摘要
传统的冲击地压危险性预测方法易受非稳定弹性冲击影响,导致预测精度降低。为此,文章研究了双层Stacking算法在冲击地压危险性预测中的应用。该方法通过冲击显现模型计算三向应力弹性冲击,划分非稳定危险状态预测分区,结合随机森林和XGBoost算法,在双层Stacking框架下推导危险发生条件。此外,引入注意力机制动态调整模型输出权重并利用莫兰指数估算预测结果的空间自相关性,以提升预测精度。实验结果表明,1#在3.01~3.2 s出现危险性信号,预测与实际一致,精度高,该方法在均方误差(Mean Squared Error, MSE)、平均绝对误差(Mean Absolute Error, MAE)和准确率方面均优于传统方法,预测精度显著提高,能够为冲击地压的预警提供可靠依据。
In traditional methods for predicting the risk of rockburst,the prediction accuracy is easily reduced due to the influence of unstable elastic impacts.To this end,the article proposes the application research of the double-layer Stacking algorithm in the prediction of rockburst risk.This method calculates the three-dimensional stress elastic impact through the impact manifestation model,divides the unstable dangerous state prediction zone,and combines the random forest and XGBoost algorithm to derive the dangerous occurrence conditions under the double-layer Stacking framework.In addition,an attention mechanism is introduced to dynamically adjust the output weights of the model,and the Moran index is used to estimate the spatial autocorrelation of the prediction results,in order to improve the prediction accuracy.The experimental results show that danger signals appear at 3.01~3.2 s for 1#,and the prediction is consistent with the actual situation,with high accuracy.This method is superior to traditional methods in terms of mean square error(MSE),mean absolute error(MAE),and accuracy,and the prediction accuracy is significantly improved.It can provide reliable basis for the warning of rockburst.
作者
孙甜甜
杨蒙蒙
彭思雨
SUN Tiantian;YANG Mengmeng;PENG Siyu(Department of Information Engineering,Changji University,Changji 831100,China)
出处
《无线互联科技》
2025年第17期15-18,共4页
Wireless Internet Science and Technology
基金
昌吉学院一般项目(自然科学),项目名称:基于BP神经网络的冲击地压预测方法研究,项目编号:No.KYLK032。
