To address the significant deviations between actual and theoretical aiming error values of the radar radome， which are difficult to compensate for precisely using ground data， this paper proposed a multi-model filtering algorithm. Considering the distribution differences of aiming error due to line-of-sight angle dispersion of the seeker body， the paper analyzed the characteristics of low-speed ground motion targets and employed an interactive multi-model approach combined with the measurement-augmented unscented Kalman filter algorithm to improve the accuracy of aiming error estimation. During the terminal guidance phase， the line-of-sight angle of the seeker body gradually decreased. This method processed large-angle data to identify target motion and then small-angle data to estimate aiming error in real time. Simulation results indicate that the proposed algorithm is an effective approach for estimating aiming error under equality constraints， significantly improving estimation accuracy compared to traditional methods.