Research On Some Key Technologies Of Underwater Target Motion Analysis

Target motion analysis (TMA) originates from hydroacoustic passive sensor observation system. The fundamental problem is to estimate the target motion parameters using bearings-only measurements. The main features are nonlinearity and system observability depending on observer maneuvering. It is extensively studied after the problem is proposed, attracts a lot of attention of science and engineering technical personnel, and is funded by many national government agencies. Though plenty developments and results have been obtained, TMA theory is still not enough perfect and TMA results cannot satisfy need in pracrice. This dissertation presents some TMA key techniques with respect to TMA observability, TMA algorithm (including maneuvering TMA) and optimal observer trajectory. The main contents are stated as follows:(1) The analytic expression of Gram matrix (GM) determinant for TMA observability is derived by linearizing the measurement equation. A sufficient and necessary condition for observability of bearings-only system is proved. The bearings-only criterion uses only the bearings-information measured by sensor, neither its differentiation nor the solution of a differential equation. Then, a judgment condition of target maneuver detection is obtained by utilizing the criterion. On the other hand, the unobservability of bearings-only observation system is analyzed. A bearings-only system is unobservable if and only if for all bearing measurements the observer trajectory is equivalent to a uniform line.(2) An iterative formula is derived about nonlinear least square (LS) algorithm with 4-dimensional states for TMA. The two sufficient and necessary conditions of NLS and LS algorithms are analyzed. Then, equivalence properties of NLS and maximum-likelihood models are presented.(3) A new formula of calculating target distance is proposed by utilizing the geometric relation between observer, target and its suppositional bearing. Then, a two-step model is established for TMA. The uniqueness of TMA solution is analyzed. A sufficient and necessary condition of unique solution of target motion parameter is proved.(4) In order to diminish the errors of distance estimation, an optimal observer maneuver scheme is presented by analyzing distance estimation error. An optimal observer maneuver scheme analytic solution is derived by using variation method and extremum principle. For the convenience of application a simplified suboptimal observer maneuver scheme is proposed. Finally, at the cost of maximizing increment of determinants of FIM and GM, two optimal trajectory equations are obtained, respectively.(5) A maneuver identification algorithm of piecewise line motion target is presented under non-maneuvering observer. The algorithm uses measurement bearings and bases on sensitivity functions. On the same time, under effective observer maneuver, a maneuver tracking of piecewise line motion target is dealt with, and an iterative integrated identification algorithm of parameters and states is presented.(6) An optimization model of parameter estimation is built for a kind of observation systems, where the right functions of state equation are piecewise continuous and differentiable. An iteration algorithm for parameter estimation is presented and a formula of Fisher Information Matrix is derived. The algorithm can be applied likewise to target tracking in piecewise line, circle arc and constant acceleration motion.