基于惯性测量与组合预测的舰船瞬时线运动计算方法研究

发布时间：2018-04-17 08:42

本文选题：SINS + 运动分离　；参考：《东南大学》2015年硕士论文

【摘要】：舰船在航行过程中受到海浪、海风以及其它干扰力的激励,将会产生六自由度的摇荡运动,对舰载机着舰、武器控制以及舰员和舰载设备等均产生不利影响。实时求取出舰船振荡信息,并对未来一段时间内的舰船振荡运动进行预测,以供舰载机及时调整路径,可避免起飞和降落过程中事故的发生,从而提高舰载机的生存能力。本文主要针对舰船振荡运动的分离与预测等相关问题展开研究。首先,详细分析了舰船在海上航行的运动状态,研究了海浪波的形成机理以及海浪波的建模方法,给出了舰船瞬时摇荡运动的模型方程,该方法比单一频率的建模要更接近实际情况。详细阐述了用于振荡运动测量的捷联式惯性导航系统(SINS)的解算原理,给出了增加振荡运动测量的SINS导航解算方法；通过对振荡运动解算结果中的频率成分的分析,给出了基于数字滤波的振荡运动提取方法。其次,分析并比较了IIR数字滤波器和FIR数字滤波器性能特点,选择了延迟时间较短的IIR数字滤波器作为振荡运动分离滤波工具。详细阐述了11R滤波器的工作原理和设计过程,以及基于解算结果频率组成的滤波器参数的选取方法。仿真验证了所设计滤波器的有效性。最后,针对舰船振荡运动的随机性及非线性特性,选取具有非线性、自适应性、自学习能力以及数据融合能力的神经网络方法对舰船振荡运动进行预测。仿真分析、比较了BP神经网络、RBF神经网络和BP-RBF组合神经网络三种预测方法对舰船三轴振荡运动的预测能力。仿真实验结果表明：BP-RBF组合预测方法既具有RBF网络的学习收敛速度的快速性,又具有BP网络的泛化性能。综合认为,在一定条件下组合预测方法比单一预测方法具有优势,可在保证预测精度的基础上提高了预测速度。
[Abstract]:During the course of navigation, the ship is stimulated by waves, sea winds and other disturbance forces, which will produce a six-degree-of-freedom rocking motion, which will have a negative impact on the landing, weapon control, crew and shipboard equipment of the carrier aircraft.The ship oscillation information can be extracted in real time and the ship oscillation motion in the future can be predicted so that the carrier aircraft can adjust the path in time to avoid the accidents during take-off and landing and thus improve the survivability of the carrier aircraft.In this paper, the separation and prediction of ship oscillating motion are studied.First of all, the motion state of ship sailing at sea is analyzed in detail, the formation mechanism of wave wave and the modeling method of wave wave are studied, and the model equation of ship instantaneous rocking motion is given.The method is closer to the actual situation than the modeling of a single frequency.The calculation principle of strapdown inertial navigation system (sins) for oscillatory motion measurement is described in detail, and the SINS navigation method for increasing oscillatory motion measurement is given, and the frequency components in the result of oscillatory motion measurement are analyzed.A method for extracting oscillatory motion based on digital filtering is presented.Secondly, the characteristics of IIR digital filter and FIR digital filter are analyzed and compared, and IIR digital filter with short delay time is chosen as the tool of oscillatory motion separation filter.The working principle and design process of 11R filter are described in detail, and the method of selecting filter parameters based on the frequency component of the calculated results is also presented.The effectiveness of the designed filter is verified by simulation.Finally, according to the randomness and nonlinear characteristics of ship oscillation motion, a neural network method with nonlinear, adaptive, self-learning ability and data fusion ability is selected to predict the ship oscillation motion.The prediction ability of BP neural network and BP-RBF combined neural network for ship triaxial oscillation motion is compared.The simulation results show that the proposed method not only has the rapidity of learning convergence rate of RBF network, but also has the generalization performance of BP network.It is concluded that the combined forecasting method is superior to the single prediction method under certain conditions, and the prediction speed can be improved on the basis of ensuring the prediction accuracy.
【学位授予单位】：东南大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U674.70

【相似文献】