Aiming at the problem of noise in UAV speed signal estimation, a filter algorithm based on strong tracking filter and Kalman filter is proposed to estimate unknown signal. When predicting the prior error covariance, the fading factor is introduced to reduce the influence of the past data on the filtering effect, and thus improve the response speed and accuracy of the filter. The speed signal estimated by the UAV autopilot is filtered. The results show that the signal response of strong tracking Kalman filter is faster, the overshoot is smaller and the precision is higher.