A Novel Adaptive Initial State Estimation Method for Underwater Bearings-Only Target Tracking

In bearings-only tracking (BOT), the uncertainty of the target’s initial state significantly impacts the tracking accuracy. To address this problem, this article proposes a new estimation method called the heading estimation-based improved grey wolf optimization (HEIGWO) algorithm, which integrates bearings-only target motion analysis with the grey wolf optimization algorithm. The method first randomly selects multiple particles within a defined range based on the initial observation measurements, and subsequently incorporates the observation data to continuously optimize the particles’ states, thereby estimating the target’s initial state. The analysis indicates that HEIGWO algorithm demonstrates low computational demand and high estimation accuracy. However, it necessitates a minimum of 300 s of observation data to estimate the target’s heading, which impacts the algorithm’s real-time performance. Therefore, we also propose a hybrid improved grey wolf optimization algorithm based on cubature Kalman filter (CKF-IGWO), which can reduce the estimation time by changing the maneuvering mode of the observation station itself. Finally, we combine HEIGWO, CKF-IGWO, and nonlinear filters to design a novel model for a bearings-only target tracking system, which can adaptively estimate the target states according to the motion mode of the observation station. The experimental results show that in different observation station maneuvering modes, the tracking error of this model is less than $\text{50}\ \text{m}$. It has fast convergence speed and precision, which meet the strict accuracy requirements of BOT.