A self-calibration Kalman filter(SKF)method

whose model and recursive algorithm were established

was presented. In most practical cases

such as deep space exploration and engine fault diagnosis

because of the effect of unknown inputs

such as gust

fault and unknown system error

the well-known Kalman filter will lead to greater filtering error in recursive process. To solve this problem

the proposed SKF

which is applied to estimate and compensate the unknown inputs

efficiently reduces the effect of the unknown inputs and enhances filtering accuracy. For some spacecraft navigation simulation

the mean and variance of estimated state errors by SKF decreased by at least 400% and 300%

respectively. The SKF method can be effective to improve the performance of filter

simple to calculate and easy to apply in engineering.