Scholarlink

<p>During the global transition to sustainable energy, coal still maintains a significant position in the energy mix. Therefore, mitigating the environmental impact of coal mining becomes crucial. The coal rock gas desorption process, influenced by the coal’s pore structure, gas pressure, and stress conditions, constitutes a complex nonlinear dynamic system involving multi-physics–field coupling. Precisely forecasting the volume of coal rock gas desorption holds great significance for optimizing gas drainage strategies, preventing gas disasters, and reducing environmental impact from a carbon–hydrocarbon exploration and production perspective. This study delved into the gas distribution patterns in coal powder samples from Guizhou’s Faer (FE) and Qinglong (QL) mines amidst mining activities. A field-emission scanning electron microscope was employed to observe pore structures and carried out methane desorption experiments. These experiments aimed to clarify the mechanism of how pore structures affect gas desorption, thereby uncovering the underlying nonlinear relationships in the coal rock gas desorption process. To tackle the gas desorption prediction challenge in carbon–hydrocarbon exploration and production, study this integrated multiple methods such as particle swarm optimization (PSO), chaos game optimization (CGO), attention mechanisms (AMs), long short-term memory (LSTM) network, and gated recurrent unit (GRU) models. The results showed that FE samples had desorption peaks in stages II and VI, while QL samples peaked at 9.7509&#xa0;cm<sup>3</sup>/g in stage VI. When comparing the prediction models, the PSO-GRU model demonstrated the highest accuracy (R2 = 0.9994) and lowest error, surpassing the CGO-LSTM model (R2 = 0.9913). Meanwhile, nonlinear curve fitting models achieved good fitting of desorption data with R2 exceeding 0.8825. SHapley Additive exPlanations analysis revealed that stages III and V significantly influenced stage VI desorption. This research not only revealed the intrinsic mechanisms and laws of coalbed methane desorption during coal rock mining but also provides a novel intelligent approach for the sustainable development of the coal industry. It helps address the nonlinear processes and complex phenomena in coal rock mining systems, promoting the development of coal rock mining toward a safer, cleaner, and more efficient direction.</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

Nonlinear Characteristics and Intelligent Prediction Method of Gas Desorption in Mining Coal Rock

  • 期刊论文

  • 发表日期：

    2026年3月3日

查看全文
  • Junjie Cai,
  • Xijian Li,
  • Shoukun Chen

摘要

During the global transition to sustainable energy, coal still maintains a significant position in the energy mix. Therefore, mitigating the environmental impact of coal mining becomes crucial. The coal rock gas desorption process, influenced by the coal’s pore structure, gas pressure, and stress conditions, constitutes a complex nonlinear dynamic system involving multi-physics–field coupling. Precisely forecasting the volume of coal rock gas desorption holds great significance for optimizing gas drainage strategies, preventing gas disasters, and reducing environmental impact from a carbon–hydrocarbon exploration and production perspective. This study delved into the gas distribution patterns in coal powder samples from Guizhou’s Faer (FE) and Qinglong (QL) mines amidst mining activities. A field-emission scanning electron microscope was employed to observe pore structures and carried out methane desorption experiments. These experiments aimed to clarify the mechanism of how pore structures affect gas desorption, thereby uncovering the underlying nonlinear relationships in the coal rock gas desorption process. To tackle the gas desorption prediction challenge in carbon–hydrocarbon exploration and production, study this integrated multiple methods such as particle swarm optimization (PSO), chaos game optimization (CGO), attention mechanisms (AMs), long short-term memory (LSTM) network, and gated recurrent unit (GRU) models. The results showed that FE samples had desorption peaks in stages II and VI, while QL samples peaked at 9.7509 cm3/g in stage VI. When comparing the prediction models, the PSO-GRU model demonstrated the highest accuracy (R2 = 0.9994) and lowest error, surpassing the CGO-LSTM model (R2 = 0.9913). Meanwhile, nonlinear curve fitting models achieved good fitting of desorption data with R2 exceeding 0.8825. SHapley Additive exPlanations analysis revealed that stages III and V significantly influenced stage VI desorption. This research not only revealed the intrinsic mechanisms and laws of coalbed methane desorption during coal rock mining but also provides a novel intelligent approach for the sustainable development of the coal industry. It helps address the nonlinear processes and complex phenomena in coal rock mining systems, promoting the development of coal rock mining toward a safer, cleaner, and more efficient direction.