Performance Analysis and Improved Algorithm of Gaussian Optimal Filtering for Underwater Target Tracking
Chengzhen Zhang, Yuanming Ding, Yang Yang · 2021
In order to accurately predict the state of underwater dynamic targets, the performance analysis of Gaussian optimal filtering and its improved algorithm are proposed in this paper. The state estimation performance of Gaussian Hermite Kalman filter (GHKF) and Gauss Hermite Rauch-Tung-Striebel smoother (GHRTS) is analyzed in detail, and the idea combined with particle filter algorithm-Gaussian Hermite Kalman particle filter (GHKF-PF) and Gauss Hermite Rauch-Tung-Striebel smoother particle filter (GHRTS-PF) is proposed. This analysis is done through the variance of Gaussian noise and the change of Gaussian mixture noise. This performance-based research is carried out under the background of bearing-only tracking (BOT) phenomenon. All the experiments are to find out the RMSE between the real state and the predicted state of the object. The numerical results based on independent Monte Carlo simulation show that under the condition of white Gaussian noise, GHRTS has better performance than GHKF, GHKF-PF is better than GHKF, GHRTS-PF and GHRTS, and GHRTS-PF is better than GHKF-PF. However, under Gaussian mixture noise, the performance of GHRTS is lower than that of GHKF, and the performance of GHKF-PF is better than that of GHKF, GHRTS-PF, which is better than that of GHRTS,GHRTS-PF and lower than that of GHKF-PF.