the correction coefficient was obtained by iteratively solving Taylor expansion linearization of the conditional equation. Then, the dynamic correction model system was constructed by combining the correction theory and numerical method. Afterwards, the system performance was evaluated according to the classification of magnetic storms and magnetostatic events. The results show that the forecast accuracy of the correction model was significantly improved compared with that of the MSIS00 model. The relative errors of the forecast decreased by 33.8% during magnetic storm and 9.1% during magnetic quiet respectively. Finally, the sequences of boundary temperature correction were analyzed and the main variation rules were obtained. Through the research and practice of the dynamic correction system of the thermospheric model, it is expected to provide a solution for the high precision forecast of themospheric density.