State Estimation With Incomplete Measurements And Observer Platform Maneuver Model For TMA

Target motion analysis (TMA) technique has become one of active research fields in the world at present. The performance of target tracking system is related to factors that include with target dynamic state, detection ability of sensor, background noise origin and state estimator. For the target tracking system on the background of such as environmental monitoring, air surveillance, maritime search and rescue, and navigational location-tracking, there exist incomplete measurements that probability of detection is less than unity owing to sensor detection ability limited and disturbance of environment meteorology. In this dissertation, two importance problems have been investigated involved in TMA. One is the estimation problem of TMA with incomplete measurements; the other is the optimal observer platform maneuver model in passive target localization and tracking system.The first half of this thesis has done some pilot work on estimation performance and properties for TMA based on incomplete measurements.First, since the stochastic detection/miss sequences are introduced, both Cramer-Rao lower bound (CRLB) and modified Riccati difference equation (MRDE) become stochastic, which their statistical properties have some new changes. The CRLB mathematical model for nonlinear systems and linear systems, as well as the nature of the statistical and compute method are discussed in depth. The condition for the statistical convergence properties of the CRLB are given based on the expected detection/miss sequences. Meanwhile, the computation of the upper and lower bounds for static CRLB is reformulated as the iteration of a linear matrix inequation (LMI) feasibility problem.Secondly, a correlation phenomenon presents between index CRLB and location of missing data (LMD) for incomplete measurements systems. A novel result is given that CRLB and MRDE for deterministic system is monotone decreasing to LMD in sense of Lyapunov stability theory. Then, a group of new upper and lower bounds of CRLB are designed for a given probability of detection, which hold smaller amount of calculation.Thirdly, the necessary and sufficient conditions for convergence of estimation error system with incomplete measurements is given in the sense of statistics, as well as the steady-state statistical error covariance method of calculation. In order to facilitate the engineering design, a solution method of filter index is given with covariance constrains of the upper and lower bounds.On the other hand, passive target location and tracking system is an important branch of the TMA. The maneuver of observer platform is necessary but not sufficient condition for system observability. The maneuver of observer platform occupies an important position due to special observability.On the latter part of this dissertation, unobservable and maneuver-mode of platform for the passive target location and tracking are investigated, and the filtering algorithms are studied.Fourthly, the unobservable results for BOT are proved by using a vectors projection method.In view of a special issue of observability in passive target tracking, the unobservable for 2D and 3D tracking system are studied. Utilizing the method of vectors transformation, the conditions of unobservable have been drived for different geometric situation of the observer and target. The proposition that the probability of detection do not influence on the unobservable of system is given.Fifthly, considering the passive target location system, a new optimal maneuver strategy is designed at current time based on maximizing the determinant of the Fisher information matrix (FIM). The result that there exist only two optimal trajectories is proved, and the optimal maneuver strategy and the optimal trajectories equations are proposed. The feasibility of the sequence of maneuver controls is described visually by geometry. It is the optimal maneuver in any time not only in the end of the observational times. Using analytical method, the optimal trajectory equations are also given based on maximizing the change rate of angles. The different characters of the optimal trajectories base on two performance indexes are compared. In addition, a concise and practical observer platform sub-optimal maneuver strategy is proposed for passive target tracking system.Finally, the model of group of EKF filter is designed based on first-range partition for BOT. The proposed theoretical is applied to a type of maneuvering target tracking analysis and extended to two-channels with different detection probability under incomplete measurements. The 3D target tracking system that the bearings and elevation angles measurements come from different channels is analyzed to construct nonlinear CRLB model in sense of theoretic and design approximate solution algorithm.