﻿ 基于53H滤波的船舶综导信息在线平滑方法
 舰船科学技术  2019, Vol. 41 Issue (1): 121-124 PDF

An online ship integrated navigation information smoothing method based on the improved 53H filtering
LV Wu, GENG Jian-ning
CSSC Marine Technology Co.,Ltd, Beijing 100070, China
Abstract: In view of the existence of outliers in navigation information received by ship integrated navigation system, an online smoothing method based on the improved 53H filtering method is proposed in this paper. The method uses the median of several consecutive information to predict the information at the current moment and then compares it with the measured value. If the difference between the two values exceeds a certain threshold, the predicted value is used to replace the measured value to achieve on-line smoothing. By applying this method to the experimental data of sea trial, the result shows that compared with the traditional least squares sliding filter and 3 times in 5 points smoothing algorithm, this method is simple, easy to implement and has a good online smoothing effect.
Key words: integrated navigation     53H filtering     online smoothing     least squares sliding filtering     3 times in 5 points smoothing
0 引　言

1 改进的53H滤波算法

53H滤波算法最先由Tukey提出[13]，其基本思想是产生一个曲线的平滑估计，然后通过将测量值与这一估计值进行比较来识别异常点。其基本步骤如下：

1）假设在线测量的数据序列为 $x(i)$ ，根据 $x(i)$ 构造一个新序列 ${x_1}(i)$ ，方法为：从 $x(1)$ $x(2)$ $x(3)$ $x(4)$ $x(5)$ 选择中值作为 ${x_1}(3)$ ，然后从 $x(2)$ $x(3)$ $x(4)$ $x(5)$ $x(6)$ 中选出中值作为 ${x_1}(4)$ ，依次类推，直到当前的数据 $x(i)$

2）按照步骤1）类似的方法，从 ${x_1}(j)$ 中相邻的3个数据中选取中值构成 ${x_2}(k)$

3）最后由序列 ${x_2}(k)$ 按照如下方式构成 ${x_3}(l)$

 ${x_3}(l) = 0.25 \cdot {x_2}(l - 1) + 0.5 \cdot {x_2}(l) + 0.25 \cdot {x_2}(l + 1){\text{，}}$ (1)

4）如果有下式成立，则用 ${x_3}(l)$ 代替 $x(l)$

 $\left| {{x_3}(l) - x(l)} \right| > T{\text{，}}$ (2)

5）将 $x(i)$ 的前8个和最后8个数据点反序排列，其余数据不变生成序列 ${x'}(i)$

6）对序列 ${x'}(i)$ 重复前4个步骤，形成新的 $x_3'(l)$ 序列，用新序列中的 $x_3'(5)$ $x_3'(6)$ $x_3'(7)$ $x_3'(8)$ $x_3'(l - 7)$ $x_3'(l - 6)$ $x_3'(l - 5)$ $x_3'(l - 4)$ 分别替代序列 $x(i)$ 中的 $x(4)$ $x(3)$ $x_3'(2)$ $x(1)$ $x(l)$ $x(l - 1)$ $x(l - 2)$ $x(l - 3)$

2 基于改进53H滤波的信息在线平滑

 图 1 船舶综合导航信息综合处理流程图 Fig. 1 Flow chart of ship integrated navigation information processing

3 试验验证

 图 2 惯导测量经纬度 Fig. 2 Longitude and latitude measured by inertial navigation system

 图 3 惯导测量速度 Fig. 3 Velocity measured by inertial navigation system

 图 4 惯导测量姿态角 Fig. 4 Attitude angle measured by inertial navigation system

 图 5 东向速度滤波结果 Fig. 5 Filtering results of eastern velocity

 图 6 北向速度滤波结果 Fig. 6 Filtering results of north velocity

4 结　语

 [1] 曲全福, 陈志刚, 高洪宇. 新型综合船桥系统[J]. 中国惯性技术学报, 2011, 19(3): 325-328. [2] 韩剑辉, 许镇琳, 赵承利, 等. 船舶综合导航系统应用技术[J]. 天津大学学报, 2010(2): 121-125. DOI:10.3969/j.issn.0493-2137.2010.02.005 [3] 韩剑辉, 许镇琳, 舒东亮. 综合船桥多源导航信息融合技术研究[J]. 控制工程, 2010(4): 466-469. DOI:10.3969/j.issn.1671-7848.2010.04.013 [4] 张亚崇, 陆志东, 雷宏杰. 多传感器容错综合导航系统技术及其应用研究[J]. 弹箭与制导学报, 2009(4): 51-54. DOI:10.3969/j.issn.1673-9728.2009.04.014 [5] 卢元磊, 何佳洲, 安瑾, 等. 几种野值剔除准则在目标预测中的应用研究[J]. 指挥控制与仿真, 2011, 33(4): 98-102. DOI:10.3969/j.issn.1673-3819.2011.04.023 [6] WANG Rong, XIONG Zhi, LIU Jian-ye, et al. Chi-square and SPRT combined fault detection for multisensor navigation[J]. IEEE Transactions on Aerospace and Electronic Systems, 2016, 52. [7] Joelle Al Hage, Nourdine Aïttmazirte, Maan E. El Najjar, et al. Fault detection and exclusion method for a tightly coupled localization system[C]// 2015. [8] Se Hyun Yun, Chul Woo Kang, Gook Park Chan. Reducing the computation time in the state chi-square test for IMU fault detection[C]// 2014: 879–883. [9] Ruan Wen. Improved residual χ~2 inspection in integrated navigation fault detection of application[J]. Electronic Measurement Technology, 2012. [10] LIU Lian-sheng, FU Jing. Improved state-χ2 fault detection of navigation systems based on neural network[C]// 2010: 3932–3937. [11] 金学军. 基于最小二乘拟合的外弹道测量数据野值剔除方法[J]. 四川兵工学报, 2011, 32(1): 20-23. DOI:10.3969/j.issn.1006-0707.2011.01.007 [12] 陈伟利. 基于最小二乘B样条逼近的观测数据野值剔除方法[J]. 飞行器测控学报, 2001, 20(4): 50-53. [13] 杨莉, 张理. 在线监测数据剔点处理算法的研究[J]. 高压电器, 2000, 36(5): 3-6. DOI:10.3969/j.issn.1001-1609.2000.05.001