Scholarlink

<p>Owing to increasing human activities and climate events over the past few decades, reliable hydrological forecasting has become a critical yet challenging task in water resources planning, reservoir operation, and water supply management. This study presents a hybrid artificial intelligence-based approach for hydrological prediction, combining signal decomposition techniques and optimization algorithms. Specifically, Variational Mode Decomposition (VMD) is employed to capture the irregular and dynamic features of the original hydrological time series. Subsequently, a Long Short-Term Memory (LSTM) network is applied to develop an effective forecasting model for each decomposed component, with the Whale Optimization Algorithm (WOA) employed to identify the optimal parameter set for each LSTM model. The proposed method is tested on runoff datasets from three hydrological stations in China, and its feasibility is demonstrated through several quantitative performance metrics. Extensive simulations reveal that the proposed model outperforms traditional models across various scenarios. With high accuracy and stability, this data-driven approach provides a robust solution to the complex problem of hydrological forecasting. Furthermore, the integration of signal decomposition and evolutionary algorithms enhances the performance of artificial intelligence models in regression tasks.</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

Compound Hydrological Forecasting Model by Long Short-term Memory Network Coupled with Adaptive Mode Decomposition and Evolutionary Algorithm

  • 期刊论文

  • 发表日期：

    2025年1月2日

查看全文
  • Zhong-kai Feng,
  • Wen-jie Liu,
  • Wen-jing Niu,
  • Tao Yang,
  • Wen-chuan Wang,
  • Sen Wang

摘要

Owing to increasing human activities and climate events over the past few decades, reliable hydrological forecasting has become a critical yet challenging task in water resources planning, reservoir operation, and water supply management. This study presents a hybrid artificial intelligence-based approach for hydrological prediction, combining signal decomposition techniques and optimization algorithms. Specifically, Variational Mode Decomposition (VMD) is employed to capture the irregular and dynamic features of the original hydrological time series. Subsequently, a Long Short-Term Memory (LSTM) network is applied to develop an effective forecasting model for each decomposed component, with the Whale Optimization Algorithm (WOA) employed to identify the optimal parameter set for each LSTM model. The proposed method is tested on runoff datasets from three hydrological stations in China, and its feasibility is demonstrated through several quantitative performance metrics. Extensive simulations reveal that the proposed model outperforms traditional models across various scenarios. With high accuracy and stability, this data-driven approach provides a robust solution to the complex problem of hydrological forecasting. Furthermore, the integration of signal decomposition and evolutionary algorithms enhances the performance of artificial intelligence models in regression tasks.