Aiming at the problems of motor failure and composite disturbance in coaxial multi-rotor UAV, a fault-tolerant control algorithm is designed by using adaptive neural network observer.Firstly, the motion model of the coaxial eight-rotor UAV including motor fault and composite disturbance is established, then the composite disturbance is approached by neural network and the fault factor is estimated by adaptive law, and the adaptive neural network observer is designed to estimate the system state.Finally, the backstepping fault-tolerant control law is designed for the attitude angle loop and angular velocity loop,and a filter is adopted to filter the virtual command signal, which suppresses the differential explosion and realizes the gradual stability of the UAV.The experimental results show that the proposed method has better stability and accuracy compared with the adaptive fault-tolerant control method, and the maximum tracking error is only 0.1°, which effectively compensates for the influence of composite interference and rotor motor fault,and improves the flight stability and fault-tolerant performance of the UAV.