Adaptive dynamic prediction model of mining subsidence aided by measured data

Area of Science:

• Geotechnical Engineering
• Mining Engineering
• Environmental Science

Background:

• Underground mining causes surface subsidence, impacting the environment and structures.
• Accurate prediction of subsidence is vital for effective maintenance and remediation planning.
• Traditional models struggle with the dynamic nature of mining, leading to prediction inaccuracies.

Purpose of the Study:

• To develop an adaptive prediction model for dynamic underground mining subsidence.
• To improve the accuracy of subsidence predictions compared to traditional methods.
• To provide technical support for the maintenance and remediation of structures affected by mining.

Main Methods:

• Developed a data-driven adaptive prediction model using historical surface subsidence measurements.
• Derived optimal model parameters for different historical mining periods.
• Analyzed parameter trends to dynamically adjust predictions for future periods.
• Validated the model using an engineering case study.

Main Results:

• The adaptive model achieved an average relative RMSE of 4.3% for dynamic subsidence prediction.
• This represents a significant improvement over traditional models, which had an average RMSE of 9.1%.
• Estimated parameter values from the new method closely matched optimal values derived from historical data.

Conclusions:

• The adaptive prediction model offers high-precision forecasting of surface dynamic subsidence caused by underground mining.
• The model's accuracy enhances decision-making for maintenance strategies, timing, and volume assessments.
• This approach provides robust technical support for mitigating adverse effects of mining-induced subsidence.