Enterprise AI Analysis
Performance evaluation and interpretation of hybrid models based on light gradient boosting machine to predict coal mining subsidence
Accurate prediction of coal mining subsidence (CMS) is pivotal for mining design, environmental preservation, safety assurance, and the formulation of preventive measures in mining areas. This study introduces an effective hybrid prediction model for coal mining subsidence based on light gradient boosting machine (LightGBM) and uses Shapley Addi-tive Explanations (SHAP) method to reveal and explain the contribution mechanism and interaction of factors affecting mining subsidence. This study collected a dataset of 163 mining subsidence cases (covering 12 features such as coal seams, rock layers, and mining conditions) to model mining subsidence prediction, and evaluated the performance of the model through multiple performance indicators. The results indicate that the hybrid prediction model developed in this paper demonstrates remarkable performance on the test set, with HGS-LightGBM standing out, achieving the coefficient of determination (R2) of 0.9589, while the single LightGBM achieves an R2 of 0.925. Finally, apply model interpreta-tion techniques to analyze the impact of input features on mining subsidence, and explain the prediction principles and decision-making process of the model. The analysis reveals that the thickness of coal seam (m) is the most influential parameter for CMS. Furthermore, a targeted interaction analysis on key parameters was conducted to clarify the impact mechanisms of each influencing factor. In summary, the model established in this study has excellent performance and exhibits significant interpretability and transparency.
This study presents a novel hybrid LightGBM model for predicting coal mining subsidence (CMS), achieving superior accuracy and interpretability compared to traditional methods. By integrating intelligent optimization algorithms and advanced explanation techniques like SHAP, the model offers unprecedented insights into the complex factors influencing subsidence. Our analysis reveals that coal seam thickness (m) is the most critical parameter, with significant non-linear impacts. This advanced predictive capability is crucial for enhanced safety, environmental protection, and optimized mining design, leading to more sustainable and responsible resource extraction. Enterprises can leverage these insights to proactively manage risks, refine operational strategies, and ensure regulatory compliance.
0
Prediction Accuracy (R²)
0
Root Mean Square Error (RMSE)
0
Mean Absolute Error (MAE)
0
Variance Accounted For (VAF)
Deep Analysis & Enterprise Applications
Select a topic to dive deeper, then explore the specific findings from the research, rebuilt as interactive, enterprise-focused modules.
Addressing Complex Mining Challenges with AI
Coal mining subsidence (CMS) poses significant challenges to infrastructure, environment, and safety. Traditional prediction methods like numerical modeling or physical simulations often simplify complex geological conditions, limiting their accuracy and applicability. This research addresses these limitations by introducing advanced machine learning, specifically a hybrid LightGBM model, to provide more precise and interpretable predictions. The goal is to enhance proactive mitigation strategies and ensure sustainable mining practices.
Leveraging LightGBM for Efficient Prediction
Light Gradient Boosting Machine (LightGBM) is a highly efficient and scalable machine learning framework built on Gradient Boosting Decision Trees (GBDT). It is particularly adept at processing large datasets quickly and accurately. Key features include its gradient-based one-sided sampling (GOSS) for improved training efficiency and a leaf-wise growth strategy that optimizes node splitting for maximum gain, preventing overfitting and maintaining high accuracy.
Boosting Accuracy with Swarm Intelligence
To further enhance LightGBM's predictive capabilities, this study integrates powerful swarm intelligence optimization algorithms for hyperparameter tuning. Algorithms like Dwarf Mongoose Optimization (DMOA), Fox Optimization Algorithm (FOX), Giant Trevally Optimizer (GTO), Grey Wolf Optimizer (GWO), Hunger Games Search (HGS), and Sparrow Search Algorithm (SSA) are employed. These intelligent algorithms guide the model to find optimal hyperparameter configurations, thereby boosting model reliability, precision, and generalization performance by iteratively searching for the best solution in the parameter space.
Enterprise Process Flow
Data Sets (163 cases, 12 features)
→
Initial LightGBM Hyperparameters
→
LightGBM Model for Data Mining (Leaf-wise Growth, Histogram)
→
Intelligent Optimization Algorithms (DMOA, FOX, GTO, GWO, HGS, SSA)
→
Best LightGBM Model (Optimized Hyperparameters)
→
Satisfy Terminal Condition
→
Performance Evaluation (RMSE, R², MAE, VAF)
Comprehensive Dataset & Robust Methodology
The study utilized a dataset of 163 coal mining subsidence cases, each characterized by 12 influencing factors (e.g., coal seam thickness, burial depth, dip angle, rock properties) and a single label representing maximum surface subsidence (w). This comprehensive dataset was split 80% for training and 20% for testing. A five-fold cross-validation strategy was applied during training and optimization to ensure robust model performance and generalization capabilities.
| Mode | RMSE | R² | MAE | VAF |
|---|---|---|---|---|
| DMOA-LightGBM | 0.3358 | 0.9526 | 0.1562 | 95.6248 |
| FOX-LightGBM | 0.3227 | 0.9562 | 0.1468 | 96.009 |
| GTO-LightGBM | 0.3222 | 0.9563 | 0.1972 | 95.9307 |
| GWO-LightGBM | 0.3292 | 0.9580 | 0.1734 | 96.1103 |
| HGS-LightGBM | 0.3126 | 0.9589 | 0.1651 | 96.1826 |
| SSA-LightGBM | 0.3159 | 0.958 | 0.1734 | 96.1103 |
0.9589
Highest R² Achieved by HGS-LightGBM Hybrid Model
Outstanding Predictive Accuracy
The hybrid prediction models, particularly HGS-LightGBM, demonstrated superior performance on the test set, achieving an R² of 0.9589, RMSE of 0.3126, MAE of 0.1651, and VAF of 96.1826. This represents a marked improvement over the standalone LightGBM model (R²=0.925), affirming the effectiveness of integrating intelligent optimization algorithms for enhanced predictive accuracy and generalization.
Key Influencing Factors: Coal Seam Thickness & Settlement Coefficient
Utilizing Shapley Additive Explanations (SHAP) and Local Interpretable Model-agnostic Explanations (LIME), this study provides deep insights into the model's decision-making process. The thickness of coal seam (m) and the mining area settlement coefficient (q) emerged as the most influential factors for CMS. Notably, the impact of 'm' exhibits strong nonlinearity: its influence on CMS is small when m < 2, significantly increases when 2 < m < 4, and reaches a maximum range for m > 4. Understanding these complex, non-linear interactions is crucial for tailoring effective prevention and control measures in mining operations.
The analysis also reveals complex interactions between 'm' and other variables like 'q', 'h' (unconsolidated strata thickness), 'n' (mining intensity), and 'H' (burial depth). For example, when 'm' is small, the impact on overlying rock layers is minimal. However, as 'm' increases, the potential for substantial detrimental impacts rises, necessitating refined mining methodologies and enhanced rock strata observation.
Enhanced Interpretability for Informed Decisions
The application of SHAP and LIME techniques allows for a transparent understanding of how individual features contribute to the CMS prediction. This interpretability is vital for identifying critical influencing factors and their interaction mechanisms, enabling mining engineers and stakeholders to make informed decisions, optimize mining designs, and develop targeted mitigation strategies with greater confidence.
Strategic Implications for Sustainable Mining
The robust and interpretable hybrid model offers significant strategic implications for the mining industry. By accurately predicting CMS and identifying key influencing factors, enterprises can:
- Optimize Mining Design: Adjust coal seam extraction plans, pillar reserves, and goaf filling strategies to minimize subsidence.
- Enhance Safety & Environment: Proactively implement measures to protect surface infrastructure and ecosystems, reducing risks of building collapse, soil erosion, and groundwater disturbance.
- Improve Decision-Making: Leverage data-driven insights to develop targeted interventions and comply with environmental regulations.
- Drive Sustainability: Promote more responsible and sustainable coal resource extraction by understanding and mitigating its impacts.
Calculate Your Potential AI Impact
Estimate the ROI of implementing advanced AI solutions like the one described in this research for your enterprise.
Annual Savings
$0
Hours Reclaimed Annually
0
Your AI Implementation Roadmap
A typical phased approach to integrate advanced AI solutions into your enterprise operations.
Phase 1: Discovery & Strategy
Comprehensive assessment of current challenges, data infrastructure, and strategic objectives. Define project scope, key performance indicators (KPIs), and a tailored AI strategy to align with business goals.
Phase 2: Data Engineering & Model Development
Collect, clean, and preprocess relevant data. Design and develop custom AI models, leveraging techniques like LightGBM and intelligent optimization, ensuring high accuracy and interpretability.
Phase 3: Integration & Deployment
Seamlessly integrate the AI solution into existing enterprise systems and workflows. Conduct rigorous testing, validation, and pilot deployments to ensure stability and performance in a live environment.
Phase 4: Monitoring, Optimization & Scaling
Establish continuous monitoring for model performance and data drift. Implement mechanisms for ongoing model retraining and optimization. Scale the solution across departments or operations to maximize enterprise-wide impact.
Ready to Transform Your Enterprise with AI?
Book a personalized consultation with our AI experts to explore how these advanced insights can drive efficiency, mitigate risks, and unlock new opportunities for your business.
Ready to Get Started?
Book Your Free Consultation.
Let's Discuss Your AI Strategy!
Lets Discuss Your Needs
Select a Date
Sun
Mon
Tue
Wed
Thu
Fri
Sat