EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

A BZOA-FS-Based Prediction Model for Rockburst Hazard Levels

Yan-Bin Wang, Kang-Ping Wang

Applied Sciences · 25.07.2026 · 16 (15) · 7452

DOI: 10.3390/app16157452 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Applied Sciences, online erschienen am 25.07.2026. Verfasst von Yan-Bin Wang und 1 weiteren Autor:innen. Die Arbeit nennt 29 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Feature selection plays an important role in improving classification performance for high-dimensional datasets. To address the limitations of the original Zebra Optimization Algorithm (ZOA) in binary optimization problems, a Binary Zebra Optimization Algorithm for Feature Selection (BZOA-FS) is proposed. The proposed algorithm is integrated with a K-nearest neighbor (KNN) classifier to construct a prediction model for rockburst hazard levels. The proposed framework consists of three main components. First, the feature importance guidance mechanism is designed, and the feature importance score is constructed and run through the initialization, foraging stage, and defense stage of BZOA-FS. The binary zebra foraging operator and binary defense operator are designed to complete the transformation from continuous optimization to discrete optimization, and the adaptive strategy is introduced to make adaptive dynamic selection according to the evolution stage and population state. It should be emphasized that the basic binary operations used in BZOA-FS, such as XOR-based difference representation and probabilistic bit flipping, are common components in binary metaheuristic optimization. The main contribution of this study is its ZOA-specific reconstruction and integration with feature-importance guidance and diversity-driven adaptive defense. Second, four UCI data sets are selected for experiments to compare the performance of GA-FS, WOA-FS, ZOA-FS, BPSO-FS, RF-FS, and BZOA-FS algorithms on the two classifiers. The proposed BZOA-FS demonstrates competitive or superior performance compared with the selected benchmark algorithms on the investigated datasets. This verifies the feasibility of the improved strategy proposed in this study. Third, the algorithm is applied to the example of rockburst, and the prediction model of rockburst hazard level based on BZOA-FS is constructed. The results show that the performance of the BZOA-FS model in the evaluation indexes is better than other models, which demonstrates the potential of the proposed BZOA-FS model in the prediction of rockburst hazard level.

Zitieren

APA

Wang, Y., & Wang, K. (2026). A BZOA-FS-Based Prediction Model for Rockburst Hazard Levels. Applied Sciences, 16(15), 7452. https://doi.org/10.3390/app16157452

Vancouver

Wang Y, Wang K. A BZOA-FS-Based Prediction Model for Rockburst Hazard Levels. Applied Sciences. 2026;16(15):7452. doi:10.3390/app16157452

BibTeX
@article{Wang2026,
  author = {Wang, Yan-Bin and Wang, Kang-Ping},
  title = {A BZOA-FS-Based Prediction Model for Rockburst Hazard Levels},
  journal = {Applied Sciences},
  year = {2026},
  volume = {16},
  number = {15},
  pages = {7452},
  doi = {10.3390/app16157452}
}
Kostenlos starten →