基于改进粒子滤波算法的水下地形辅助导航方法

Underwater Terrain Aided Navigation Method Based on Improved Particle Filter Algorithm

目的解决当前的粒子滤波算法用于水下航行器(AUV)基于极地区域的低分辨率海图时导航精度较低的问题。方法提出了一种带有自抖动及修正的粒子滤波方法(SJCPF),在状态转移过程中引入粒子抖动,每次粒子位置更新时,引入额外的过程噪声,使得传统算法中过度集中的粒子适当向周围发散,改善算法本身及海图分辨率低带来的粒子多样性匮乏。在重采样步骤中,引入相关系数用于修正权值,进一步增加粒子多样性及算法的鲁棒性。结果对传统PF及SJCPF进行仿真,相较于传统PF算法,SJCPF的导航均方根误差降低了27.7%,导航精度及鲁棒性都有显著的提升。结论SJCPF的导航性能优于传统PF,选用皮尔逊相关系数,并在适当范围内选择较大的粒子数量和较高的测量频率,可以兼顾AUV的续航与导航精度。

This paper aims to solve the problem of low navigation accuracy when the current particle filter algorithm is used for autonomous underwater vehicles(AUV)based on low-resolution underwater maps of polar regions.A particle filter method with self-jitter and correction(SJCPF)is proposed,which introduces particle jitter in the state transition process,and introduces additional process noise every time the particle position is updated,so that the over-concentrated particles in the traditional algo-rithm are appropriately moved to the surroundings.It improves the lack of particle diversity caused by the algorithm itself and the low resolution of the chart;in the re-sampling step,the correlation coefficient is introduced to modify the weights to further increase the particle diversity and the robustness of the algorithm.The simulation of traditional PF and SJCPF shows that,com-pared with traditional PF algorithm,SJCPF navigation root mean square error is reduced by 27.7%,navigation accuracy and ro-bustness have been significantly improved.The navigation performance of SJCPF is better than traditional PF.The Pearson cor-relation coefficient is selected,and the larger number of particles and higher measurement frequency is chosen within an appro-priate range,which can take into account the endurance and navigation accuracy of AUV.