【作者】 贾均；

【导师】 徐德民；

【作者基本信息】 西北工业大学 ， 武器系统与运用工程， 2004， 硕士

【摘要】 多传感器信息融合是指对来自多个传感器的数据进行多级别、多方面、多层次的处理，从而产生新的有意义的信息，而这种新信息是任何单一传感器所无法获得的。多传感器信息融合在国防上已发展成为一个十分活跃的热门研究领域。 本论文介绍了多传感器信息融合技术的发展过程及其在鱼雷等水中兵器中的研究意义和研究现状，综述了信息融合算法以及国内外研究动向。研究了多传感器信息融合技术，并对多传感器的状态估计和航迹融合算法在线导鱼雷中的应用进行了深入研究。主要研究工作有以下几个方面： (1) 系统研究了信息融合的基本模型，从功能和结构上对多传感器信息融合进行了合理的分类。给出了四级数据融合处理功能模型，进而给出了第一级处理中的检测级、跟踪级和属性级的结构模型。 (2)对线性和非线性离散卡尔曼滤波进行了研究，在此基础上针对线性离散多传感器信息融合系统中的状态估计问题，分别研究多层系统中单传感器的状态估计、集中式多传感器状态估计以及分布式多传感器系统的状态估计。 (3)基于卡尔曼滤波算法，在各传感器估计误差相互独立和不独立两种假设的前提下，给出了卡尔曼滤波加权融合算法，在此基础上，提出了修正的track-to-track航迹融合算法，并将改进后的算法应用于鱼雷线导加末自导系统的仿真研究。仿真结果表明，该算法提高了估计精度。 (4) 将多传感器量测噪声自动加权卡尔曼滤波应用在线导鱼雷制导中，融合来自制导站声纳测量到的目标信息和来自鱼雷惯性测量组件的鱼雷位置和姿态等信息进行了仿真研究。仿真中在线加入量测噪声的指数加权时变噪声估计器，使滤波器的稳定性和自适应能力显著增强。更多还原

【Abstract】 Multi-sensor information fusion means processing data from more than one sensor to get new efficient information, which is not available from any single sensor. Multi-sensor information fusion has been developed into an active research field. So it is significant to further get access to this field and apply this technology to the guidance of underwater weapon such as torpedo.In this dissertation, we mainly research into multi-sensor information fusion technology and apply it to line-guidance torpedo, and multi-sensor state estimation and track fusion algorithm are discussed. Research works is shown as follows:(1) Basic model of information fusion is studied in detail, and information fusion is classified in terms of function and structure. Four-level data processing function model is given.(2) On the basis of studying linear and nonlinear decentralized kalman filter algorithm, aiming at state estimation of linear decentralized multi-sensor data fusion system, state estimations of single sensor and centralized multi-sensor and decentralized multi-sensor system of multi-level system are studied respectively.(3) On the basis of kalman filer algorithm and on the premise of two assumptions, namely, state estimation error of each sensor is dependent or independent, kalman filter weight fusion algorithm and modified track-to-track fusion algorithm are proposed, then improved algorithm is applied to the simulation of line-guidance and terminal-self guidance of torpedo.(4) Applying measurement noise automatic weighting kalman filter to the guidance of line-guidance torpedo, we compromised target information and torpedo information, which in turn come from sonar of guidance station and inertial measurement units of torpedo. In the simulation, we added index weighting noise estimation of measurement noise to improve stability and adaptability of filter.更多还原