An Adaptive Kalman Filter for Mars Spacecraft Acquisition Phase

Celestial navigation is an energy saving and efficient way of autonomous navigation for deep space probes. Kalman filter has been successfully applied in the Celestial navigation system. During the acquisition phase, due to the complex dynamic environment, unmolded acceleration error and the ephemeris error etc. may cause the statistics of process noise uncertainty. To overcome the problem, a method named AQSCKF based on the trend of the innovation sequences and residual sequences to scale the process noise covariance matrix is proposed in this paper. In the first place, it calculates the scale factor based on the innovation and residual separately. Then, by comparing the trend of the two factors, the scale factor of the new method is set as the average. In addition, the navigation performance of traditional SCKF, the method only using innovation or residual to scale Q and AQSCKF is also compared by simulation. The simulation results show that the new method yields better performance than the traditional methods in solving the problem caused by the uncertainty of the process noise, furthermore it also shows a good stability.