RF-BiLSTM神经网络在海浪预测中的应用  

Application of RF-BiLSTM Neural Network in Ocean Wave Forecast

作　　者：李海涛[1] 孙亚男 付建浩 LI Hai-Tao;SUN Ya-Nan;FU Jian-Hao(College of Information Science and Technology,Qingdao University of Science and Technology,Qingdao 266061,China)

机构地区：[1]青岛科技大学信息科学技术学院,青岛266061

出　　处：《计算机系统应用》2022年第6期331-338,共8页Computer Systems & Applications

基　　金：国家自然科学基金(61806107);青岛市创新创业领军人才(15-07-03-0030);农业部水产养殖数字建设试点项目(2017-A2131-130209-K0104-004)。

摘　　要：由于风浪数据的随机性,复杂性,影响因素多,多为时间序列的特点,造成了传统预测模型预测难度大,精确率低,构建了基于随机森林的注意力机制与双向长短期记忆神经网络相结合的海浪预测模型.该模型对输入进行优化,可以使用过去和未来的数据信息进行预测,提高了海浪波高的预测精度.该模型利用随机森林对输入变量筛选优化,降低网络复杂度,然后将注意力机制与双向长短期记忆神经网络相结合建立预测模型,并利用实际数据进行验证.结果显示,和BP, LSTM, BiLSTM模型比较, RF-BiLSTM模型的预测精度更高,拟合程度更好,在海浪数值的预测预报中有重要意义.Due to the randomness, complexity, many influencing factors, and dominant time series of wind and wave data,traditional prediction models have great prediction difficulty and low prediction accuracy. In response, this study proposes an ocean wave prediction model combining the attention mechanism of the random forest with the bidirectional long short-term memory(BiLSTM) neural network. The model optimizes the inputs and can predict ocean waves with past and future data to improve the prediction accuracy on the wave height. It uses the random forest to filter and optimize the input variables and thereby reduce the network complexity. Then, the attention mechanism is combined with the BiLSTM neural network to build a prediction model, which is subsequently verified on actual data. The results show that compared with the BP, LSTM, and BiLSTM models, the RF-BiLSTM model has higher prediction accuracy and fitting degree and thereby has good application value in the prediction of ocean wave values.

关 键 词：海浪预测 RF-BiLSTM 随机森林 注意力机制 双向长短期记忆 时间序列 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] P731.33[自动化与计算机技术—控制科学与工程]