Research On Signal Processing Method Of X-ray Pulsar-based Navigation

Pulsars are rapidly rotating neutron stars which emit signals with a stable period.The pulsar signals radiated from the X-ray band of the electromagnetic spectrum have the advantage of easy recording by X-ray detectors which are installed on spacecraft and can be miniaturized.In recent years,as a novel autonomous navigation method,X-ray pulsar-based navigation(XPNAV)which can provide the position,velocity,attitude and timing information for vehicles travelling near Earth,into deep space or on interplanetary missions has been extensively researched.At present,the new technique is still in the phase of theoretical research and feasibility verification,and some key techniques still need further research.Concentrating on the key technical problems unresolved in the XPNAV system,this dissertation investigates the period estimation of X-ray pulsar,pulse profile reconstruction with high signal-to-noise ratio(SNR),X-ray pulsar pulse time-of-arrival(TOA)estimation and the navigation filtering method of XPNAV.The main research works include:1.This dissertation summarized the basic physical characteristics of pulsars,and studied the influence of the changes of pulsar frequency and its derivatives on pulsar pulse profile.Then,we analyzed the statistical properties of the photons of X-ray pulsar,and established the X-ray pulsar signal model.And then,the transformation equation of the photon arrival time from the spacecraft to solar system barycenter(SSB)is elaborated in detail.2.A new method for period estimation of X-ray pulsars with the short-time observations based on frequency subdivision is proposed.We analyzed the unevenly spaced characteristics of the arrival time of X-ray pulsar photons.Based on this characteristics,a continuous Lomb Periodogram(CLP)is constructed.Then the initial frequency of X-ray pulsar is firstly calculated using Fast Lomb Periodogram,and the frequency subdivision is performed near the initial frequency layer by layer.Finally,a refined period,which has a higher precision,is achieved by calculating CLP in the high precision frequencies.This algorithm takes into account the characteristics of the pulsar signal and estimates the pulsar period in less subdivision frequencies,which can avoid the computational complexity increased tremendously caused by oversampling as well as is helpful for the quasi real-time update of the pulsar period.Experiment results of the RXTE observation data indicate that the proposed method is excellent in pulsar period estimation with better performance than that of the fast Lomb method,FFT and the chi square method(efsearch),and show that the observed data of Crab pulsar within 120?130 s can help to achieve the period estimation precision of 10-8s.3.An efficient pulse profile reconstruction method based on the idea of Compressed Sensing(CS)is proposed to improving the SNR of pulsar pulse profile of the short-time observation data and reducing computation complexity.We studied the sparse representation of the pulse profile and analyzed the conditions that the sensing matrix needs to be satisfied when the signal is reconstructed accurately.Based on the randomness of the diagonal matrix with the element of ±1,and the orthogonality of the Hadamard matrix as well as the screening matrix designed in this paper,an improved Hadamard sensing matrix is constructed with the controllable dimensions and random independence.In addition,we designed the algorithm for pulse profile reconstruction.Experiment results of the RXTE observation data demonstrate the superiority of the proposed algorithm.The results also indicate that the pulse profile can be reconstructed with a 100%probability when the sampling rate is greater than or equal to 0.6.Compared with the Epoch Folding method,the proposed CS method has a better performance with the improvement of the SNR at least 15.859 dB.4.A pulse TOA estimation algorithm based on generalized cross correlation is proposed to overcome the contradiction between the estimation accuracy and the computational complexity.We analyzed the statistical properties of the photons of X-ray pulsar and analyzed the distribution properties of epoch folding noise.Based on the noise variance as the optimal weight value,the pulse TOA estimation method is deduced,and the estimation performance of the proposed method is proved theoretically.Experiment results of the RXTE observation data indicate that the precision of the proposed method is higher than the method of the FML,NLS,WNLS and CC.With the increase of observation time,the mean variance of the proposed method decreases fastest.5.Aiming to reduce model errors and the noises for improving the accuracy of the XPNAV,this dissertation proposed a nonlinear predictive strong tracking unscented Kalman filter(NPSTUKF).This method uses the model errors and observation errors to construct criterion function.Then,the minimum model error can be obtained by minimizing this criterion function,and the system state model can be modified by using the minimum model error.And then,forecast the state of the next moment.In order to realize high precision navigation filter,in the process of status updates,the real-time adjustment of gain matrix need to be done,which will ensure that the state estimation error variance of the next moment is the smallest and the residual sequence of different time is orthogonal.The simulation results show that the accuracy of the proposed mothod is significantly higher than that of EKF,UKF and STUKF,especially when the model error and the noise are large.