Hazard evaluation of goaf based on DBO algorithm coupled with BP neural network.

Area of Science:

• Geological Engineering
• Artificial Intelligence
• Optimization Algorithms

Background:

• Goaf formation poses significant economic and developmental challenges in China due to extensive mineral resource exploitation.
• Effective assessment and management of goafs are crucial for sustainable development and mitigating associated risks.

Purpose of the Study:

• To introduce and evaluate the efficacy of the Dung Beetle Optimizer (DBO) algorithm for goaf hazard assessment.
• To compare the performance of the DBO-BP model against other heuristic algorithms coupled with back-propagation neural networks.

Main Methods:

• Development of a DBO-BP model integrating the Dung Beetle Optimizer with a back-propagation neural network.
• Comparative analysis with Particle Swarm Optimization (PSO)-BP, Whale Optimization Algorithm (WOA)-BP, and Sparrow Search Algorithm (SSA)-BP models.
• Application of the models to evaluate goaf hazards, including a case study on a tungsten mine goaf.

Main Results:

• The DBO-BP model achieved the highest accuracy on both training (at least 2.7% higher) and test datasets.
• The DBO-BP model demonstrated superior effectiveness and stability compared to PSO-BP, WOA-BP, and SSA-BP models.
• Successful validation of the DBO-BP model's practicability in a real-world tungsten mine goaf hazard assessment.

Conclusions:

• The DBO-BP model presents a highly accurate and stable approach for goaf hazard assessment.
• This research provides a valuable reference for addressing nonlinear engineering problems in mining and resource management.
• The study highlights the potential of the DBO algorithm in environmental and geological risk assessment.