Scholarlink

<p>Accurately forecasting regional carbon emissions is crucial for devising effective climate mitigation policies. However, existing models often struggle to capture heterogeneous characteristics inherent in megalopolis emission data. To address this, this study introduces a novel multi-source, multi-process framework that integrates a unified new information accumulation generation operator (<i>AGO</i>), the nonlinear grey Bernoulli model (<i>NGBM</i>), and the Crested Porcupine Optimizer (<i>CPO</i>), designed to tackle multi-characteristic nonlinear time series in regional carbon emission prediction. Building upon <i>NGBM</i>, the adaptability of the new <i>AGO</i>, combined with the <i>CPO</i> algorithm’s ability to adjust the population dynamically, ensures the flexibility and robustness of the proposed framework. Empirically, the proposed framework is evaluated against nine top-performing models from four key dimensions: the proposed framework’s accuracy, the <i>AGO</i>’s superiority, the optimization algorithm’s applicability, and the proposed framework’s generalizability. The results show that the proposed framework demonstrates an average forecasting <i>MAPE</i> value below 6%, significantly outperforming the existing counterparts. Furthermore, consistency tests between one-step and two-step ahead recursive predictions confirm the model’s accuracy in forecasting results over the next four years. Thus, this novel framework presents a robust instrument and is applied to forecast the Beijing-Tianjin-Hebei region’s carbon emissions, offering valuable insights for informed policies and management.</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 novel integrated framework of the innovative data-preprocessing technique, grey Bernoulli model, and advanced crested Porcupine optimizer for forecasting carbon emissions with improved accuracy

  • 期刊论文

  • 发表日期：

    2025年11月18日

查看全文
  • Song Ding,
  • Yi Wang,
  • Zhijian Cai,
  • Mingfei Zhang,
  • Fangyuan Qian

摘要

Accurately forecasting regional carbon emissions is crucial for devising effective climate mitigation policies. However, existing models often struggle to capture heterogeneous characteristics inherent in megalopolis emission data. To address this, this study introduces a novel multi-source, multi-process framework that integrates a unified new information accumulation generation operator (AGO), the nonlinear grey Bernoulli model (NGBM), and the Crested Porcupine Optimizer (CPO), designed to tackle multi-characteristic nonlinear time series in regional carbon emission prediction. Building upon NGBM, the adaptability of the new AGO, combined with the CPO algorithm’s ability to adjust the population dynamically, ensures the flexibility and robustness of the proposed framework. Empirically, the proposed framework is evaluated against nine top-performing models from four key dimensions: the proposed framework’s accuracy, the AGO’s superiority, the optimization algorithm’s applicability, and the proposed framework’s generalizability. The results show that the proposed framework demonstrates an average forecasting MAPE value below 6%, significantly outperforming the existing counterparts. Furthermore, consistency tests between one-step and two-step ahead recursive predictions confirm the model’s accuracy in forecasting results over the next four years. Thus, this novel framework presents a robust instrument and is applied to forecast the Beijing-Tianjin-Hebei region’s carbon emissions, offering valuable insights for informed policies and management.