In the process of engine starting, the airflow of every section is in subcritical condition. It's hard to construct a model by using the traditional thermodynamics method. A dynamic identification model is set up in this paper based on Radial Basis Function network, using the flight data records of aeroengine parameter in the process of engine starting as learning samples. The simulation results show that the model has a good dynamic performance and high accuracy, which opens a new way to build the model of engine in low speed.