文献类型：期刊文献

英文题名：An adaptive tidal forecasting model using EMD-PCA decomposition and LSTM-RBF nonlinear optimization

作者：Wang, Rui[1];Yin, Jianchuan[1,2];Xu, Dongxing[1,2];Wang, Nini[3]

机构：[1]Guangdong Ocean Univ, Naval Architecture & Shipping Coll, Zhanjiang 524009, Peoples R China;[2]Guangdong Prov Key Lab Intelligent Equipment South, Zhanjiang 524088, Peoples R China;[3]Guangdong Ocean Univ, Coll Math & Comp, Zhanjiang 524088, Peoples R China

年份：2026

卷号：355

期号：P1

外文期刊名：OCEAN ENGINEERING

收录：SCI-EXPANDED(收录号：WOS:001724220700002)、、EI(收录号：20261420425390)、WOS

语种：英文

外文关键词：Tide prediction; Radial basis function; Nonlinear mapping; Empirical mode decomposition; Long short-term memory network; Principal components analysis

外文摘要：High-precision tidal forecasting poses significant challenges in ocean engineering applications attributed to the nonlinearity and uncertainty of tidal systems caused by environmental disturbances. To address this issue, this study proposes an adaptive real-time tidal forecasting framework that integrates Empirical Mode Decomposition (EMD), Principal Component Analysis (PCA), Long Short-Term Memory (LSTM) networks, and Radial Basis Function (RBF) nonlinear optimization. Non-astronomical factors significantly correlated with tidal variations are first identified using Pearson correlation analysis and incorporated into the tidal residual series derived from harmonic analysis. The reconstructed residuals are decomposed into intrinsic mode functions using EMD, and PCA is subsequently applied to extract dominant features and reduce dimensionality. LSTM is employed to model temporal dependencies, while an RBF network is further used to nonlinearly refine the LSTM outputs. The proposed model is validated using real tidal data from Canaveral Port, Old Port Tampa, and Gulf of Mexico Port. Experimental results demonstrate that the proposed approach significantly outperforms harmonic analysis, conventional neural networks, and Transformer-based models in terms of MAE and RMSE, while maintaining lower computational cost. These results confirm the effectiveness and robustness of the proposed framework for high-accuracy real-time tidal forecasting under complex environmental conditions.