Flexible air-breathing hypersonic vehicle (FAHV) inevitably suffers unknown aerodynamic disturbances as well as measurement noises in real flight. Although fixed-gain state estimator proposed in our previous work can effectively restrain the measurement noises, it may not perform well when large model disturbances and uncertainties exist. Focusing on this problem, we first study how the state estimator gains affect the transient process of FAHV. Then, the contradiction between accurate and fast state reconstruction versus high-frequency measurement noise suppression is analyzed. Based on the knowledge in choosing the state estimator gains, we design two fuzzy logic systems (FLSs), for the velocity channel and altitude channel respectively, to adapt the gains online. Finally, a novel knowledge-based fuzzy adaptive state estimator (FASE) is developed, which can realize accurate and fast state reconstruction in the transient process as well as high-frequency noise suppression in the steady-state process. Comparative simulation results verify the improvements of our proposed FASE on FAHV tracking performances.