Rockburst intensity grading prediction based on the LOF-ENN-KNN model
Abstract Rockburst is a typical dynamic disaster in deep underground engineering, and its accurate prediction is of great significance to ensure the safety of engineering. Aiming at the key problems in rockburst prediction, such as insufficient analysis of nonlinear correlation characteristics, sign...
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2025-08-01
|
| Series: | Scientific Reports |
| Subjects: | |
| Online Access: | https://doi.org/10.1038/s41598-025-15603-7 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849342997324365824 |
|---|---|
| author | Haoran Ge Jiyong Zhang Congbo Ma Kai Hui Yicong Li Ziniu Wu |
| author_facet | Haoran Ge Jiyong Zhang Congbo Ma Kai Hui Yicong Li Ziniu Wu |
| author_sort | Haoran Ge |
| collection | DOAJ |
| description | Abstract Rockburst is a typical dynamic disaster in deep underground engineering, and its accurate prediction is of great significance to ensure the safety of engineering. Aiming at the key problems in rockburst prediction, such as insufficient analysis of nonlinear correlation characteristics, significant discreteness of small sample data and limited generalization ability of traditional models, this study innovatively constructs a LOF-ENN-KNN coupling prediction model. The model achieves a breakthrough in prediction performance through a three-level progressive processing architecture: noise samples are effectively eliminated by LOF (Local Outlier Factor), category distribution is optimized by ENN (Edited Nearest Neighbour), and high-precision prediction is finally achieved by KNN(K-Nearest Neighbors).In the rockburst database composed of 299 sets of measured data in the world, the LOF-ENN-KNN model shows significant advantages: its overall accuracy rate is 98.93%,which has an overwhelming advantage over the traditional model. Through the comparison and exploration of different sampling methods and different combinations of LOF algorithm, single resampling technology (SMOTE, ADASYN) or simple technology superposition (LOF-SMOTE, LOF-ADASYN) is easy to introduce over-fitting or negative coupling effect, while LOF-ENN-KNN significantly improves the robustness and generalization ability of the model through modular design. In addition, compared with LR, SVM, DT, NBs and DNN intelligent algorithms, LOF-ENN-KNN has significant advantages. Through engineering examples, it is further confirmed that high-precision prediction ability can still be preserved in complex nonlinear parameters, which provides an efficient and reliable technical scheme for rockburst warning in deep engineering. |
| format | Article |
| id | doaj-art-60ea24c784d14ede8b702c3759331aae |
| institution | Kabale University |
| issn | 2045-2322 |
| language | English |
| publishDate | 2025-08-01 |
| publisher | Nature Portfolio |
| record_format | Article |
| series | Scientific Reports |
| spelling | doaj-art-60ea24c784d14ede8b702c3759331aae2025-08-20T03:43:11ZengNature PortfolioScientific Reports2045-23222025-08-0115111610.1038/s41598-025-15603-7Rockburst intensity grading prediction based on the LOF-ENN-KNN modelHaoran Ge0Jiyong Zhang1Congbo Ma2Kai Hui3Yicong Li4Ziniu Wu5School of Resources and Environmental Engineering, Jiangxi University of Science and TechnologySchool of Resources and Environmental Engineering, Jiangxi University of Science and TechnologySchool of Resources and Environmental Engineering, Jiangxi University of Science and TechnologySchool of Resources and Environmental Engineering, Jiangxi University of Science and TechnologySchool of Architecture and Design, Jiangxi University of Science and TechnologySchool of Resources and Environmental Engineering, Jiangxi University of Science and TechnologyAbstract Rockburst is a typical dynamic disaster in deep underground engineering, and its accurate prediction is of great significance to ensure the safety of engineering. Aiming at the key problems in rockburst prediction, such as insufficient analysis of nonlinear correlation characteristics, significant discreteness of small sample data and limited generalization ability of traditional models, this study innovatively constructs a LOF-ENN-KNN coupling prediction model. The model achieves a breakthrough in prediction performance through a three-level progressive processing architecture: noise samples are effectively eliminated by LOF (Local Outlier Factor), category distribution is optimized by ENN (Edited Nearest Neighbour), and high-precision prediction is finally achieved by KNN(K-Nearest Neighbors).In the rockburst database composed of 299 sets of measured data in the world, the LOF-ENN-KNN model shows significant advantages: its overall accuracy rate is 98.93%,which has an overwhelming advantage over the traditional model. Through the comparison and exploration of different sampling methods and different combinations of LOF algorithm, single resampling technology (SMOTE, ADASYN) or simple technology superposition (LOF-SMOTE, LOF-ADASYN) is easy to introduce over-fitting or negative coupling effect, while LOF-ENN-KNN significantly improves the robustness and generalization ability of the model through modular design. In addition, compared with LR, SVM, DT, NBs and DNN intelligent algorithms, LOF-ENN-KNN has significant advantages. Through engineering examples, it is further confirmed that high-precision prediction ability can still be preserved in complex nonlinear parameters, which provides an efficient and reliable technical scheme for rockburst warning in deep engineering.https://doi.org/10.1038/s41598-025-15603-7Rockburst predictionEliminating outliersUndersamplingKNN algorithmMachine learning |
| spellingShingle | Haoran Ge Jiyong Zhang Congbo Ma Kai Hui Yicong Li Ziniu Wu Rockburst intensity grading prediction based on the LOF-ENN-KNN model Scientific Reports Rockburst prediction Eliminating outliers Undersampling KNN algorithm Machine learning |
| title | Rockburst intensity grading prediction based on the LOF-ENN-KNN model |
| title_full | Rockburst intensity grading prediction based on the LOF-ENN-KNN model |
| title_fullStr | Rockburst intensity grading prediction based on the LOF-ENN-KNN model |
| title_full_unstemmed | Rockburst intensity grading prediction based on the LOF-ENN-KNN model |
| title_short | Rockburst intensity grading prediction based on the LOF-ENN-KNN model |
| title_sort | rockburst intensity grading prediction based on the lof enn knn model |
| topic | Rockburst prediction Eliminating outliers Undersampling KNN algorithm Machine learning |
| url | https://doi.org/10.1038/s41598-025-15603-7 |
| work_keys_str_mv | AT haorange rockburstintensitygradingpredictionbasedonthelofennknnmodel AT jiyongzhang rockburstintensitygradingpredictionbasedonthelofennknnmodel AT congboma rockburstintensitygradingpredictionbasedonthelofennknnmodel AT kaihui rockburstintensitygradingpredictionbasedonthelofennknnmodel AT yicongli rockburstintensitygradingpredictionbasedonthelofennknnmodel AT ziniuwu rockburstintensitygradingpredictionbasedonthelofennknnmodel |