Monitoring and optimizing the positioning of the UAV in the airport can reduce the security risk of airport operation. Due to the low, slow and small features of UAV flight, the surveillance signals of airport scene are easily obscured by ground obstructions. The traditional location method mainly uses the reference feature point location, and for the non-moving object with different reference features, a large number of positioning errors will be generated during the iteration process. According to the features of intruding UAVs, a TDOA positioning algorithm based on Taylor series for adaptive bad pixels removal is proposed. TDOA measurements were used to locate multiple points and an adaptive filter was constructed to remove the bad pixels from the initial reference points of the Taylor series expansion, which effectively improved the convergence and efficiency of the algorithm. The simulation results show that the accuracy of adaptive bad pixel reject algorithm is close to the Taylor series expansion algorithm based on the actual position, and the convergence speed is faster. When the position of the ground station relative to the UAV is poor, the stability and positioning accuracy of the new algorithm are more Excellent, to meet the airport on the invasion UAV surveillance positioning accuracy requirements.