基于EMD分解的海浪有效波高短期预测研究

Short-Term Prediction of Significant Wave Height Based on EMD Decomposition

摘要: 海浪与人类的海上活动密切相关，其在航海运输、海洋开发、防灾减灾等领域中具有重要地位。海浪预报，尤其是海浪有效波高的预报一直是海洋研究的重点。由于海浪具有很多的随机特性和不确定因素，海浪预报的准确度依然是目前要解决的关键问题。本文深入研究了渤海海浪数据，拟采用基于经验模式分解(EMD)和支持向量回归模型(SVR)联合的混合模型进行短期海浪有效波高预测。经验模式分解能够自适应地将非平稳时间序列分解为频率由高到低的一系列本征模函数(Intrinsic Mode Function, IMF)和残差，然后根据本征模函数各自趋势变化的剧烈程度选择不同的核函数进行支持向量回归预测，接着对各预测分量进行加权组合，得到原始序列的准确预测值。渤海2012年的海浪有效波高短期预测实验表明，基于经验模式分解(EMD)和支持向量回归模型(SVR)联合的混合模型预测结果比单一的支持向量回归模型预测结果更准确。

Abstract: Waves are closely related to human activities at sea, and play an important role in the fields of na-vigation, marine development, disaster prevention and mitigation. Wave prediction, especially the prediction of significant wave height, has always been the focus of ocean research. Because of the random characteristics and uncertainties of waves, the accuracy of wave prediction is still the key problem to be solved. In this paper, the wave data in Bohai Sea are deeply studied to predict the short-term significant wave height by a hybrid model based on empirical mode decomposition (EMD) and support vector regression (SVR). Empirical mode decomposition (EMD) can adaptively decompose non-stationary time series into a series of intrinsic mode functions (IMFs) and residuals with frequencies from high to low. According to the severity of trend changes of the IMFs, different kernel functions are selected for support vector regression prediction. Then, the weighted combina-tion of each prediction component is used to obtain the accurate prediction value of the original se-ries. The short-term prediction experiment of significant wave height in Bohai Sea in 2012 shows that the hybrid model based on empirical mode decomposition (EMD) and support vector regres-sion model (SVR) is more accurate than the single support vector regression model.