The region passive locating system, named Beidou, nowadays is under extensive research, the inaccuracy of receiver, environmental noise, man-made interference and other factors will cause false data in the relevant measuring arrays. The difference between these data and the normal value in terms of quantity is larger and exceeds the allowed range of error, which will obviously affect the Kalman filtering process and cause the decrease of the position precision. All these will lead to instability of the system and even the disperse of filtering. This paper presents an improved algorithm, the principle of which is to reject the dynamic fault tolerant by using Kalman filter. This method upgrades the tracing ability of the filter under the airborne environment. The real-time character is improved because of the decrease in the dimension numbers of the state equation. Verified from the trying out data, the locating precision of the system can reach up to 92 meters, filtered by the algorithm, thus it can meet the general needs of the users in the air.