基于变分模态分解与注意力机制的海洋风速预测

Ocean wind speed prediction based on variational mode decomposition and attention mechanism

海洋风速预测对远洋航行安全与航线规划具有重大影响.风速同时受多种外在自然因素影响,表现出强烈的非线性、非平稳性与随机性等特性,使得预测准确性受到极大考验.为提高风速预测准确性,创新性地提出一种基于变分模态分解与融合注意力机制的神经网络的风速预测方法.首先,利用变分模态分解将风速序列分解为一系列调幅调频信号,以降低数据复杂度,有效提取特征并提高噪声鲁棒性,减少风速自身对预测准确性的影响.其次,对分解后的不同模态子序列利用融合注意力机制的神经网络进行风速预测.最后,用实测数据验证所提方法的有效性.与其他典型风速预测模型相比,所提方法可有效提高风速预测准确性.

Ocean wind speed prediction has a major impact on the safety of ocean navigation and route planning. Wind speed is affected by a variety of external natural factors, showing strong nonlinearity, non-stationarity and randomness, which greatly challenges the prediction accuracy. In order to improve the accuracy of wind speed prediction, a wind speed prediction method based on variational mode decomposition(VMD) and neural network integrating attention mechanism is creatively proposed. Firstly, VMD is used to decompose the wind speed sequence into a series of amplitude-modulated and frequency-modulated signals to reduce data complexity, effectively extract features, improve noise robustness, and reduce the impact of wind speed itself on the accuracy of prediction. Then, the decomposed different modal subsequences are used to predict the wind speed by the neural network integrating attention mechanism. Finally, the effectiveness of the proposed method is verified by practical data. Compared with other typical wind speed prediction models, the proposed method can effectively improve the accuracy of wind speed prediction.