基于LSTM-RNN的苏格兰鱼群迁移问题研究

The Fish Migration Problem in the Scotland Based on LSTM-RNN

摘要: 本文旨在通过建立北大西洋研究海域海洋温度时间序列模型预测苏格兰鲱鱼和鲭鱼鱼群未来50年的迁移轨迹，给予周边渔业公司经营战略。首先，本文通过对影响鲱鱼和鲭鱼鱼群迁移的相关因素进行机理分析，选取适宜温度最佳阈值，接着结合研究海域的相关情况选取合适的样本点建立基于LSTM-RNN的海洋温度时间序列模型，从而给出了未来50年鱼群最可能出没的位置。同时根据预测得到的海水温度变化情况，给出鱼群位置变化的极端情况，进而为周边渔业公司提出经营策略。最后，综合考虑军事政治因素，验证了我们模型具有较强的适用性和合理性。

Abstract: In this paper, the migration of herring and mackerel in Scotland in the next 50 years is studied based on LSTM-RNN theory. By analysis of the corresponding factors, one model named time series model based on LSTM-RNN is given under some proper assumptions; based this model, the management strategy for the surrounding fishery companies is given according to the extreme situation of fish position. Ultimately, considering the military and political factors, our model is relatively applicability and feasible.