【作者】 官兴华；

【导师】 李建龙；

【作者基本信息】 浙江大学 ， 信息与通信工程， 2011， 硕士

【摘要】 自主水下航行器(Autonomous Underwater Vehicle, AUV)是目前海洋工程领域研究的热点,在军用和民用两方面都发挥着极其重要的作用。导航技术是实现AUV自主航行的关键,而AUV所具有的工作时间长、环境复杂、信息源少、隐蔽性要求高等特点,给稳定、精确的导航实现带来很大的挑战。Bayes滤波是一种可融合多种传感器观察数据的组合导航算法。由于利用概率表示海底声纳图像特征,因此对于缺少点、线等图像特征的非结构化海底地形以及较低质量的海底成像数据,该方法具有较好的宽容性,同时也具有广泛的适用性。本文基于AUV载声纳系统,研究以Bayes滤波为核心的同时定位和水底特征检测组合导航技术,侧重于把水底特征检测作为导航的关键信息。该方法的观察数据主要来源于四种途径,即通过GPS获取AUV下潜之前水面初始位置信息,通过声纳成像技术获取相邻脉冲间隔之间的水底地貌特征信息,通过多普勒计程仪获取AUV的速度信息,通过电子罗盘/智能罗经获取AUV的姿态信息,利用Bayes滤波对上述四类观察数据的融合达到高精度组合导航的目的。同时,考虑到观察数据可能存在非高斯特性,本文对基于粒子滤波(Particle Filter, PF)的AUV导航技术进行了研究。此外,还对多AUV协同导航技术进行了初步的研究。论文对上述方法开展了充分的理论推导、仿真分析,并通过实验室研制的AUV载声纳系统对相关算法开展了湖上试验验证,证明了算法的有效性。更多还原

【Abstract】 AUV (Autonomous Underwater Vehicle) is a hot research point of ocean engineering and plays an extremely important role in both military and civilian fields. Navigation technology is the key of AUV autonomous navigation, while the characteristics of AUV such as the long working time, complex environments, fewer sources of information, high confidentiality requirement bring a great challenge to the achievement of stable and accurate navigation.Bayes filtering is a integrated navigation algorithm to fuse multiple sensor observations. The probability description of water bottom image features is robust, in the cases of the unstructured seabed topography with the lack of points, lines and lower-quality imaging data. In this paper, based on sonar system of AUV, we discuss the integrated navigation method of simultaneous localization and bottom feature detection using Bayes filtering as a core technology, with focusing on the detection of underwater features as key navigation information.The method has four main ways of getting the observation data, namely through the GPS to get the initial location information before AUV diving, through the sonar imaging technology to obtain the seabed topography feature information between adjacent pulse intervals, through the doppler velocity log (DVL) to obtain the AUV speed information, through the electronic compass and intelligent gro compass to obtain the AUV attitude information. We employ Bayes filtering to integrate these four types of observation data to achieve the purpose of high-precision navigation.At the same time, in terms of the non-Gaussian characteristics of the observed data, we investigate the AUV navigation technology based on the particle filter (PF). In addition, navigation technology for multiple AUVs is also discussed.In this paper, we perform full theoretical analysis, simulation and at lake experiment through AUV loading sonar system for the above approaches, which prove the effectiveness of the algorithms.更多还原