The invention provides a ship track prediction method based on improved dogvessel multi-output support vectors. The method specifically comprises the following steps: screening and preprocessing AIS track data provided by a navigation center of an American coast police team; carrying out dogvessel parameter optimization on hyper-parameters in the model; constructing a final multivariate support vector model, and performing evaluation; and other optimization methods are selected to be compared with various support vector machines. Model training and verification are carried out on AIS data collected by a navigation center of the American coast police team, and a result shows that the method has good prediction performance under the condition of strong coupling of longitude and latitude and still has excellent performance under the condition of weak coupling of the longitude and latitude of the model; an SSA optimization method is improved, appropriate parameters are found through the algorithm, interference of subjective human factors is avoided, and the algorithm has few control parameters and is easy to implement; the characteristics of an adaptive weight algorithm and an outlier image algorithm are introduced for improvement.
本发明提供了一种基于改进樽海鞘多输出支持向量的船舶航迹预测方法。具体为对美国海岸警卫队导航中心提供的AIS航迹数据进行筛选和预处理;对模型中的超参数进行樽海鞘参数寻优;构建最终多元支持向量模型,并进行评估;选取其他的寻优方法与各种支持向量机与之做对比。本发明在美国海岸警卫队导航中心采集到的AIS数据进行模型训练及验证,结果显示本方法在经纬度强耦合的情况下的预测表现良好,并且在模型经纬度弱耦合的情况下的表现仍然优秀;并改进了樽海鞘群(SSA)寻优方法,由算法寻找合适的参数,避免主观人为因素的干扰,算法控制参数少且易于实现;引入了自适应权重算法与离群象算法的特点进行改进。
Ship track prediction method based on improved dogvessel multi-output support vector
一种基于改进樽海鞘多输出支持向量的船舶航迹预测方法
12.05.2023
Patent
Elektronische Ressource
Chinesisch
IPC: | G06N COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS , Rechnersysteme, basierend auf spezifischen Rechenmodellen / G08G Anlagen zur Steuerung, Regelung oder Überwachung des Verkehrs , TRAFFIC CONTROL SYSTEMS / G06F ELECTRIC DIGITAL DATA PROCESSING , Elektrische digitale Datenverarbeitung |
Research on ship track prediction method based on improved PSO-BP algorithm
British Library Conference Proceedings | 2023
|