Aiming at the problem that unmanned aerial vehicle（ UAV） often encounter large maneuverable flight and magnetic interference

and the attitude angle error of UAV will fluctuate greatly

a quaternion extended Kalman filtering adaptive algorithm based on inertial measurement unit（ IMU） and magnetometer is designed. The algorithm separates the measurement updates of accelerometer and magnetometer

and adjusts the fusion strategy according to measurement precision of magnetometer in real time. At the same time

residual information is introduced to adaptively adjust navigation parameters and improve the performance of the attitude and heading reference system（ AHRS）. Results of simulation analysis on actual flight data show that this method can effectively improve the attitude precision and ensure the flight stability of small UAV in various situations.