基于改进无迹卡尔曼滤波的动力定位系统状态估计  

State Estimation of the Dynamic Positioning System Based on the Improved Unscented Kalman Filter

作　　者：王帅洋 杨宣访[1] 胡致远 王家林[1] 李厚朴[1] WANG Shuaiyang;YANG Xuanfang;HU Zhiyuan;WANG Jialin;LI Houpu(School of Electrical Engineering,Naval University of Engineering,Wuhan 430033,China)

机构地区：[1]海军工程大学电气工程学院,武汉430033

出　　处：《船舶工程》2022年第3期102-109,178,共9页Ship Engineering

基　　金：国家自然科学基金资助项目(41771487);湖北省杰出青年科学基金资助项目(2019CFA086)。

摘　　要：动力定位(DP)船舶状态估计的准确性是影响其在海面上沿期望航迹运行或位置固定的关键因素。在DP状态估计研究中,当遇到观测噪声反常或噪声协方差与算法不符等情况时,无迹卡尔曼滤波(UKF)无法调整算法参数以适应海洋环境的变化,严重影响着状态估计的精度。鉴于此,提出一种基于误差序列协方差估计的自适应UKF,利用观测变量残差更新观测噪声协方差矩阵R。设计基于自适应UKF的状态估计器,对DP船舶纵荡、横荡和艏摇3个重要状态变量进行估计。数值仿真结果表明,提出的自适应UKF能明显降低纵荡、横荡和艏摇3个状态变量的估计误差,状态估计的准确性和滤波平滑性均优于传统UKF算法。The accuracy of dynamic positioning(DP) ship state estimation is the key factor to achieve the desired track or fixed position on the sea. In the dynamic positioning state estimation research, if the observation noise is abnormal or the noise covariance is inconsistent with the algorithm, Unscented Kalman Filter(UKF)cannot adjust the algorithm parameters to adapt to the changes of marine environment, which seriously affects the accuracy of state estimation. In view of this, an adaptive UKF based on the covariance estimation of error sequence is proposed, which updates the covariance matrix of observation noise(R) by using the residual of observation variables. A state estimator based on adaptive UKF is designed to estimate the three important state variables of dynamic positioning ship, namely, surge, sway and yaw. The numerical simulation results show that the proposed adaptive UKF can significantly reduce the estimation error of the three state variables of surge, sway and yaw, and the state estimation accuracy and filtering smoothness are better than the traditional UKF algorithm.

关 键 词：船舶动力定位 状态估计 自适应无迹卡尔曼滤波 误差序列 

分 类 号：U664.82[交通运输工程—船舶及航道工程] TP273[交通运输工程—船舶与海洋工程]