Scholarlink

<p>As the number of closed mines increases, the problem of surface deformation is becoming increasingly prominent. Traditional monitoring methods have limitations, and this paper proposes a precise prediction method combining SBAS-InSAR and the DBO–CNN–LSTM (dung beetle optimization–convolutional neural network–long short-term memory) model for closed mine surface deformation. Utilizing SBAS-InSAR for high-precision data acquisition and DBO–CNN–LSTM for complex feature learning and time series analysis, the method provides accurate long-term predictions. Based on five years of Huainan mining area data, post-closure deformation trends vary dynamically. The Lizuizi Mine transitioned from subsidence to uplift, while the Xinzhuangzi Mine experienced significant uplift, influenced by geological conditions, closure time, and subsequent activities. The DBO–CNN–LSTM model outperformed other neural networks. For the Xieyi Mine, 26,165 corresponding points between model predictions and SBAS-InSAR results showed significant correlation, with <i>R</i><sup>2</sup> of 0.91 and correlation coefficient of 0.95 on March 31, 2022, and <i>R</i><sup>2</sup> of 0.92 and correlation coefficient of 0.97 on February 6, 2023. This paper reveals surface deformation mechanisms in closed mines, including groundwater level changes, and provides a scientific basis for ecological restoration, development, geological disaster warning, and improved mining area management and policy formulation.</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

High-Precision Temporal Monitoring and Prediction of Surface Deformation at Closed Mines Using Time Series InSAR and the Deep Learning DBO–CNN–LSTM Model

  • 期刊论文

  • 发表日期：

    2025年11月14日

查看全文
  • Jin Luo,
  • Qingbiao Guo,
  • Yingming Li,
  • Songbo Wu,
  • Xin Lyu

摘要

As the number of closed mines increases, the problem of surface deformation is becoming increasingly prominent. Traditional monitoring methods have limitations, and this paper proposes a precise prediction method combining SBAS-InSAR and the DBO–CNN–LSTM (dung beetle optimization–convolutional neural network–long short-term memory) model for closed mine surface deformation. Utilizing SBAS-InSAR for high-precision data acquisition and DBO–CNN–LSTM for complex feature learning and time series analysis, the method provides accurate long-term predictions. Based on five years of Huainan mining area data, post-closure deformation trends vary dynamically. The Lizuizi Mine transitioned from subsidence to uplift, while the Xinzhuangzi Mine experienced significant uplift, influenced by geological conditions, closure time, and subsequent activities. The DBO–CNN–LSTM model outperformed other neural networks. For the Xieyi Mine, 26,165 corresponding points between model predictions and SBAS-InSAR results showed significant correlation, with R2 of 0.91 and correlation coefficient of 0.95 on March 31, 2022, and R2 of 0.92 and correlation coefficient of 0.97 on February 6, 2023. This paper reveals surface deformation mechanisms in closed mines, including groundwater level changes, and provides a scientific basis for ecological restoration, development, geological disaster warning, and improved mining area management and policy formulation.