一种有效的水下目标运动分析算法与仿真研究

Simulation Study on Algorithm of Underwater Target Motion Analysis

研究了基于纯方位观测的水下目标运动分析(TMA)原理及方法,针对常用粒子滤波算法在水下目标运动分析中存在的后验概率选择问题及粒子退化现象,通过分析问题产生的原因,提出了一种基于改进粒子滤波算法的水下目标运动分析方法。该算法首先结合了扩展卡尔曼滤波与粒子滤波算法(EKF-PF)各自的优势,同时考虑到粒子退化现象,并将马尔科夫链蒙特卡洛方法(MCMC)应用于EKF-PF算法中。与传统粒子滤波算法相比较,该算法不仅提高了重要性密度函数准确度,同时还克服了粒子退化问题,而且对重采样带来的采样枯竭也有很好的抑制作用。通过仿真实验表明该算法有效,且估计精度有较大的提高。

The basic principle and method for underwater target motion analysis based on bearing was studied. For the problem of posterior probability and particle degeneracy for SIR, a modified particle filter algorithm was proposed, which perfectly combined the extended kalman and particle filter （EKFPF） and adopted the Markov chain Monte Carlo method to overcome the effect of particle degeneracy and solve the problem of the sample impoverishment as re-sampling. Simulation results show the feasible of the algorithm and better estimation performance for underwater bearing-only target motion analysis （BO-TMA）.