The Application Of UKF Algorithm In An Underwater Target Tracking System

In both military and civilian fields, some moving targets need to be tracked more reliably and more accurately. It's a challengeable problem to track maneuvering targets which have the complexity, the randomness and the multiplicity performance. Targets tracking are more difficult in theory and engineering.Firstly, this dissertation investigates the Kalman filter (KF) in the linear systems, whose processing model and measurement model are linear, but the most of real systems are nonlinear. We usually use the extended Kalman filter (EKF) for nonlinear state model and observation model. First, linearization process is necessary for these models of nonlinear state and observation, and then the signals are estimated by means of the Kalman Filter. Unfortunately, the larger error will be produced after the linearization processing, the error will lead EKF filter to divergence easily. To resolve this problem, unscented Kalman filter (UKF) algorithm is introduced in the dissertation. The probability density of the nonlinear function is approximated by some typical sigma points in the UKF way. UKF compensates the flaws of the EKF, and has better accuracy.For the underwater tracking target system based on UKF, the dissertation has completed the simulations for tracking of underwater moving-target in different conditions. UKF was also compared with EKF according to simulating results, and UKF is better than EKF on tracking-filter performance. Underwater target tracking with the higher maneuvering ability can be well solved by the algorithm of interactive multiple model. The combination of IMM and UKF is called as IMM-UKF. By the computer simulation for target tracking of higher maneuverability, the simulation results shows that the IMM-UKF algorithm is more convergence and better accuracy. But the process time for calculating IMM-UKF is longer than for UKF, so IMM-UKF will rely on faster arithmetic speed then UKF for the real-time processing. To a high-speed underwater target tracking system, it is necessary to real-time processing. The UKF program for the real-time tracking to high-speed underwater target has be run on the TMS320C64x DSP hardware platform. The result of experiments shows a better real-time processing performance and filter precision.