Scholarlink

<p>Short-Term Load Forecasting (STLF) predicts electricity demand over a short period, typically ranging from hours to a few days. STLF plays a critical role in power system planning and operations, aiding in efficient resource allocation and reliable grid operations. This paper presents a novel approach for STLF, leveraging the Chaotic Class Topper Optimization (CCTO) algorithm and a Delayed Gated Recurrent Unit (DGRU) neural network. The proposed methodology integrates the robust optimization capabilities of CCTO with the enhanced memory and learning abilities of the modified DGRU architecture to improve forecasting accuracy. The CCTO algorithm, inspired by chaotic dynamics, efficiently explores the solution space, while the DGRU model captures complex temporal dependencies within the load data. Experiments are conducted using real-world load datasets to evaluate the performance of the proposed approach against traditional methods. The evaluation is based on an aggregated, system-level hourly electricity load dataset from Panama city, making the proposed work directly relevant to area-level STLF. Results demonstrate that proposed CCTO-DGRU model outperforms some models in terms of forecasting accuracy and computational efficiency. Moreover, the model exhibits resilience to noisy and nonlinear load patterns, making it suitable for practical deployment in power system operations. The proposed CCTO-DGRU model achieved 1.30-25.10% reduction in RMSE when compared with other models. This research contributes to advance STLF techniques, offering valuable insights for enhancing the reliability and sustainability of power systems.</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

Harnessing Load Dynamics: A Novel CCTO-Enhanced DGRU for Short-Term Load Forecasting

  • 期刊论文

  • 发表日期：

    2026年2月13日

查看全文
  • Maloy Kumar Dey,
  • Yogeeshwar Charasala,
  • Dushmanta Kumar Das

摘要

Short-Term Load Forecasting (STLF) predicts electricity demand over a short period, typically ranging from hours to a few days. STLF plays a critical role in power system planning and operations, aiding in efficient resource allocation and reliable grid operations. This paper presents a novel approach for STLF, leveraging the Chaotic Class Topper Optimization (CCTO) algorithm and a Delayed Gated Recurrent Unit (DGRU) neural network. The proposed methodology integrates the robust optimization capabilities of CCTO with the enhanced memory and learning abilities of the modified DGRU architecture to improve forecasting accuracy. The CCTO algorithm, inspired by chaotic dynamics, efficiently explores the solution space, while the DGRU model captures complex temporal dependencies within the load data. Experiments are conducted using real-world load datasets to evaluate the performance of the proposed approach against traditional methods. The evaluation is based on an aggregated, system-level hourly electricity load dataset from Panama city, making the proposed work directly relevant to area-level STLF. Results demonstrate that proposed CCTO-DGRU model outperforms some models in terms of forecasting accuracy and computational efficiency. Moreover, the model exhibits resilience to noisy and nonlinear load patterns, making it suitable for practical deployment in power system operations. The proposed CCTO-DGRU model achieved 1.30-25.10% reduction in RMSE when compared with other models. This research contributes to advance STLF techniques, offering valuable insights for enhancing the reliability and sustainability of power systems.