Abstract: n response to the variability of fishing boat trajectories, this study aims to improve the accuracy of the prediction model by optimizing the characteristic parameters of fishing boats during the data preprocessing stage, in order to enhance the accuracy of predicting fishing boat berthing trajectories. Propose a fishing vessel berthing trajectory prediction model based on Beidou ship position data and combined with Long Short Term Memory (LSTM) network. Collect Beidou fishing vessel position data through a Vessel Monitoring Systems (VMS) onboard terminal, extract spatiotemporal position information and other feature parameters, preprocess the collected Beidou fishing vessel position data, select input feature parameters for the prediction model using correlation analysis, classify the feature parameters according to fishing vessel size and type, and train the model. Finally, compare the predicted trajectory with the actual berthing trajectory. Exploring the practicality of Beidou ship position data in ship trajectory prediction and the impact of fishing vessel types on berthing trajectory prediction. The final experimental results showed that the accuracy of the model prediction reached 92.3%, proving the superiority of Beidou ship position data in ship trajectory prediction research. At the same time, it proved the conclusion that the type of fishing captain is positively correlated with the longitude of trajectory prediction, providing a new method for port and fishery management.

Key words: fishing boat, trajectory prediction, beidou ship position data: LSTM, network model

摘要： 针对渔船轨迹多变性的特点，本研究通过在数据预处理阶段对渔船特征参数的优化提高预测模型的精度，旨在能够提高渔船停泊轨迹预测的精度，提出一种基于北斗船位数据并结合长短期记忆网络的（LongShort-TermMemory，LSTM）的渔船停泊轨迹预测模型。通过船舶监控系统（Vessel Monitoring Systems，VMS）船载终端收集北斗渔船船位数据，提取时空位置信息等特征参数，将所收集的北斗船位数据进行预处理，通过使用相关性分析法选择预测模型的输入特征参数，将特征参数按渔船大小类型分类输入并训练模型，最终通过对比预测轨迹与实际停泊轨迹。探究北斗船位数据在船舶轨迹预测的实用性以及渔船类型对停泊轨迹预测的影响。结果显示，模型预测的精度达到92.3%，证明了北斗船位数据在船舶轨迹预测研究的优越性，同时证明了渔船类型与轨迹预测精度呈正相关。本研究为港口以及渔业管理提供了一种新方法。

关键词: 渔船, 轨迹预测, 北斗船位数据：LSTM, 网络模型