基于惯性测量与组合预测的大型舰船甲板态势计算方法研究

项目名称： 基于惯性测量与组合预测的大型舰船甲板态势计算方法研究

项目编号： No.61273056

项目类型： 面上项目

立项/批准年度： 2013

项目学科： 自动化技术、计算机技术

项目作者： 刘锡祥

作者单位： 东南大学

项目金额： 80万元

中文摘要： 从（超）大型舰船实际应用需求出发，以惯性测量方法解决舰船与甲板兴趣点（武器安装点及舰载机起降点）的振荡及摇荡耦合线运动实时计算问题；以基于神经网络与灰色系统的组合预测方法解决甲板兴趣点的运动（摇荡）态势计算问题。主要研究思路有：1）从惯性系统比力方程出发，根据舰船摇荡运动相对于航行运动及其牵连运动等的高频特点，提出以频率为特征的摇荡与航行运动分离算法；2）从子惯导实际测量值出发，寻找子惯导相对于摇摆中心的摇摆牵连线加速度值提取算法，解算并利用多子惯导组网匹配方法优化摇摆中心计算值，构建甲板兴趣点相对于摇摆中心的杆臂运动；3）从舰船兴波运动的灰色特征出发，提出利用灰色系统方法对甲板兴趣点运动态势进行预测，并针对灰色预测拟合能力弱与预测鲁棒性差的不足，提出基于神经网络与灰色系统优势互补的组合预测方法。

中文关键词： 甲板态势预测；摇荡与航行运动分离；Online-SVM预测；惯性导航与测量；

英文摘要： For the requirements in (super) large ship, strapdown inerital navigation system (SINS) is used to calculate oscillating motion of ship and transport motion caused by ship sway of the interest points in deck (where weapons are installed or carrier-based aircraft take off and land), and combination forecasting method based on neural network and gray system is used to predict the deck motion. Main ideas include: 1) Force equation is analyzed and the high-frequency characteristics of ship swing motion is pointed out compared with navigation and other transport motion, and a separation algorithm based on frequency is brought out in which line movements caused by swing and navigation are separated. 2) Transport acceleration in installating position of SINS relative to the swing center is extracted, and swing point is calculated from transport acceleration, also transport accerleration from multi SINSs are used to optimize the swing point. Finally lever-arm movements of interest points relative to the swing center are constructed. 3) Considering the grey characteristics of ship wave-making movement, grey prediction is used to predict the deck motion. According to the shortcomings of grey prediction, combination forecasting method based on neural network and gray system is used to improve the fitting ability and ensure

英文关键词： Prediction for Deck-motion；Separation of Sway and Navigation Motions；Prediction based on Online-SVM；Inertial Navigation and Measurement；