DOI: 10.11799/ce202511017
Construction and application of a multi-parameter intelligent comprehensive prediction model for rockburst
GUAN Xinbang
ZHAO Shankun
WANG Yin
LI Yunpeng
KONG Linghai
LI Yizhe
LIU Junhong
Abstract：To construct an intelligent early warning model for rockburst that integrates"mechanism constraints and data-driven"approaches,enabling integrated prediction of hazard"location-time-intensity",a multi-parameter index system was established based on multi-source heterogeneous data.This system integrates real-time monitoring,mining techniques,coal-rock strength,and geological conditions,with temporal and spatial parameters introduced to build a spatio-temporally coupled warning module.Data normalization,K-means clustering imputation,and principal component analysis were applied for feature processing.Predictive models were developed using L2-regularized multiple linear regression,BP neural network,and RNN-GRU neural network,and their performances were compared.The results indicate that the L2-regularized model can interpret factor weights,supporting mechanistic studies and single-indicator warnings.The RNN-GRU model significantly enhances the ability to capture temporal dynamic features through gated recurrent units.Combining with monitoring point location encoding and improved K-means clustering,it enables spatio-temporal distribution prediction of rockburst hazards.In field applications at the Hujierte mining area,the model's predicted magnitudes were close to actual events,with time errors of less than 2.5 hours,location errors of less than 50 meters,and an overall accuracy of 85%,providing a reliable method for intelligent rockburst early warning.
Keywords：rockburstmonitoring and early warninglinear regressionneural networkhazard prediction
Publication Date：2025-11-20
Online Publishing Date：2026-01-15(First online date of this platform, not the publication date of the document)
Pages：10( 131-140 )