A novel structural adaptive seasonal grey Bernoulli model in natural gas production forecasting

Abstract

Energy data are characterized by nonlinearity, seasonality and growth, which render prediction challenging, making it difficult for Chinese government to formulate relevant policies. Considering that the grey Bernoulli model can deal with nonlinear and growing data, and utilizing the newly proposed seasonal factor, we propose a novel structural adaptive seasonal grey Bernoulli model, which is limited to application to nonlinear, seasonal, and growing time sequence. Firstly, the traditional seasonal factors are logarithmically transformed to ensure more refined seasonal factor sequence. Then, the grey Bernoulli model and the grey structural adaptive model are combined in a multiplicative manner to create a new model, thereby refining its structural adaptability. And the Grey Wolf Optimization algorithm is used to optimize the parameters in the model. Finally, the natural gas production prediction is implemented to verify the feasibility of this model. The mean absolute percentage errors of the proposed model are below 5% both in the experiment and in the robustness test. Compared with other models, this model demonstrates an improvement of 29.75% in prediction accuracyand 15.80% in robustness test. Additionally, the Diebold & Mariano statistics exhibit values that are essentially significant at the 10% level. The experimental results substantiate the validity, superiority, and robustness of the model proposed in this paper, which can be effectively applied to the prediction of seasonal natural gas production.