A fractional grey reservoir computing prediction model and its application in clean energy forecasting

The prediction of clean energy is of great significance for the transformation of the power system and the achievement of the carbon neutrality goal. However, due to the characteristics of small sample size and strong volatility of clean energy data, as well as the deficiencies of existing prediction models in capturing nonlinear features and short-term prediction accuracy, accurate prediction faces great challenges. This study proposes an innovative fractional-order grey reservoir computing prediction model. The model effectively combines the fractional-order calculus theory with the reservoir framework, effectively solving the limitations of traditional grey models in nonlinear scenarios and the sensitivity of machine learning models to small sample data. Experimental results show that in the prediction of solar and wind energy, the proposed model’s accuracy is superior to that of the classical grey model and machine learning models, demonstrating its stability and predictive accuracy. The work of this study provides a new solution for complex time series prediction by integrating fractional-order calculus theory with reservoir computing. Meanwhile, the prediction results facilitate scientific planning and precise regulation of clean energy production and consumption for decision-makers.