Scholarlink

<p>Natural gas is a primary energy source in the energy landscape, with grey modeling serving as the mainstream method for natural gas production forecasting. Cumulative generation is the key technology for enhancing grey modeling performance. To address the limitations of existing cumulative operators—namely their rigid weight structures that struggle to adaptively match historical data patterns and lack flexibility—this paper introduces the sine-based adaptive accumulation generation operator. Based on optimal angle interval partitioning via sine functions, sine-based adaptive accumulation generation operator incorporates a dynamic starting angle <i>θ</i>. By leveraging the monotonic properties of sine functions, it assigns appropriate weights to cumulative data. Integrating sine-based adaptive accumulation generation operator accumulation into a Bernoulli model establishes the NGBM<sub>sin</sub>(1,1) model. Nonlinear parameters are optimized using a differential evolution algorithm to enhance the model's adaptability and accuracy. The model's effectiveness is validated through two experimental approaches: First, the constructed model is compared with seven superior models for forecasting natural gas consumption in the Middle East; Second, the model was compared with a Bernoulli model incorporating four distinct accumulation methods, along with time series models and machine learning models, for forecasting natural gas production and reserves across China and Shaanxi province. Four primary evaluation metrics were selected for comparative analysis. Across seven experimental settings, NGBM<sub>sin</sub>(1,1) ranked first in predicted mean absolute percentage error in five groups, ranked first in the mean absolute error metric in six groups, and ranked first in the root mean square error metric in four groups. It also achieved the top overall ranking in the comprehensive metric across six comparison experiments. This validates that the proposed NGBM<sub>sin</sub>(1,1) possesses highly efficient predictive performance and overall error control capabilities, effectively capturing the nonlinear fluctuation characteristics of natural gas data. It provides a reliable methodological foundation for forecasting natural gas production, consumption, and reserves.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Grey Prediction Model Based on Sine Cumulative Generation and Its Application in Natural Gas Forecasting

  • 期刊论文

  • 发表日期：

    2026年5月2日

查看全文
  • Wenyan Guo,
  • Xiaoqi Xiao,
  • Xiujuan Hua,
  • Linbo Zhang

摘要

Natural gas is a primary energy source in the energy landscape, with grey modeling serving as the mainstream method for natural gas production forecasting. Cumulative generation is the key technology for enhancing grey modeling performance. To address the limitations of existing cumulative operators—namely their rigid weight structures that struggle to adaptively match historical data patterns and lack flexibility—this paper introduces the sine-based adaptive accumulation generation operator. Based on optimal angle interval partitioning via sine functions, sine-based adaptive accumulation generation operator incorporates a dynamic starting angle θ. By leveraging the monotonic properties of sine functions, it assigns appropriate weights to cumulative data. Integrating sine-based adaptive accumulation generation operator accumulation into a Bernoulli model establishes the NGBMsin(1,1) model. Nonlinear parameters are optimized using a differential evolution algorithm to enhance the model's adaptability and accuracy. The model's effectiveness is validated through two experimental approaches: First, the constructed model is compared with seven superior models for forecasting natural gas consumption in the Middle East; Second, the model was compared with a Bernoulli model incorporating four distinct accumulation methods, along with time series models and machine learning models, for forecasting natural gas production and reserves across China and Shaanxi province. Four primary evaluation metrics were selected for comparative analysis. Across seven experimental settings, NGBMsin(1,1) ranked first in predicted mean absolute percentage error in five groups, ranked first in the mean absolute error metric in six groups, and ranked first in the root mean square error metric in four groups. It also achieved the top overall ranking in the comprehensive metric across six comparison experiments. This validates that the proposed NGBMsin(1,1) possesses highly efficient predictive performance and overall error control capabilities, effectively capturing the nonlinear fluctuation characteristics of natural gas data. It provides a reliable methodological foundation for forecasting natural gas production, consumption, and reserves.