A transfer alignment algorithm based on multi-source dynamic error compensation is proposed to solve the problem that the performance of airborne optoelectronic payload is short due to the accumulation of device noise and the inability of static alignment caused by installation error in complex dynamic environment. The "velocity + attitude" matching model between the main inertial navigation system of UAV and the micro-inertial navigation system of photoelectric payload is constructed, and the navigation information of the high-precision main inertial navigation system is used to correct the error of the low-precision sub-inertial navigation system. Combined with the 21-dimensional state space equation and measurement equation, the real-time estimation and compensation of multi-source errors, such as attitude misalignment angle, sensor bias, installation error and flexure deformation, are realized. Simulation results show that the proposed method can significantly improve the alignment accuracy. After compensating the lever arm effect, the speed error can be controlled within 0.01 m/s. The flexure deformation angle compensation significantly reduces the fluctuation amplitude of the misalignment angle in each direction, enhances the system stability, and effectively controls the misalignment angle at a lower level. The error compensation strategy is effective.