Recurrence Plots and Cross Recurrence Plots

Engineering Applications of Artificial Intelligence, 145, 110218p. (2025) DOI:10.1016/j.engappai.2025.110218

A hybrid prediction model for marine wind speed considering internal temporal features recombination and external variables association

Marine wind exists widely as a major environmental disturbance in the process of marine resource exploration. Accurate short-term prediction of marine wind speed can ensure the safety of offshore exploration operations. The current wind speed prediction model has prediction lag, and lacks consideration of chaotic characteristics and the influence of external variables. Analyzing the chaotic time series characteristics of marine wind speed and their causative correlations, a hybrid prediction model for marine wind speed is presented in this paper, which incorporates the recombination of internal temporal features and external variables association. Firstly, the marine wind speed is decomposed using the improved complete ensemble empirical mode decomposition with adaptive noise, with features recombined based on a comprehensive evaluation of multiple entropies and hierarchical clustering. Subsequently, for the chaotic components characterized by strong randomness and nonlinearity, successive variational mode decomposition is used for a secondary decomposition, with chaotic reconstruction based on their recurrence plot characteristics. Then, a hybrid prediction model for marine wind speed considering internal temporal features recombination and external variables association is proposed. Finally, experiments are performed by using actual data from a marine buoy. The result shows that the hybrid prediction model can accurately predict the marine wind speed for the next time step, providing advanced environmental perception for marine resource exploration.