Intelligent rockburst level prediction model based on swarm intelligence optimization and multi-strategy learner soft voting hybrid ensemble
Abstract Rockbursts are highly destructive geological events that pose serious risks to the safety of underground engineering projects, including tunnels, mines, and other subterranean structures. Accurate prediction of rockburst occurrence and intensity is crucial for preventing and mitigating thei...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Springer
2025-01-01
|
| Series: | Geomechanics and Geophysics for Geo-Energy and Geo-Resources |
| Subjects: | |
| Online Access: | https://doi.org/10.1007/s40948-024-00931-1 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Abstract Rockbursts are highly destructive geological events that pose serious risks to the safety of underground engineering projects, including tunnels, mines, and other subterranean structures. Accurate prediction of rockburst occurrence and intensity is crucial for preventing and mitigating their potentially catastrophic impacts. Such predictions are vital not only for ensuring the safety of workers and infrastructure but also for optimizing construction and operational strategies in underground environments. This research is developing a smart model to predict rockburst levels by combining advanced swarm intelligence with a hybrid ensemble method that uses multiple strategies and classifiers for soft voting. By collecting and analyzing 287 rockburst cases from diverse geological settings around the world, a comprehensive and representative database was constructed. This database was subsequently subjected to in-depth statistical and correlation analyses to identify key patterns and relationships. The data preprocessing method proposed in this study, based on an improved version of the Student t-SNE algorithm, effectively reduced the negative impact of data noise on model performance, enhancing the reliability of predictions. Furthermore, the prediction model developed in this study not only demonstrated exceptional prediction accuracy on the test set but also provided valuable insights into key geological parameters influencing rockbursts through rigorous sensitivity analysis. The proposed hybrid ensemble model greatly improves generalization and interpretability over traditional single models. It offers an efficient, data-driven approach to rockburst prediction, providing a strong scientific foundation for safety management and risk mitigation in underground engineering worldwide. |
|---|---|
| ISSN: | 2363-8419 2363-8427 |