Research on rock burst early warning technology based on NOA-CNN
Wu Guandong
Wen Yingyuan
Li Huwei
Cao Anye
Guo Wenhao
Guo Jisen
Abstract：As a typical geological condition that induces rock burst,hard roofs are characterized by high strength,high bearing capacity,high integrity,and low jointing,posing a serious threat to safe and efficient production in mines.To accurately predict rock burst disas-ters under this geological condition,a NOA-CNN rock burst early warning model was constructed.The NOA algorithm was utilized to optimize three hyperparameters of CNN:learning rate,batch size,and regularization coefficient,with their optimal values being 0.016 3,175,and 0.023 6,respectively.When training the NOA-CNN model,the accuracy rates of its training set and test set were 98.12％and 98.62％respectively.When training the CNN model without hyperparameter optimization,the prediction accuracy rates of both its train-ing set and test set were lower than those of NOA-CNN model,and the false negative rate and false positive rate of CNN model in-creased by 94.4％and 0.2％respectively.Research indicates that the NOA-CNN model excels in mining deeper potential features from microseismic data,delivering superior early warning performance,and is more suitable for predicting rock burst,thereby ensuring coal mine safety.
Keywords：rock burstearly warning modelhard roofcoal mine safety
Publication Date：2025-12-30
Online Publishing Date：2026-01-31(First online date of this platform, not the publication date of the document)
Pages：5( 202-206 )