GNSS拒止环境下的全源全维高精度导航定位机制与关键技术研究

Focusing on the urgent need for precise navigation and positioning capabilities of all types of military & civilian vehicles under GNSS denial environment and the shortage of current integrated navigation system , the project analyzes essential features of navigation needs of the all-source , full-dimensional , plug and play and configuration on demand. It proposes the mechanisms which uses uncertainty reasoning of navigation parametric and correlation of decision-making in the navigation system. Meanwhile, it researches the core issues of all-source full-dimensional navigation which combines topological information fusion, use of the sparsity information and probabilities correlation. Furthermore, it establishes theoretical models and architectures of innovational all-source full-dimensional navigation system. And it settles topological probability integration method and maximum entropy compensation mechanisms for transferring and sharing all-source full-dimensional navigation information, asynchronous solve & update mechanism, and optimal posterior probability measure mechanism of posterior information. It also reveals inherent statistical correlation rule between the state parameters( position, velocity , attitude ,etc.) and the measurements of various types navigation sensor; obtains unsteady coupling mechanism between navigation & position solutions and sensing measurements; creates new navigation theories and methods; achieves design of all-source full-dimensional navigation receiver which is full compatibility , interoperability and mutual replace for all-source full-dimensional navigation information . Expected results which will further promote the theory and technologies of navigation to maturity, the theoretical and practical value are very significant.

针对军民各型载体对GNSS信号拒止环境下精确导航定位能力的迫切需求，深入分析全源、全维、即插即用、随遇组合等导航需求的本质特征，提出在导航系统中采用导航参量非确定性推理与决策的相关性机制，通过对全源全维导航信息的拓扑融合机制、稀疏性利用和概率关联等核心问题的研究，建立起即插即用、随遇配置的新型全源全维导航理论模型和体系架构，突破全源全维导航信息概率传递与共享的拓扑融合方法及最大熵补偿机制、信息异步解算和更新机制以及信息后验概率最优后验测度机制等核心理论，揭示出全源全维导航理论架构下载体位置、速度、姿态等状态参量与各型导航传感单元量测量之间固有统计相关性的本质规律，获得导航传感单元量测信息与导航解算信息间的非定常耦合机制，创造出导航定位新理论和方法，完成全兼容、互操作与互替换的全源全维导航接收机设计。 预期成果将进一步促进导航理论和导航对抗技术的完善与成熟，具有十分重要的理论意义和实用价值。

以大疆M100为载荷平台构架了构建了融合声(超声波)、光(视觉、LiDAR、)电(Wi-Fi、GNSS)、磁(地磁)、力(IMU)异构传感器的全维全源导航理论与技术实验验证系统，在理论研究方面通过对全源全维导航信息的拓扑融合机制、稀疏性利用和概率关联等核心问题的研究，提出了一种高斯模型下基于因子图框架的子系统观测量均值向量和协方差矩阵的动态估计方法，用来实现微智能终端的室内外无缝导航，并验证了基于WLAN多载波相位测距的精密定位、自适应行人航迹推算和地磁辅助的姿态解算、基于单目视觉/WSN融合的室内精密定位与构图以及多通道-单维卷积的全卷积神经网络故障检测方法等新型技术路线，这些研究成果与M100硬件平台共同构建起了即插即用、随遇配置的新型全源全维导航理论模型和体系架构。此外，为了更好实现全源导航的信息融合，还研究了类脑智能自主导航模型，并对经典仿生SLAM进行了改进。 .研究过程中还突破了全源全维导航信息概率传递与共享的拓扑融合方法及最大熵补偿机制、信息异步解算和更新机制以及信息后验概率最优后验测度机制以及仿生智能信息融合等核心理论，提出的因子图估计方法，利用因子图最优化过程中每个迭代周期内的观测量残差，实时地更新各个子系统观测量的均值向量和协方差矩阵的最大后验估计，从而得到更加准确的导航状态估计值，仿真测试与实验测试结果均表明，与已有的因子图算法相比，本项目所提出的基于迭代最大后验估计的因子图算法能够有效提高子系统观测状态变化时的多源组合导航估计精度。在仿生智能研究方面提出的模板识别阈值动态调整的局部视图细胞生成方法，提升了视觉模板的正确匹配能力，改善了视觉感知仿生SLAM经验图的生成质量，在算法运行时间方面也表现良好。.研究中，共发表论文25篇，其中EI源刊12篇、SCI源刊2篇、国际会议4篇(SCI索引2篇，EI索引1篇)、中文核心期刊5篇;申请专利2项；出版著作2部。