摘要
为研究基于船位数据的南极磷虾中层拖网船的作业状态特征提取方法,选取2021年福荣海、龙腾和龙发等3艘中国南极磷虾中层拖网船的VMS船位数据,采用航速、航向的阈值综合判别方法初步判断渔船作业状态(分为捕捞、航行和漂流),并与渔捞日志数据进行对比验证,确定渔船实际作业状态。依据船位数据中的航速、航向、时间区间、经纬度等特征,构建了基于深度学习网络的船位状态识别算法,并将数据集按照8∶2的比例分为训练集和验证集代入训练,提取渔船的作业天数、作业日期以及对应的作业位置信息,并与实际填写的渔捞日志数据进行对比验证。结果表明:提取的结果和实际记录的结果误差较小,其中,作业天数每月差值为0~2 d,平均准确率为93.88%;作业位置距离差值较小,距离差值在20 km以内的占比94%;与k近邻、逻辑回归、贝叶斯和决策树等传统机器学习算法相比,基于深度学习模型的南极磷虾拖网船作业状态特征提取正确率最高,达87.04%。研究结果对识别南极磷虾拖网渔船作业状态、捕捞行为和渔业管理具有一定的参考价值。
The operation characteristics of Antarctic krill midwater trawlers based on the location data were investigated.The VMS location data were collected from three Chinese Antarctic krill vessels:Furonghai,Longteng,and Longfa.The operational statuses were classified as fishing,sailing and drifting,and they were evaluated preliminarily according to a method combining the threshold of speed and heading data.The evaluation results were compared with fishing logbooks for verification,and finally,the actual status of the vessels was determined as a dataset.The dataset was divided into training set and test set by 8∶2.A deep neural networks(DNN)multiple operational status identification algorithm was constructed based on speed,heading,duration,latitude and longitude.The operation days,operation date and corresponding operation location information of the fishing vessels were extracted,compared and verified with the actual recorded fishing log data.The results show that:the errors between the extracted result and the actual recorded result are small,and the monthly difference in the number of operation days is within 0-2 d,the average accuracy rate is 93.88%;the difference in operation distance is small,and 94%of the total is within 20 km;compared with k-nearest neighbor(KNN),logistic regression,Bayesian algorithm and decision tree,DNN performed better in vessel operational status identification,with the best accuracy of 87.04%.The research can provide some references for identifying the operational status,fishing and fisheries management of Antarctic krill trawlers.
作者
苏冰
何瑞麟
赵国庆
蒋沛雯
李阳
商宸
韩海斌
沈烈
张衡
SU Bing;HE Ruilin;ZHAO Guoqing;JIANG Peiwei;LI Yang;SHANG Chen;HAN Haibin;SHEN Lie;ZHANG Heng(College of Navigation and Ship Engineering,Dalian Ocean University,Dalian Liaoning 116023,China;Laoshan Laboratory of Qingdao Marine Science and Technology Center,Qingdao Shandong 266237,China;Key Laboratory of East China Sea&Oceanic Fishery Resources Exploitation and Utilization,Ministry of Agriculture and Rural Affairs,East China Sea Fisheries Research Institute,Chinese Academy of Fishery Sciences,Shanghai 200090,China;Key Laboratory of Fisheries Remote Sensing,Ministry of Agriculture and Rural Affairs,Shanghai 200090,China;Henan Vocational University of Science and Technology,Zhoukou Henan 461300,China;Anhui Normal University,Wuhu Anhui 241000,China;Shanghai Ocean University,Shanghai 201306,China)
出处
《海洋渔业》
北大核心
2025年第5期650-661,共12页
Marine Fisheries
基金
青岛海洋科技中心山东省专项经费(2022QNLM030002-1)
国家重点研发计划(2022YFC2807504)
中国水产科学研究院东海水产研究所中央级公益性科研院所基本科研业务费专项资金(2021M06)。
关键词
南极磷虾
船位数据
作业状态
深度学习
Antarctic krill
vessel location data
operational status
deep learning
