Research On Anti-Disturbance Filter Theory And Its Application In Polarization Integrated Navigation System

Anti-disturbance filter as a class of method to fulfill disturbance estimation and rejection, and have attracted more and more attention form industrial area and academia in recent years. It has become a new hot topic in the information fusion area. But there are still many problems are waiting to be solved. This thesis is main focused on the filter design for Polarization light integrated navigation systems with multiple disturbances and missing measurements, and the designed filters are applied to initial alignment. The main contributions of this thesis can be summarized as follows:(1) For a class of linear stochastic systems with unknown disturbance or input, the problem of filter design with little prior information about initial state is researched, and a class of recursive three-step information filter is proposed. By virtue of the proposed filter, the unknown disturbances and states can be estimated simultaneously. Consid-ering that the measurement dropout may occur, the measurement model is modeled as Bernoulli process with a binary variable. When measurements are intermittent and the systems with unknown disturbance, a class of three-step recursive information fil-ter are designed. The stability analysis of the proposed filter is also conducted. The relationships between proposed filter and some existing results are also given.(2) For a class of nonlinear stochastic systems, there may exist multiple types of disturbances, such as measurement noise, norm-bounded disturbance and disturbance that can be described by given model. Considering the measurement may occur dropout, the design of controller under many types of disturbances and missing measurement are researched. For the nonlinear function is available or not, by virtue of the Bernoulli pro-cess with a binary variable, a class of anti-disturbance filter is designed, and a controller which can be used to compensate and reject the multiple types of disturbances is put forward. Based on the LMI. a sufficient condition had been derived, under which the augmented systems are pth moment exponentially stable.(3) For the initial alignment under static base, According to the horizontal mis-alignment angles and azimuth alignment angle have different velocity to reach stable, the state to be estimated are separated into two parts, the estimations of horizontal misalignment angles are obtained by Kalman filter, then they can be used as input for designing disturbance observer which is used to estimate the azimuth alignment angle. Then, a new form of DOBKF approach has been proposed. It is applied to two kinds of error model in this study, which are ten-state error model and five-state error models. The problem of fast estimating azimuth alignment angles has been solved.(4) There are many types of disturbances in calculating polarization information from polarization sensors. Three types of disturbance from different physical source are given and analyzed. The model of calculating polarization degree and polar azimuth angle under multiple types of disturbances is developed. For the six channels polariza-tion sensors, the affection of error propagation from installed angles and scale are also researched.(5) For the initial alignment of Polarization/INS integration navigation systems under stationary, the polarization information can be used as complimentary to INS to improve the accuracy and shorten the alignment time. For one set of polarization sensor, a class of nonlinear measurement model is given. While, for two sets of polarization sen-sors a class of linear measurement model is developed. Based on disturbance estimation, three-step recursive information filter is applied to initial alignment. Compared to the standard Kalman filter, the proposed method is effective and available.