为自动提取数据中的复杂特征，提升状态检验的鲁棒性，设计深度学习算法下舰船导航设备状态检验方法。通过建立包含陀螺仪漂移误差、加速度计随机误差及GPS接收机定位误差的多源误差模型，系统量化姿态角、速度及位置等关键状态指标；在深度信念网络内，输入关键指标数据，自动提取数据中反映舰船导航设备状态的复杂特征；以极限学习机为深度信念网络的回归层，结合提取的复杂特征，输出导航设备状态检验结果。实验证明，该方法可有效自动提取检查导航设备状态的复杂特征，完成设备状态检验；在不同负载情况下，导航设备状态检验的决定系数达0.93以上，即状态检验的鲁棒性较优。
To automatically extract complex features from data and improve the robustness of state checking, a deep learning algorithm is designed for ship navigation equipment state checking method. By establishing a multi-source error model that includes gyroscope drift error, accelerometer random error, and GPS receiver positioning error, the system quantified key state indicators such as attitude angle, velocity, and position; In the deep belief network, key indicator data is input to automatically extract complex features reflecting the status of ship navigation equipment from the data; Using an extreme learning machine as the regression layer of a deep belief network, combined with the extracted complex features, output the navigation device state verification results. Experimental results have shown that this method can effectively automatically extract complex features for checking the status of navigation devices and complete device status verification; Under different load conditions, the determination coefficient of navigation device state inspection is above 0.93, indicating that the robustness of state inspection is relatively good.
2025,47(14): 164-168 收稿日期：2025-2-24
DOI：10.3404/j.issn.1672-7649.2025.14.025
分类号：TP391
基金项目：山西省基础研究计划项目（202303021221014）
作者简介：李永俊(1990-),男,高级工程师,研究方向为人工智能、信息安全及算法优化
参考文献：
[1] 祁帅, 贾继超, 寇得民, 等. 基于Allan方差的改进自适应滤波SINS/GNSS导航算法[J]. 传感技术学报, 2024, 37(5): 818-824.
QI S, JIA J C, KOU D M, et al. Improved adaptive filtering SINS/GNSS navigation algorithm based on allan variance[J]. Journal of Sensing Technology, 2024, 37 (5): 818-824
[2] 张闯, 贾春林, 郭沐壮. 基于因子图的智能船舶INS/GPS组合导航算法[J]. 上海海事大学学报, 2024, 45(4): 8-16.
ZHANG C, JIA C L, GUO M Z. Intelligent ship INS/GPS integrated navigation algorithm based on factor graph[J]. Journal of Shanghai Maritime University, 2024, 45 (4): 8-16
[3] 高小鹏, 柯欢欢, 陈浩然. 基于组合导航和互补滤波器的舰船升沉测量方法研究[J]. 导航与控制, 2024, 23(2): 82-87.
GAO X P, KE H H, CHEN H R. Research on ship heave measurement method based on combined navigation and complementary filters[J]. Navigation and Control, 2024, 23 (2): 82-87
[4] 周芸, 周凌柯, 李胜, 等. SINS/GNSS/OD组合导航系统的容错机制研究[J]. 电子测量与仪器学报, 2024, 38(11): 25-32.
ZHOU Y, ZHOU L K, LI S, et al. Research on fault tolerance mechanism of SINS/GNSS/OD integrated navigation system[J]. Journal of Electronic Measurement and Instrumentation, 2024, 38 (11): 25-32
[5] 刘子龙, 楼家祺. 基于正交性原理的组合导航系统故障检测[J]. 控制工程, 2024, 31(10): 1841-1848.
LIU Z L, LOU J Q. Fault detection of integrated navigation system based on orthogonality principle[J]. Control Engineering, 2024, 31 (10): 1841-1848
[6] 赵宾, 曾庆化, 刘建业, 等. 基于平滑有界层的故障检测及抗差自适应滤波算法[J]. 中国惯性技术学报, 2023, 31(3): 245-253.
ZHAO B, ZENG Q H, LIU J Y, et al. Fault detection and robust adaptive filtering algorithm based on smooth bounded layers[J]. Chinese Journal of Inertial Technology, 2023, 31 (3): 245-253
