基于CALS算法的船舶航迹和航行行为特征研究  

Research of Ship Trajectory and Navigation Behavior Characteristics Based on CALS Algorithm

作　　者：杨悦 刘天一 张美丽[1] YANG Yue;LIU Tian-yi;ZHANG Mei-li(Dalian Naval Academy,Dalian 116018,China)

机构地区：[1]海军大连舰艇学院,辽宁大连116018

出　　处：《数学的实践与认识》2025年第1期127-136,共10页Mathematics in Practice and Theory

基　　金：海军大连舰艇学院科研基金(DJYKT2021-D18)。

摘　　要：为解决船舶异常航行轨迹检测,为安全航行和军事任务决策提供信息支撑,提出一种基于深度学习和滑动窗口的可疑船只预警算法(CALS).基于一年内中国海区AIS数据,运用卷积神经网络(CNN)提取船舶运动参数的多维特征,将AIS数据以时序形式输入LSTM网络,引入注意力机制赋予隐含层权重,以区分不同时序点对预测舰船运动参数的影响程度,优化预测模型,最后利用滑动窗口算法进行流数据异常检测,实现船舶异常航迹预警.选取2021年某日中国海区共8576条船舶AIS报文信息进行测试,结果表明引入注意力机制的神经网络预测模型比LSTM和CNN-LSTM预测准确率更高,预警准确率稳定在90%左右.To address the challenge of detecting abnormal navigation trajectories of vessels and to provide informational support for safe navigation and military decision-making,this paper proposes a suspicious vessel early warning algorithm CALS that leverages deep learning and sliding window techniques.Utilizing one year's worth of AIS data from Chinese maritime areas,Convolutional Neural Networks(CNN)is employed to extract multidimensional features related to vessel movement parameters.The AIS data is formatted temporally and input into a Long Short-Term Memory(LSTM)network,where an attention mechanism is introduced to assign weights to the hidden layers.This approach enables us to differentiate the impact levels of various time points on predicting vessel movement parameters,thereby optimizing the prediction model.Ultimately,a sliding window algorithm is implement for anomaly detection in streaming data,achieving timely warnings for abnormal ship trajectories.In the experiments,8576 AIS messages are selected from vessels operating in Chinese maritime areas on a specific day in 2021 for testing purposes.The results indicate that the neural network prediction model incorporating the attention mechanism significantly outperforms both LSTM and CNN-LSTM models regarding prediction accuracy,with an early warning accuracy consistently around 90%.

关 键 词：CALS算法 船舶航迹 航行行为 AIS系统 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]