摘要
针对自主水下航行器(autonomous underwater vehicle,AUV)导航定位技术的发展需求,提出了水下目标的3种非线性滤波估计方法.首先,分别介绍了扩展卡尔曼滤波(extended Kalman filter,EKF)、无迹卡尔曼滤波(unscented Kalman filter,UKF)和粒子滤波(particle filter,PF)的基本原理和实现步骤.其次,针对PF算法存在粒子退化现象,提出了重采样算法,并通过数值仿真验证该算法的有效性.然后,为了解决PF算法中粒子贫化现象,提出了一种基于萤火虫算法的粒子滤波算法(FA-PF).最后,在非线性环境下通过仿真实验对EKF、UKF、FA-PF算法的滤波性能进行对比分析,重点研究非线性强度及噪声特性对估计精度的影响.研究结果表明:重采样能够在一定程度上减轻粒子退化问题.在弱非线性高斯环境下,EKF、UKF、FA-PF算法的估计精度较为接近;在强非线性高斯环境下,UKF和FA-PF算法的跟踪性能明显优于EKF;在非线性非高斯环境下,FA-PF算法跟踪精度最高.
Aiming at the development of navigation and positioning technology of autonomous underwater vehicle(AUV),three nonlinear filtering estimation methods for underwater targets are proposed.First,the basic principles and implementation steps of Extended Kalman Filter(EKF),Unscented Kalman Filter(UKF)and Particle Filter(PF)are introduced respectively.Secondly,in view of the particle degradation phenomenon of the PF algorithm,a resampling algorithm is proposed,and the effectiveness of the algorithm is verified by numerical simulation.Then,in order to solve the particle depletion phenomenon in the PF algorithm,a particle filter algorithm based on the Firefly algorithm(FA-PF)is proposed.Finally,in a nonlinear environment,a comparative analysis of the filtering performance of EKF,UKF,and FA-PF algorithms is carried out through simulation experiments,focusing on the impact of nonlinear strength and noise characteristics on the estimation accuracy.The research results show that:resampling can alleviate the problem of particle degradation to a certain extent.In a weakly nonlinear Gaussian environment,the estimation accuracy of the EKF,UKF,and FA-PF algorithms are similar;in a strongly nonlinear Gaussian environment,the tracking performance of the UKF and FA-PF algorithms is significantly better than that of EKF;in a nonlinear non-Gaussian environment the FA-PF algorithm has the highest tracking accuracy.
作者
许奕
邢传玺
万志良
姜佳圆
XU Yi;XING Chuan-xi;WAN Zhi-liang;JIANG Jia-yuan(School of Electrical and Information Technology,Yunnan Minzu University,Kunming 650500,China)
出处
《云南民族大学学报(自然科学版)》
CAS
2023年第2期231-239,共9页
Journal of Yunnan Minzu University:Natural Sciences Edition
基金
国家自然基金(61761048)
云南省基础研究专项面上项目(202101AT070132).
关键词
自主水下航行器
粒子滤波
萤火虫算法
重采样
autonomous underwater vehicle
particle filter
firefly algorithm
resampling