Abstract:

To address the existing and limited original data and lower prediction robustness, a new multivariate regression forecasting model combined with gliding data barycenter was proposed. In this new forecasting method, the original data was interpolated and the corresponding data barycenter was optimized. Then, the important impact factors of ration consumption were analysed and chosen for the multivariate regression model. In simulation experiments, the training data of 35 years (1981-2015) were used, and the results have shown that the proposed method can greatly improve prediction accuracy and robustness.