A Recognition Method for Ship Motion Pattern Based on Nine-axis IMU
-
摘要: 运动模式识别是实现船舶智能航行的重要研究方向。针对现有方法中数据更新速度慢、受环境约束大的问题，研究了1种基于九轴惯性测量单元（inertial measurement unit，IMU）的船舶运动模式识别方法。分析了当前船舶运动感知技术的不足，提出利用加速度计、陀螺仪和磁力计构成的九轴IMU来识别船舶运动参数。为处理包含多种运动模式的长时间连续信号，提出了1种基于隐马尔可夫模型的数据分割算法，并采用期望最大化算法估算模型参数，实现按运动模式分割信号，提取单一稳态模式信号。分析分割后的运动模式信号，提取能够表征船舶运动模式的时域特征。为提高识别精度，设计了1种基于二叉树结构的支持向量机（support vector machine，SVM）算法，利用最大割问题构建二叉树结构，在决策节点使用SVM分类器，并通过粒子群优化算法优化模型参数。实验基于实船采集的运动数据进行验证，结果表明：所提识别算法只需训练5个SVM子分类器，能够对6种船舶运动模式进行有效识别，平均识别精度达到96.498%。相比传统的一对一和一对多SVM多分类方法，提出的方法平均识别精度分别提高了13.835%和21.305%，且所需训练的子分类器数量更少，验证了方法的优越性与高效性。Abstract: Motion pattern recognition is an important issue for achieving intelligent navigation of ships. To address the limitations of existing methods, including slow data update rates and strong environmental constraints, a ship motion pattern recognition method based on nine-axisinertial measurement unit (IMU) is proposed. The shortcomings of current ship motion sensing technologies are analyzed, and a nine-axis IMU consisting of an accelerometer, gyroscope, and magnetometer is utilized to identify ship motion parameters. To process long-duration continuous signals encompassing multiple motion patterns, a data segmentation algorithm based on hidden Markov model is developed. The expectation maximization algorithm is employed to estimate model parameters, enabling signal segmentation according to motion patterns and the extraction of single steady-state motion signals. The time-domain features that characterize ship motion patterns are then extracted from the segmented signals. To improve recognition accuracy, a support vector machine (SVM) algorithm based on a binomial tree structure is designed. The binary tree structure is constructed using the maximum cut problem, with SVM classifiers employed at decision nodes. The particle swarm algorithm is applied to optimize the model parameters. Experiments conducted using ship motion data collected from real ships validate the proposed method. Results show that the proposed recognition algorithm requires training only five SVM sub-classifiers for the recognition of six ship motion patterns, achieving an average recognition accuracy of 96.498%. Compared to traditional one-to-one and one-to-rest SVM multi-classification methods, the proposed method improves average recognition accuracy by 13.835% and 21.305%, respectively, while requiring fewer sub-classifiers for training. These findings demonstrate the superiority and efficiency of the proposed approach.
-
图 4 x、y、z轴方向以及合加速度信号分割结果
Figure 4. Segmentation results of x、y、z axis signals and combined acceleration signals
图 5 基于二叉树的支持向量机识别算法流程图
Figure 5. Flow chart of support vector machine recognition algorithm based on binary tree
图 7 基于二叉树支持向量机算法的识别结果
Figure 7. Graph of recognition results based on binary tree support vector machine algorithm
表 1 船舶各运动模式明显特征变化统计
Table 1. Statistics on the change of distinctive features of each movement mode of the ship
运动模式 加速度 角速度 姿态 ax ay az wx wy wz pitch roll yaw 静止 加速 √ √ 减速 √ √ 匀速 √ √ 左转 √ √ √ √ 右转 √ √ √ √ 
下载: 导出CSV
表 2 船舶运动参数特征分析
Table 2. Characterization of ship motion parameters
运动模式 运动参数特征分析 静止 ax，ay，roll，pit ch均值基本为0，航向角基本保持不变 加速 ay均值大于0，且roll均值在[3° 5°] 之间 减速 ay均值小于0，且roll均值在[1° 2°] 之间 匀速 ay均值基本为0，且roll均值在[2° 3°] 之间 左转向 ax先增大后减小且大于0，wz小于yaw，yaw变小，pitch先减小后增大 右转向 ax先减小后增大且小于0，wz大于2°/s，yaw变大，pitch先增大后减小
下载: 导出CSV
表 3 船舶转向运动的组合方式
Table 3. Combination of lateral and longitudinal modes of ship steering motion
运动模式 纵向 横向 加速 匀速 减速 静止 左转向 右转向 船舶左转向 √ -2°/s √ 船舶左转向 √ -2°/s √ 船舶右转向 √ √ 船舶右转向 √ √
下载: 导出CSV
表 4 船舶运动特征
Table 4. Ship motion feature
特征 提取对象 均值 ax，ay，wz，roll 方差 ax，ay 极大值极小值 ax，ay，wz 四分位间距 ax，ay 变化差值 ax，ay，wz，yaw
下载: 导出CSV
表 5 第1组识别结果
Table 5. Identification results of the first group
运动模式 样本数量/个 本文识别算法 1-V-1 1-V-R 误判数/个 识别率/% 误判数/个 识别率/% 误判数/个 识别率/% 静止 539 0 100 0 100 0 100 匀速 183 0 100 2 98.90 14 92.34 加速 116 0 100 0 100 18 84.48 左转向 106 12 88.67 83 21.69 40 62.26 右转向 102 11 89.52 30 70.58 77 24.50 减速 70 0 100 5 92.85 0 100
下载: 导出CSV
表 6 第2组识别结果
Table 6. Identification results of the second group
运动模式 样本数量/个 本文识别算法 1-V-1 1-V-R 误判数/个 识别率/% 误判数/个 识别率/% 误判数/个 识别率/% 静止 287 0 100 0 100 2 99.30 匀速 294 2 99.31 16 94.55 64 78.23 加速 302 0 100 0 100 70 76.82 左转向 66 3 95.45 20 69.69 43 34.84 右转向 146 19 86.99 51 65.06 4 97.72 减速 103 2 98.05 22 78.64 50 51.45
下载: 导出CSV
表 7 分类算法复杂度比较
Table 7. Complexity comparison of classification algorithms
多分类算法 子分类器训练个数 子分类器测试个数 本文算法 5 3 1-V-1 15 15 1-V-R 6 6
下载: 导出CSV
-
[1] CHEN Z, XUE J, WU C, et al. Classification of vessel motion pattern in inland waterways based on automatic identification system[J]. Ocean Engineering, 2018, 161: 69-76. doi: 10.1016/j.oceaneng.2018.04.072 [2] 曾毅, 刘成林, 谭铁牛. 类脑智能研究的回顾与展望[J]. 计算机学报, 2016, 39 (1): 212-222.ZENG Y, LIU C L, TANG T N. Retrospect and outlook of brain-inspired intelligence research[J]. Chinese Journal of Computers, 2016, 39 (1): 212-222. (in Chinese) [3] 马文耀, 吴兆麟, 李伟峰, 等. AIS数据中单船操纵模式辨识[J]. 中国航海, 2017, 40 (4): 51-55, 80.MA W Y, WU Z L, LI W F, et al. Recognition of individual ship maneuvering pattern from AIS data[J]. Navifation of China, 40 (4): 51-55, 88. (in Chinese) [4] PATROUMPAS K, ALEVIZOS E, ARTIKIS A, et al. Online event recognition from moving vessel trajectories[J]. Geoinformatica, 2017, 21 (2): 1-39. doi: 10.1007/s10707-016-0266-x [5] 陈影玉, 杨神化, 索永峰. 船舶行为异常检测研究进展[J]. 交通信息与安全, 2020, 38(5): 1-11. doi: 10.3963/j.jssn.1674-4861.2020.05.001CHEN Y Y, YANG S H, SUO Y F. Research progress of ship behavior anomaly detection[J]. Journal of Transport Information and Safety, 2020, 38 (5): 1-11. doi: 10.3963/j.jssn.1674-4861.2020.05.001 [6] 陈影玉, 杨神化, 索永峰. 船舶行为异常检测研究进展[J]. 交通信息与安全, 2020, 38 (5): 1-11. doi: 10.3963/j.jssn.1674-4861.2020.05.001CHEN Y Y, YANG S H, SUO Y F. Research progress of ship behavior anomaly detection. [J]. Journal of Transport Information and Safety, 2020, 38 (5): 1-11. (in Chinese) doi: 10.3963/j.jssn.1674-4861.2020.05.001 [7] 马文耀, 吴兆麟, 李伟峰. 船舶异常行为的一致性检测算法[J]. 交通运输工程学报, 2017, 17 (5): 149-158.MA W Y, WU Z L LI W F. Conformal detection algorithm of anomalous behaviors of vessel[J]. Journal of Traffic and Transportation Engineering, 2017, 17 (5): 149-158. (in Chinese) [8] 文元桥, 宋荣鑫, 黄亮, 等. 船舶行为的语义建模与表达[J]. 哈尔滨工业大学学报, 2021, 53 (8): 109-115.WEN Y Q, SONG R X, HUANG L, et al. Semantic modeling and expression for ship behaviors. [J]. Journal of Haebin Institute of Technology, 2021, 53 (8): 109-115. (in Chinese) [9] 黄亮, 刘益, 文元桥, 等. 内河渡船异常行为识别[J]. 大连海事大学学报, 2017, 43 (1): 8-13.HUANG L, LIU Y, WEN Y Q, et al. Abnormal behavior recognition of inland river ferryboat[J]. Journal of Dalian Maritime University, 2017, 43 (1): 8-13. (in Chinese) [10] YANG D, WU L, WANG S, et al. How big data enriches maritime research-a critical review of automatic identification system (AIS) data applications[J]. Transport Reviews, 2019, 39 (6): 755-773. doi: 10.1080/01441647.2019.1649315 [11] 汪荣琪. 内河航道船舶检测与跟踪算法研究[D]. 武汉: 武汉理工大学, 2017WANG R Q, Research on detection and tracking algorithm for ship in inland waterway[D]. WuHan: Wuhan University of Technology, 2017. (in Chinese) [12] 刘俊. 基于红外图像的内河运动船舶目标检测和跟踪技术研究[D]. 重庆: 重庆大学, 2008.LIU J. Moving ships detection and tracking from infrared image in inland waterway[D] ChongQing: Chongqing University, 2008. (in Chinese) [13] CAI Z, YU L, FAN Q, et al. Development of a vision-based ship behavior characteristics extraction approach and comparison with K-means clustering[C]. ISOPE International Ocean and Polar Engineering Conference, Shanghai: ISOPE, 2022. [14] GAO L, BOURKE A, NELSON J. Evaluation of accelerometer based multi-sensor versus single-sensor activity recognition systems[J]. Medical Engineering & Physics, 2014, 36 (6): 779-785. http://www.xueshufan.com/publication/2013582641 [15] 孟光磊, 张慧敏, 朴海音, 等. 自动化飞行训练评估中的战机机动动作识别[J]. 北京航空航天大学学报, 2020, 46 (7): 1267-1274.MENG G L, ZHANG H M, PIAO H Y, et al. Recognition of fighter maneuver in automatic flight training evaluation[J]. Journal of Beijing University of Aeronautics and Astronautics, 2020, 46 (7): 1267-1274. (in Chinese) [16] 徐西蒙, 杨任农, 于洋, 等. 基于运动分解和H-SVM的空战目标机动识别[J]. 控制与决策, 2020, 35 (5): 1265-1272.XU X M, YANG R N, YU Y, et al. Targetmaneuver recognitioninair combat based on motion decompositionand H-SVM[J]. Control and Decision, 2020, 35 (5): 1265-1272. (in Chinese) [17] HE Z, JIN L. Activity recognition from acceleration data using AR model representation and SVM[C]. 2008 International Conference on Machine Learning and Cybernetics, Kunming: IEEE, 2008. [18] GUERBAI Y, CHIBANI Y, HADJADJI B. The effective use of the one-class SVM classifier for handwritten signature verification based on writer-independent parameters[J]. Pattern Recognition, 2015, 48 (1): 103-113. http://www.onacademic.com/detail/journal_1000036617188210_317c.html [19] 白金辉. 面向能量回收的MIMU人体运动检测识别方法研究及系统实现[D]. 南京: 东南大学, 2015.BAI J H. Research and implementation of human motion detection and recognition system with MIMU for energy harvesting[D]. NanJing: Southeast University, 2015. (in Chinese) [20] MORILLO L M S, GONZALEZ-ABRIL L, RAMIREZ J A O, et al. Low energy physical activity recognition system on smartphones[J]. Sensors, 2015, 15 (3): 5163-5196. http://www.xueshufan.com/publication/2072737142 [21] FAN X, ZHAO S, SUN J. Nonparametric test for stratum effects in the cox model with interval-censored data[J]. Communicationsin Statistics-Theory and Methods, 2019, 48 (2): 300-310. http://www.xueshufan.com/publication/2774284810 [22] CHEN J Y, XIONG H, ZHENG H. Parameters optimization for SVM based on particle swarm algorithm[J]. Computer Science, 2018, 45 (6): 197-203. http://en.cnki.com.cn/Article_en/CJFDTotal-JSJA201806035.htm -
点击查看大图
计量
- 文章访问数: 37
- HTML全文浏览量: 9
- PDF下载量: 10
- 被引次数: 0