A Study On The X-ray Pulsar Navigation And Its Enhancement Method

X-ray pulsars Navigation(XPNAV) has attracted much research attention in the recent years. XPNAV is based on the celestial navigation, which has the advantages of the full information, whole airspace, long time, high precision and independence, etc. It not only can be applied in the auxiliary GPS and Beidou satellite navigation system, but also can realize absolute autonomous navigation in deep space. XPNAV plays a vital role in future aerospace development,. In addition, it is of a great significance to the Navigation Warfare in the complicated environment and to the development of the deep space exploration. This thesis studies the phase measurement and enhancement method of XPNAV mainly in four aspects, i.e. the signal denoising, phase measurement, navigation filter algorithm, and navigation augmentation methods. This piece of work is supported by the National Natural Science Foundation of China(Grant No. 61172138) and the National High Technology Research and Development Program of China(Grant No. 2007AA12Z323). The main contribution of this dissertation is summarized as follows.In the first part of the thesis, we study the method to denoise the pulsar signal. The pulsar signal usually contains strong noise because of the interference from the interstellar space material, receiving equipment as well as other factors in the process of communication. A bispectrum denoising algorithm is proposed because the bispectrum is not sensitive to the Gauss noise. This method used the a-Trimmed Phase Filter(TPF) to smooth the bispectrum phase, and an a-Trimmed Median Average Filter(TMAF) is designed to deal with the bispectrum amplitude. Then the power spectrum and phase of signal are reconstructed from the filtered bispectrum values. The signal reconstructed not only eliminates the Gauss noise, but also effectively smooth the non Gauss noise. The pulsar signal noise has significant pulse character, and it is not completely in line with the Gauss distribution. In this thesis, the noise distribution is extended to the a-stable distribution. Based on the a-stable noise distribution, we proposed an estimation method called Quantum Covariance Shaping Least-Squares(QCSLS) and an algorithm called Variable Step Normalized Least Mean P-Norm(VSNLMP). The simulation experiments have demonstrated the good denoising effect of the proposed methods.In the second part of the thesis, we study the TOA measurement method to improve the measurement accuracy of the TOA of X-ray pulsar. The phase estimation of pulsar signal is treated as a general problem of the estimation of the cyclic shift parameter under the multinomial distributed observations. We also proposed a parallel Maximum Likelihood(ML) estimation method to estimate the pulsar phase. The proposed method gives the Poisson model of the X-ray pulsar signal. It uses the multi-Gaussian fitting method to determine the X-ray pulsar profile, and uses a parallel estimation method to improve the performance of the numerical search. In the simulation experiment, we use Crámer-Rao Lower Bound(CRLB) to analyze the performance of the maximum likelihood estimation. In addition, we propose a parallel ML estimation method and compare it with the current ML estimator, and the Nonlinear Least Squares(NLS) estimator. We have proved that the parallel ML estimation can reduce much more computational complexity than the current ML estimator, and has many advantages over the NLS estimator, not only of the asymptotically unbiased, but also of the computational complexity.In the third part of the thesis, we use a Central Difference Kalman Filter(CDKF) algorithm and a robust Kalman algorithm to filter the XPNAV system. We have also designed a simulation experiment to compare CDKF, Extended Kalman Filtering(EKF), Robust CDKF(RCDKF) and Robust Extended Kalman Filtering(REKF) algorithms. The simulation results have proved that the filtering accuracy of CDKF is better than EKF, and the filtering accuracy of RCDKF is higher than that of REKF. The simulation results also show that the filtering accuracy becomes higher with the increasing number of pulsars. Moreover, the pulse time observation is longer as the filtering accuracy becomes higher.In the last part of thesis, we proposed the augmentation method using pulsar position vector to improve the performance of XPNAV. We construct a collimator model to select the vector search strategy. The vector procedure is transformed into a vector optimization problem with the pulse angle of the vector collimator and collimator effective area by using collimator model, and vector search strategy was put forward based on the improved Powell algorithm. Then,the position vectors obtained from the searching procedure were used to formulate the observation equation together with theobservation on the pulsar timing, which were used in the integrated Kalman filtering and federated Kalman filtering. Simulation results indicate that the proposed filtering methods have better performance compared with the pulsar navigation using only the timing data, which also demonstrate the feasibility of the augmentation method. Finally, the federated Kalman filtering had better performance than the integrated Kalman filtering in estimating the position and velocity.