摘要
海底地形匹配导航(terrain aided navigation, TAN)无需卫星或水声等精确定位手段辅助,可实现无人水下航行器(unmanned underwater vehicle, UUV)水下长期、隐蔽、精确导航。但TAN系统易在地形特征不明显区域出现匹配失效问题,通过地形特征量化分析获取地形适配性从而进行路径规划,是保障TAN系统可靠性的前提。本文针对现有地形适配性统计学变量耦合关系复杂、综合分析权重选取困难等问题,提出地形适配性的神经网络分析方法。基于历史海测子地图,通过蒙特卡罗迭代生成训练集;构建面向TAN系统地形匹配的PointNetKL网络模型,实现子地图适配性的准确、高效判断。最终,结合模板匹配算法对判断结果进行评估。本算法有效性和计算效率已通过海试数据回放试验验证。
Terrain aided navigation(TAN) can yield long-term, covert and accurate underwater navigational results for unmanned underwater vehicles(UUVs) without the supplements of accurate positioning methods, such as satellite or underwater acoustic positioning. However, TAN system might fail catastrophically in areas with poor terrain features. Thus, the calculation of TAN-suitability for areas using quantitative analysis and the subsequent path planning, is the premise to ensure the reliability of TAN systems. The complex coupling relationship of TAN-suitability statistical variables makes it difficult to select the weight of each variable in the TAN-suitability comprehensive analysis. To solve this problem, this paper proposed a TAN-suitability analysis method based on neural network. Training set are constructed using Monte-Carlo iteration with real bathymetric data. PointNetKL network was constructed to calculate TAN-suitability for areas. Finally, the PointNetKL results are evaluated using the template matching algorithm. The proposed algorithm has been proved in playback experiment using sea-trial data.
作者
高靖萱
杜昊翰
张亚
GAO Jing-xuan;DU Hao-han;ZHANG Ya(School of Navigation and Naval Architecture,Dalian Ocean University,Dalian 116000,China;Science and Technology on Underwater Vehicle Laboratory,Harbin Engineering University,Harbin 150001,China;College of Shipbuilding Engineering,Harbin Engineering University,Harbin 150001,China)
出处
《舰船科学技术》
北大核心
2022年第24期114-118,共5页
Ship Science and Technology
基金
中央高校基本科研业务费资助项目(3072022TS0102)。
