摘要：AbstractIn this paper, parameterised state estimation approach has been discussed for a typical passive 2-D bearings only target tracking (BoT) problem. Existing nonlinear state estimation algorithms may not provide good estimation performance in case of initial uncertainty in the range and speed of the target since uncertain and way-off initial conditions can lead to loss of accuracy. Here, the target velocity parameterisation along with range is embedded in extended Kalman filter (EKF), unscented Kalman filter (UKF), new sigma point Kalman filter (NSKF) and shifted Rayleigh filter (SRF). This approach involves partitioning the initial target range and velocity estimates into a number of sub-areas. Then separate weighted filter is applied to each sub-area to estimate the target position and speed. The likelihood of each sub-area is computed based on a Gaussian assumption for the residuals, and the final estimate is calculated by the weighted sum of each filter estimates over all the sub-areas. This overcomes the practical large initial range variation scenarios usually encountered in underwater tracking problems. Then, comparison analysis is performed on the basis of track loss and root mean square error (RMSE) in position.
关键词：KeywordsBearings only target tracking (BoT)Nonlinear filteringParameterised state estimation
