The study examines the accurate establishment of an attitude angle calculation model for UAVs influenced by large motion accelerations. Accelerometer output shows a decreased proportion of effective gravity acceleration information， while polarization sensors exhibit jitter. First， polarization sensor integrated navigation was introduced， along with the orientation principle of polarization navigation. Then， a cascaded attitude solution algorithm was built using the double-vector Gauss-Newton method and the SHKF （Sage-Husa Kalman） algorithm. This algorithm observed UAV attitude information based on multiple sensors. Next， based on analyzing the accuracy of accelerometer measurements under motion acceleration， a trust factor was proposed to mitigate the impact of motion acceleration on attitude calculation. This algorithm could suppress the effects of motion acceleration generated by high-speed UAV bodies. To verify this algorithm's feasibility， experiments were conducted on a polarization/MIMU （Micro Inertial Measurement Unit） integrated navigation platform. Results indicate a 30% improvement over PI and EKF algorithms in static and dynamic environments. The algorithm suppressed attitude deviation caused by non-gravity acceleration under motion acceleration influence. It improved the accuracy of attitude calculation under motion acceleration， ensuring normal UAV flight.