Abstract: In allusion to the high precision requirements for flight parameters, a subsonic flush air data sensing(FADS) system was developed and flight-tested. First of all, the FADS pressure data was obtained based on computational fluid dynamics(CFD) modeling, and error distribution bounds for pressure inputs was analyzed based on CFD modeling accuracy and wind tunnel calibration data. Secondly, a real-time solution algorithm based on a surrogated model was analyzed, and a principle prototype was implemented. Finally, flight tests were conducted to validate the FADS principle prototype. In order to evaluate the reliability of the real-time FADS angle of attack solving algorithm in flight test validation, the angle of attack solving algorithm based on the aerodynamic model was rebuilt afterwards. Systematic comparisons of the angle of attack between FADS system and INS were made. The results show that similar results are obtained from the aerodynamic model and the surrogated model, and the reliability of the angle of attack real-time solving algorithm is validated during flight tests. The accuracy of the real-time angle of attack solving algorithm is less than 1°, and even less than 0.5° during the key flight phase. Abnormal fluctuations in pressure inputs exceeding error distribution bounds are the main factors causing the angle of attack data abnormalities. The FADS solving algorithm determined by CFD modeling technology can set allowable pressure input error bounds, and reasonable fluctuations in pressure inputs have no influences on the angle of attack. Therefore, angle of attack data abnormalities are caused by abnormal fluctuations in pressure inputs exceeding error distribution bounds. For the FADS system applied to a subsonic vehicle, pressure transducer accuracy and fine engineering implementation are necessary for high precision solution.