Scholarlink

<p>Modern dam safety management requires comprehensive instrumentation systems to monitor structural behavior and process critical data. Such systems play a vital role in identifying both immediate anomalies and gradual trends that may indicate potential safety hazards. This study presents an innovative spatiotemporal modeling approach that leverages data from properly functioning instruments to reconstruct missing measurements through panel data analysis. The research focuses on developing a robust pore pressure prediction system for Eyvashan Dam by integrating multiple artificial intelligence techniques—including Feed-Forward Neural Networks (FFNN), Support Vector Regression (SVR), Group Method of Data Handling (GMDH), and Ensemble Artificial Neural Networks (EANN)—with Fuzzy C-Means Clustering (FCM) methodology. The proposed framework was evaluated across three distinct operational scenarios, analyzing clustered measurement points to generate accurate pore pressure forecasts. Key findings demonstrate that combining data from a target piezometer with two neighboring instruments significantly enhances monitoring accuracy across all AI models tested. Comparative analysis revealed the EANN model to be particularly effective, establishing it as a reliable tool for pore pressure zone identification and monitoring in earth-fill dam structures. These results highlight the potential of integrated AI approaches to improve dam safety assessment, especially in situations with partial instrumentation failures or data gaps.</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

Optimizing pore pressure prediction in earth dams through the integration of panel data and intelligent models

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年7月27日

查看全文
  • Behrang Beiranvand,
  • Taher Rajaee,
  • Mehdi Komasi

摘要

Modern dam safety management requires comprehensive instrumentation systems to monitor structural behavior and process critical data. Such systems play a vital role in identifying both immediate anomalies and gradual trends that may indicate potential safety hazards. This study presents an innovative spatiotemporal modeling approach that leverages data from properly functioning instruments to reconstruct missing measurements through panel data analysis. The research focuses on developing a robust pore pressure prediction system for Eyvashan Dam by integrating multiple artificial intelligence techniques—including Feed-Forward Neural Networks (FFNN), Support Vector Regression (SVR), Group Method of Data Handling (GMDH), and Ensemble Artificial Neural Networks (EANN)—with Fuzzy C-Means Clustering (FCM) methodology. The proposed framework was evaluated across three distinct operational scenarios, analyzing clustered measurement points to generate accurate pore pressure forecasts. Key findings demonstrate that combining data from a target piezometer with two neighboring instruments significantly enhances monitoring accuracy across all AI models tested. Comparative analysis revealed the EANN model to be particularly effective, establishing it as a reliable tool for pore pressure zone identification and monitoring in earth-fill dam structures. These results highlight the potential of integrated AI approaches to improve dam safety assessment, especially in situations with partial instrumentation failures or data gaps.