Game mechanism and event-triggering based Natural Gas Demand Prediction (GMET-NGDP) model for Chemical and Fertilizer Industry.

In order to improve the accuracy of the natural gas demand forecast of complexed chemical fertilizer industry, the Game mechanism and event-triggering combined machine learning model was proposed, and the natural gas demand of chemical fertilizer enterprises in Sichuan province was taken as examples for validation analysis. The results show that: (1) the prediction results of GMET-NGDP model are more accurate than those of the single prediction models, which avoids the prediction risk of the single prediction model; (2) The errors of GMET-NGDP combined forecasting model are lower than those of statistical forecasting models, which improves the prediction accuracy. It is concluded that the combined GMET-NGDP forecasting model has a good forecasting performance in the prediction of natural gas demand in the nonlinear chemical and fertilizer industry, and the prediction results of natural gas demand can be used as the basis for the formulation of industrial natural gas demand policies. The forecast results show that the chemical fertilizer industry in China has stable demand for natural gas under different events.