Accurate combination forecasting of wave energy based on multiobjective optimization and fuzzy information granulation

Abstract

Wave energy forecasting modeling is critical for promoting renewable energy storage technology as well as for energy sustainability and global carbon neutrality goals. However, due to the irregular volatility and complexity in wave energy data, all the effective information cannot be fully utilized by a traditional forecasting model; moreover, the point forecasting results cannot be used to effectively analyze the uncertainty of the time series. To overcome these shortcomings, a multistep point-interval combined significant wave height forecasting system based on the multiobjective grasshopper optimization algorithm and the fuzzy information granulation strategy is proposed to forecast the half-hour actual wave height at different buoy locations. Applying this system, Pareto optimal weights can be obtained to integrate the respective advantages of deep learning and neural network models in the combined forecasting module, achieve the best point and interval forecasting accuracy and accurately analyze the uncertainty of point forecasting results. Among the combined models, the proposed system has a more comprehensive and scientific prediction performance than other models (MAPE = 4.9866 for Site 1, MAPE = 4.9138 for Site 2, and MAPE = 3.9572 for Site 3). The forecasting outcomes indicate that the developed system significantly improves forecasting accuracy and stability, which provides reliable technical support for the sustainable development of wave power generation.