Research On Key Techniques Of Multi-sensors Integrated Navigation

Due to the high precision, high system stability and better fault tolerance, multi-sensor integrated navigation has been extensively used in many fields such as location, target tracking, weapon guidance and other fields. With the development of multi-information fusion technology, integrated navigation system has been widely applied to obtain more precise status information of vehicle.The research work in this thesis is mainly about:(1) Integrated navigation filter theory: There are two kinds of Sensor registration algorithm, i.e. time registration and space registration. Formulae of two algorithm of time registration are derived in sense of Least Square(LS) and interpolation extrapolation. Maximum likelihood registration algorithm of space registration is also implemented. A simulated experiment is carried out to evaluate the performances of both time and space registration algorithm. In GPS/INS system with colored and cross-correlated noises, the optimal filter solution algorithm is proposed and tested with a numerical example for its accuracy and computation cost aspects.(2) Multi-sensor data fusion algorithm. In this section, several typical kinds of data fusion structure including centralized Kalman filter, distributed Kalman filter and federated Kalman filter are analyzed. The evaluation of state correlation exists in distributed Kalman filter is presented. Both centralized and distributed asynchronous data fusion algorithms are presented and discussed. Alouni’s asynchronous data fusion algorithm which assumes no correlation between observation noise and dynamic noise is investigated. The simulation of different algorithms by Matlab is carried out to evaluate Alouni’s asynchronous data fusion algorithm.(3) Integrated navigation computation strategy. LS principle, Linear or nonlinear Kalman filters are discussed. In addition, the LS-SVM aided Kalman filter algorithm is developed to eliminate dynamic nonlinear errors. A simulation is further implemented to validate its navigation accuracy and computation efficiency.