3D Bearings-Only Tracking Using a Range-Parameterized Shifted Rayleigh Filter Based on a Novel Innovation Variance

A Range-Parameterized (RP) improvement for the Shifted Rayleigh Filter (SRF) was proposed to address the problem of three-dimensional bearings-only tracking (3D-BOT) under single-station and clutter-free conditions, with a focus on resolving the dependency of nonlinear algorithms on prior information in 3D-BOT problems, thereby enhancing the accuracy and error convergence speed of estimation algorithms. To perform RP improvement for SRF in 3D-BOT based on the unique "augmented" measurement model, a novel innovation variance is introduced using the tangent plane approach to update the weights of the range intervals. To effectively compare the performance of filters in 3D-BOT, the observability metric was extended to 3D to determine the observability of 3D-BOT, and simulations were conducted comparatively under different observability conditions. Using root mean square error (RMSE) and runtime as evaluation metrics to evaluate the simulation results, the performance of SRF3D, RP-SRF3D, EKF, and EPKF is compared. The results demonstrate that the RP method based on a novel innovation variance exhibits good numerical stability in the weight update process. It can achieve fast convergence of weights under conditions of high noise, high nonlinearity, and lack of prior information.