【作者】 陈小龙；

【导师】 庞永杰；

【作者基本信息】 哈尔滨工程大学 ， 船舶与海洋结构物设计制造， 2013， 博士

【摘要】 水下地形匹配导航是利用地形匹配技术来实现水下精确定位的自主、全天候的导航方式，对于水下航行器长时间水下航行时的精确定位具有重要的作用。本文以AUV长航时的水下精确导航定位为背景，利用多波束测深仪作为水下地形测量手段，结合真实海上试验数据对水下地形匹配导航定位方法进行了深入的研究。论文的主要成果和工作有：（1）研究了水下地形匹配导航的原理、模型以及适用的方法。从理论上探讨了水下地形匹配导航系统的组成，将其分为基本导航单元、水深测量单元以及地形匹配单元三个部分，并以此分析了水下地形匹配导航的原理。分析了水下地形匹配导航的模型，包括状态空间模型以及数字地形模型，由此导出了影响水下地形匹配导航定位性能的因素。在比较了陆地地形匹配导航与水下地形匹配导航的区别基础上，探讨了利用多波束测深进行水下地形匹配导航的优势，并从理论上分析了适用于多波束测深进行水下地形匹配导航的方法。（2）研究了局部高精度水下数字地形建模方法。详细介绍了多波束测深系统的组成以及原理，给出了本文中使用到的多波束测深系统的具体参数。在此基础上，分析了多波束测深数据后处理原理，并给出了多波束测深数据后处理的流程。针对多波束测深数据的滤波问题，提出了基于动态聚类的单声脉冲(ping)测深数据滤波方法。对滤波后的多波束测深数据进行了空间归位和网格化处理，建立了基于多波束测深的局部高精度数字地形图。（3）研究了实时水下地形特征的提取和匹配单元的构建方法。建立了单波束和多波束测深下的实时海底地形匹配模型，构建了一种多测点下的自适应地形特征的波束选择模式。为了进行地形的插值重构，在分析地形插值模型的基础上，利用分形原理对线性插值进行补偿，提出了一种基于分形补偿的双线性插值方法，对地形特征具有良好的适应能力。为了提高地形的插值精度和效率，提出了一种基于索引的快速插值策略。（4）研究了基于相关性的地形匹配定位方法。在相关性原理的基础上，利用多波束测深下的面地形特征，提出了一种基于极大似然估计的水下地形匹配定位方法。建立了极大似然估计的水下地形匹配导航模型。针对地形匹配中的伪波峰问题，提出了基于费希尔判据的去伪方法，对地形匹配中出现的伪波峰进行有效的剔除。利用真实多波束测深数据进行了极大似然估计下的定位试验。（5）研究了基于信息融合的AUV水下地形匹配导航。从信息融合的观点出发，将地形匹配定位技术与基本导航系统相结合，建立了地形匹配导航的关联模型，利用最近邻法建立了地形匹配定位的有效性检测流程，提出了一种基于关联可靠性的连续导航定位方式。建立了贝叶斯估计下的水下地形匹配导航模型，利用粒子滤波方法来对贝叶斯估计问题进行近似最优解算，提出了一种基于改进的粒子滤波的水下地形匹配航迹跟踪方法。探讨了影响水下地形匹配导航误差的因素，最后，从模块化设计思想出发，建立了AUV水下地形匹配导航系统结构，针对AUV水下导航，构建了基于双重模式的水下地形匹配导航策略，利用真实多波束测深试验数据进行了仿真研究。本文深入研究了AUV水下地形匹配导航的原理、方法以及实施步骤，并进行了真实多波束测深数据的回放式仿真试验。结果表明：本文提出的方法具有较高的精度，对地形特征具有较好的适应能力，可以利用来进行长时间下的水下导航定位修正，具有较高的实际应用价值。更多还原

【Abstract】 Underwater terrain matching navigation (UTMN) is an independent and all-weathernavigation mode using terrain matching technology to gain precise positioning of underwater,which plays an important role on the navigation of underwater vehicle for long period of timesubmerged. For the precise navigational positioning of Automous underwater vehicle (AUV),with multibeam echo sounder (MBES) as the measuring means, underwater terrain matchingnavigational positioning method is thoroughly investigated with real sea test data. The mainwork is as follows:(1) The principle, model and suitable method of UTMN are investigated. Thecomposition of UTMN system is probed into in theory, which is divided into basic navigationunit, depth measuring unit and terrain matching unit. According to which, the principle ofUTMN is analysed. The model of UTMN is analysed, including state space model and digitalterrain model, follow which the factors influence the performance of terrain matchingnavigation are educed. The difference between land and underwater terrain navigation iscompared, from which the advantage of using MBES for UTMN is discussed, and the methodsuitable for UTMN with MBES are analysed in theory.(2) The method for modeling of local high-accuracy underwater digital terrain isinvestigated. The composition and principle of MBES measuring system is introduced andthe parameter of the MBES used in this paper is presentd. The principle of post-processing ofMBES data is analysed and the flow of it is presented. Aim at the filtering of MBES data, asingle ping filtering based on dynamic clustering is proposed. After space homing and gridingare applied to the filtered data, the local high-accuracy underwater digital terrain model(UDTM) is made.(3) The real-time characteristics extraction and construction of matching unit ofunderwater are investigated. Real-time underwater terrain matching model of single beamand multibeam are established, and a beam selection mode for multi measuring beams basedon adaptive of terrain characteristic is proposed. For the interpolation of terrain, on theanalysis of terrain interpolation model, a bilinear method based on fractal compensation isproposed, which has good adaptability for different terrain characteristic. For the improvement of interpolation accuracy and efficiency, a method for storage of large terrainbased on mixing resolution is proposed, form which a fast interpolation strategy based onindex is proposed. The interpolation time decreases rapidly.(4) The terrain matching positioning method based on correlation is investigated. Basedon the correlation principle and the surface terrain characteristic get by multibeam echosounder, a method based on maximal likelihood estimation (MLE) is proposed for underwaterterrain matching positioning. The UTMN model for MLE is presented, for the false peaks interrain matching, a method based on Fischer criterion is proposed for false peak elimination,which is effective for the false peak elimination appeard in the terrain matching. Positioningtest based on MLE with real MBES data is carried out.(5) The UTMN of AUV bsed on information fusion is investigated. From the perspectiveof information fusion, the terrain matching technology is integrated with the basic navigationsystem, the association model of terrain matching navigation and the validity flow of terrainmatching positioning with nearest neighbour (NN) are presented, from which a continuousnavigational positioning mode base on association reliability is proposed. The UTMN modelbased on Bayes estimation is presented and particle filter is used for near optimal resolvationof Bayes estimation problem, and an improved particle filter is proposed for path tracking ofunderwaer terrain matching. The factors that influence the UTMN accuracy are analysed. Atlast, the UTMN system construction of AUV is presented from the modularization view. Forthe UTMN of AUV, a navigation strategy with two models is proposed and simulations arecarried out with MBES data.This paper investigated the principle, method and implementation steps of UTMN ofAUV, and the play back mode simulation is carried out with real MBES data. The resultsshow that the method proposed in this paper has good accuracy and adaptability for terraincharacteristic, which can be used for position correction of underwater navigation for longperiod of time submerged. The proposed method is proved effective, reliable and practical.更多还原