A novel seasonal adaptive grey model with the data-restacking technique for monthly renewable energy consumption forecasting

To provide accurate renewable energy forecasts that adapt to the country's sustainable development, a novel seasonal model combined with the data-restacking technique is proposed in this paper. Specifically, the data-restacking technique is initially utilized to eliminate the seasonal fluctuations of the collected observations, which can eliminate the fundamental flaws in conventional seasonal grey models. Subsequently, the time function term is originally designed to incorporate into the dynamic structure to reflect the cumulative time effects, which can smoothly describe the dynamic changes and significantly improve the robustness of the novel model. Further, the self-adaptive parameters optimized using particle swarm optimization can effectively enhance the adaptability and generalization of the proposed model. For elaboration and verification purposes, experiments on forecasting American monthly renewable energy consumption in the commercial sector and industrial solar energy have been implemented compared to a range of benchmark models, including other prevalent grey prediction models, statistical approaches, and machine learning methods. Experimental results demonstrate that this new model presents more successful outcomes than the other benchmarks in overall and restacking performance.