基于大数据驱动的船舶航行轨迹异常检测研究

Research on abnormal detection of ship navigation path based on big data drive

作者单位:

1. 广西船联网工程技术研究中心，广西 南宁 530007;2. 广西感知物联网生产力促进中心，广西 南宁 5300071. Guangxi Ship Networking Engineering Technology Research Center, Nanning 530007, China;2. Guangxi Perceived Internet of Things Productivity Promotion Center, Nanning 530007, China

大数据驱动;船舶航行轨迹;异常检测;Spark Streaming框架;聚类方法big data-driven; ship navigation path; abnormal detection; SSF; clustering method

船舶安全航行是航海领域重点关注的问题之一，为此研究基于大数据驱动的船舶航行轨迹异常检测方法。该方法利用不同类型传感器获取船舶航行大数据，然后使用船舶观测大数据相似度方程计算船舶航行大数据之间的相似度，得到来自同一船舶的航行大数据；再利用大数据驱动技术中的聚类方法建立船舶正常轨迹模型，获取船舶航行正常轨迹；依据船舶航行正常轨迹，利用大数据驱动技术内的Spark Streaming数据实时计算框架，通过计算船舶航行轨迹点与实际轨迹采样点之间的距离、航向角等，得到船舶航行轨迹异常检测结果。实验结果表明，该方法获取船舶航行实际轨迹精度较高，可有效检测船舶航行轨迹异常，具备较好的应用效果。The safe navigation of ships is one of the key issues in the navigation field. Therefore, a method of ship navigation path anomaly detection based on big data-driven is studied. This method uses different types of sensors to obtain ship navigation big data, and then uses the ship observation big data similarity equation to calculate the similarity between ship navigation big data, and obtains the navigation big data from the same ship. Then, the cluster method in big data-driven technology is used to establish the normal trajectory model of the ship and obtain the normal navigation trajectory of the ship; According to the normal ship navigation track, using the real-time computing framework of Spark Streaming data in big data-driven technology, the abnormal ship navigation track detection results are obtained by calculating the distance and heading angle between the ship navigation track point and the actual track sampling point. The experimental results show that this method has high accuracy in obtaining the actual ship navigation trajectory, and can effectively detect the ship navigation trajectory anomalies, and has good application effect.