The recursive grey model and its application

The fixed structure parameters limit the performance of grey prediction algorithm in unsmoothed time series prediction tasks. Based on the mechanism of recursive iteration, this study proposes a recursive grey model to predict the future trend of renewable energy generation in China. The proposed method has the following improvements: the first is the priority of new information. By introducing a memory factor parameter into the objective function, recursive grey model gives more influence to new observations. The second is dynamic updating scheme of structure parameters and improved iteration base values. The proposed model uses recursive method to solve the structure parameters of each observation, and the new parameters can inherit the previous experience. The third is stable forecasting performance. Based on the principle of minimizing the prediction error of in-sample data, the memory factor optimization method is given. Without additional verification set, recursive grey model can synchronize training and test of in-sample data to avoid over fitting. Finally, the proposed model is applied to predict China's renewable energy generation. The forecast results show that in the next few years, China's hydropower, wind power and solar power generation will grow at an average annual growth rate of 2.6%, 20.7% and 24.9%, respectively. It can be predicted that China's hydropower generation, wind power generation and solar power generation will reach 1475 billion kWh, 3094 billion kWh and 1352 billion kWh, respectively in 2030.