To resolve the shortcoming of the traditional Kalman filtering, a new method which combines the fuzzy inference system with the traditional Kalman filter was presented by monitoring the covariance between abstract residual and actual residual, this algorithm modified recursively the gain coefficient of the filter to control the Kalman filter. Finally, the optimal estimation was achieved. The simulation results of the anti-ship missile altitude measurement show that this algorithm can improve the precision of the result, adapt the system's noise variation better, and it is effective and practical.