-
摘要: 通过挖掘海量AIS数据, 提出了一种新的航道水深信息获取方法, 即构建船舶安全航行水深参考图; 采用数据预处理的方法对历史与在线的AIS数据进行清洗和修补, 生成船舶运动轨迹; 选定船舶航行区域的时间与经纬度, 采用K-means聚类算法对船舶航行过程中的吃水数据进行聚类分析, 得到不同安全航行区域的船舶分类, 运用BP神经网络模型预测并补齐AIS数据中缺失的船舶最大吃水信息; 分割船舶历史轨迹, 当子轨迹的时间间隔在10~20min时, 采用Spline插值方法对船舶轨迹中的丢失数据进行插值; 采用凸包构建同类船舶的安全航行水深区域图, 将不同吃水类型船舶的安全航行水深区域图合并, 得到船舶安全航行水深合并图; 将不同吃水类型的船舶安全航行水深合并图与航道图叠加, 得到船舶安全航行水深参考图。试验结果表明: 当聚类算法参数设置为4时, 聚类后得到4类船舶, 对应的船舶最大吃水范围分别为0.1~4.8、4.8~6.6、6.6~10.0、10.0~13.0m, 对应的至少可通航船舶吃水分别为1.8、2.4、3.3、5.0m, 说明船舶最大吃水与至少可通航船舶吃水呈正相关关系; 构建的船舶安全航行水深参考图在电子航道图中覆盖了86%的航道, 并与航道图的深水部分重合率为80%, 因此, 构建的船舶安全航行水深参考图能反映航道水深的真实情况, 满足不同类别船舶的导航需求。Abstract: A new method of obtaining channel depth information was proposed by mining massive AIS data, which was used to construct ship safe navigation depth reference map. The historical and online AIS data were cleaned and mended by data preprocessing method, and ship motion trajectories were generated. The time, longitudes and latitudes of ship navigation areas were selected, and K-means clustering algorithm was used to cluster and analyze the draft data during ship navigation process, then the ship classifications in different safe navigation areas were obtained. The BP neural network model was applied to predict and complete the missing maximum ship draft information from the AIS data. The historical trajectory of the ship wassegmented, and when the time interval of the sub-trajectory was 10-20 min, the spline interpolation method was used to interpolate the missing data in the ship trajectories. The safe navigation depth area maps for similar types of ships were constructed by using convex hulls, and the maps with different draft types were combined to obtain a combined safe navigation depth map. The combined safe navigation depth map was superimposed on the channel chart to obtain a safe navigation depth reference map. Experimental result shows that when the clustering algorithm parameter is 4, four ship types are obtained by clustering. The corresponding maximum draft ranges of the ships are 0.1-4.8, 4.8-6.6, 6.6-10.0, and 10.0-13.0 m, and the corresponding least navigable ship drafts are 1.8, 2.4, 3.3, and 5.0 m, respectively. Thus, the maximum drafts of the ships are positively correlated with the least navigable drafts. The constructed ship safe navigation depth reference map covers 86% of the target channels in the electronic channel chart, and the overlap rate with the deep-water areas in the channel chart is 80%. Therefore, the constructed ship safe navigation depth reference map can reflect the true conditions of the channel depths, and meet the navigation needs of different types of ships.
-
表 1 船长、船宽和最大吃水的相关性
Table 1. Correlation between ship length, ship breadth and maximum draft
表 2 船舶类型和最大吃水的相关性
Table 2. Correlation between ship type and maximum draft
表 3 不同输入神经个数下平均绝对误差和均方误差Fig.3 Mean absolute errors and mean squared errors under different numbers of input neurons
表 4 部分预测值与真实值比较
Table 4. Comparison of partial predicted values and real values
表 5 船舶AIS动态信息
Table 5. AIS dynamic information of ship
表 6 插值点的船舶数据
Table 6. Ship data of interpolation points
表 7 安全航行船舶类别
Table 7. Category of safe navigation ship
-
[1] 林祎珣. 数据挖掘技术在海上交通特征分析中的应用研究[D]. 厦门: 集美大学, 2011.LIN Yi-xun. Application of data mining technology in analysis of marine traffic characteristics[D]. Xiamen: Jimei University, 2011. (in Chinese). [2] 郑滨, 陈锦标, 夏少生, 等. 基于数据挖掘的海上交通流数据特征分析[J]. 中国航海, 2009, 32 (1): 60-63, 90. https://www.cnki.com.cn/Article/CJFDTOTAL-ZGHH200901013.htmZHENG Bin, CHEN Jin-biao, XIA Shao-sheng, et al. Analysis of marine traffic flow characteristics based on data mining[J]. Navigation of China, 2009, 32 (1): 60-63, 90. (in Chinese). https://www.cnki.com.cn/Article/CJFDTOTAL-ZGHH200901013.htm [3] RHODES B J, BOMBERGER N A, ZANDIPOUR M. Probabilistic associative learning of ship motion patterns at multiple spatial scales for maritime situation awareness[C]∥IEEE. 10th International Conference on Information Fusion. New York: IEEE, 2007: 1-8. [4] 孟范立. 利用AIS数据挖掘建立船舶到达规律模型[J]. 舰船科学技术, 2016, 38 (5A): 28-30. https://www.cnki.com.cn/Article/CJFDTOTAL-JCKX201610011.htmMENG Fan-li. The use of AIS data mining to establish the ship arrives law model[J]. Ship Science and Technology, 2016, 38 (5A): 28-30. (in Chinese). https://www.cnki.com.cn/Article/CJFDTOTAL-JCKX201610011.htm [5] PALLOTTA G, VESPE M, BRYAN K. Vessel pattern knowledge discovery from AIS data: a framework for anomaly detection and route prediction[J]. Entropy, 2013, 15 (6): 2218-2245. [6] PAN Jia-cai, JIANG Qing-shan, HU Jin-xing, et al. An AISdata visualization model for assessing maritime traffic situation and its applications[J]. Procedia Engineering, 2012, 29: 365-369. doi: 10.1016/j.proeng.2011.12.724 [7] TU En-mei, ZHANG Guang-hao, RACHMAWATI L, et al. Exploiting AIS data for intelligent maritime navigation: a comprehensive survey from data to methodology[J]. IEEETransactions on Intelligent Transportation Systems, 2018, 19 (5): 1559-1582. doi: 10.1109/TITS.2017.2724551 [8] GOLDSWORTHY B. Spatial and temporal allocation of ship exhaust emissions in Australian coastal waters using AISdata: analysis and treatment of data gaps[J]. Atmospheric Environment, 2017, 163: 77-86. doi: 10.1016/j.atmosenv.2017.05.028 [9] HANSEN G M, JENSEN T K, LEHN-SCHI∅LER T, et al. Empirical ship domain based on AIS data[J]. Journal of Navigation, 2013, 66 (6): 931-940. doi: 10.1017/S0373463313000489 [10] SU Hai-bin, LIU Hong-xing, WU Qiu-sheng. Prediction of water depth from multispectral satellite imagery-the regression kriging alternative[J]. IEEE Geoscience and Remote Sensing Letters, 2015, 12 (12): 2511-2515. doi: 10.1109/LGRS.2015.2489678 [11] YOUNIS B A, SOUSA V, MEIRELES I. Prediction of the asymptotic water depth in rough compound channels[J]. Journal of Irrigation and Drainage Engineering, 2009, 135 (2): 231-234. doi: 10.1061/(ASCE)0733-9437(2009)135:2(231) [12] KISI O, SHIRI J. Wavelet and neuro-fuzzy conjunction model for predicting water table depth fluctuations[J]. Hydrology Research, 2012, 43 (3): 286-300. doi: 10.2166/nh.2012.104b [13] 侯朋, 许文海, 王俊生, 等. 基于双基地声纳的港口航道水深实时监测系统[J]. 仪器仪表学报, 2009, 30 (10): 2155-2160. https://www.cnki.com.cn/Article/CJFDTOTAL-YQXB200910025.htmHOU Peng, XU Wen-hai, WANG Jun-sheng, et al. Realtime monitoring system of water depth based on bistatic sonar in harbor channel[J]. Chinese Journal of Scientific Instrument, 2009, 30 (10): 2155-2160. (in Chinese). https://www.cnki.com.cn/Article/CJFDTOTAL-YQXB200910025.htm [14] 吴建华, 李红祥, 文元桥. 航道水深实时监控系统的原理及实现方法[J]. 船海工程, 2009, 38 (3): 153-156. https://www.cnki.com.cn/Article/CJFDTOTAL-WHZC200903042.htmWU Jian-hua, LI Hong-xiang, WEN Yuan-qiao. The principal and accomplishing method of real-time supervise system of channel depth[J]. Ship and Ocean Engineering, 2009, 38 (3): 153-156. (in Chinese). https://www.cnki.com.cn/Article/CJFDTOTAL-WHZC200903042.htm [15] 卫桂荣, 杨春. 船舶AIS数据错误检测方法[J]. 中国航海, 2016, 39 (4): 11-14. doi: 10.3969/j.issn.1000-4653.2016.04.003WEI Gui-rong, YANG Chun. Detection of AIS data error[J]. Navigation of China, 2016, 39 (4): 11-14. (in Chinese). doi: 10.3969/j.issn.1000-4653.2016.04.003 [16] 熊木地, 朱四印, 李禄, 等. 通航船舶吃水实时检测系统数据处理方法研究[J]. 仪器仪表学报, 2012, 33 (1): 173-180. https://www.cnki.com.cn/Article/CJFDTOTAL-YQXB201201026.htmXIONG Mu-di, ZHU Si-yin, LI Lu, et al. Research on data processing method of real-time detection system for dynamic ship draft[J]. Chinese Journal of Scientific Instrument, 2012, 33 (1): 173-180. (in Chinese). https://www.cnki.com.cn/Article/CJFDTOTAL-YQXB201201026.htm [17] ADLER J, PARMRYD I. Quantifying colocalization by correlation: the Pearson correlation coefficient is superior to the Mander's overlap coefficient[J]. Journal of the International Society for Advancement of Cytometry, 2010 (77A): 733-742. [18] SAAD Z S, GLEN D R, CHEN Gang, et al. A new method for improving functional-to-structural MRI alignment using local Pearson correlation[J]. Neuroimage, 2009, 44 (3): 839-848. doi: 10.1016/j.neuroimage.2008.09.037 [19] DAVIS A M, PEARSON R G, KNEIPP I J, et al. Spatiotemporal variability and environmental determinants of invertebrate assemblage structure in an Australian dry-tropical river[J]. Freshwater Science, 2015, 34 (2): 634-647. doi: 10.1086/681303 [20] 陈功平, 王红. 改进Pearson相关系数的个性化推荐算法[J]. 山东农业大学学报: 自然科学版, 2016, 47 (6): 940-944. doi: 10.3969/j.issn.1000-2324.2016.06.026CHEN Gong-ping, WANG Hong. A personalized recommendation algorithm on improving Pearson correlation coefficient[J]. Journal of Shandong Agricultural University: Natural Science Edition, 2016, 47 (6): 940-944. (in Chinese). doi: 10.3969/j.issn.1000-2324.2016.06.026 [21] JIA Wei-kuan, ZHAO De-an, SHEN Tian, et al. An optimized classification algorithm by BP neural network based on PLS and HCA[J]. Applied Intelligence, 2015, 43 (1): 176-191. doi: 10.1007/s10489-014-0618-x [22] CHENG Cheng, CHENG Xiao-sheng, DAI Ning, et al. Prediction of facial deformation after complete denture prosthesis using BP neural network[J]. Computers in Biology Medicine, 2015, 66: 103-112. doi: 10.1016/j.compbiomed.2015.08.018 [23] LI Jing-song, TIAN Yu, LIU Yan-feng, et al. Applying a BPneural network model to predict the length of hospital stay[J]. Lecture Notes in Computer Science, 2013, 7798: 18-29. [24] 邢晓敏, 刘洪涛, 丁震宇, 等. 基于BP神经网络的高精度基波频率检测方法研究[J]. 东北电力大学学报, 2015, 35 (1): 42-45. https://www.cnki.com.cn/Article/CJFDTOTAL-DBDL201501008.htmXING Xiao-min, LIU Hong-tao, DING Zhen-yu, et al. Research on high precision of fundamental frequency detection method based on BP neural network[J]. Journal of Northeast Dianli University, 2015, 35 (1): 42-45. (in Chinese). https://www.cnki.com.cn/Article/CJFDTOTAL-DBDL201501008.htm [25] FREEDEN W. On spherical spline interpolation and approximation[J]. Mathematical Methods in the Applied Sciences, 1981, 3 (4): 551-575. [26] YI Long-tao, LIU Zhi-guo, WANG Kai, et al. A new background subtraction method for energy dispersive X-ray fluorescence spectra using a cubic spline interpolation[J]. Nuclear Instruments and Methods in Physics Research, 2015, 775: 12-14. [27] MARIANI M C, BASU K. Spline interpolation techniques applied to the study of geophysical data[J]. Physica A: Statistical Mechanics and its Applications, 2015, 428: 68-79. [28] 余翔宇, 孙洪, 余志雄. 改进的二维点集凸包快速求取方法[J]. 武汉理工大学学报, 2005, 27 (10): 81-83, 92. https://www.cnki.com.cn/Article/CJFDTOTAL-WHGY200510023.htmYU Xiang-yu, SUN Hong, YU Zhi-xiong. An improved algorithm to determine the convex hull of 2-D points set[J]. Journal of Wuhan University of Technology, 2005, 27 (10): 81-83, 92. (in Chinese). https://www.cnki.com.cn/Article/CJFDTOTAL-WHGY200510023.htm