Application of coupled IGWO-GMDH model in the prediction of significant wave height

Ocean waves are characterised as random and non-linear. Predicting significant wave height is critical for ensuring the safety of ship navigation and route planning. In the present study, the Grey Wolf optimiser was improved by optimising the search mechanism and coupled it with the Grouping Method Data Handling model to construct an effective significant wave height prediction model. This novel prediction model was validated using a significant wave height dataset. The weights of the different model variables were also explored. The results show that the IGWO-GMDH model is more accurate. The mean square error decreased by 2.65%, and the root mean square error decreased by approximately 1.35%. The standard deviation was reduced by 2.14%. Additionally, the weights of the wave characteristic parameters and the wind field data are relatively high; combining these would significantly impact the model's accuracy. The IGWO-GMDH model will provide more robust predictions of significant wave height and support research into ship navigation safety and route planning and optimisation.