A prototype-based rockburst types and risk prediction algorthim considering intra-class variance and inter-class distance of microseismic data

Xiufeng Zhang¹Guoying Li¹Yang Chen¹Hao Wang¹Haikuan Zhang^2*Haitao Li²Weisheng Du²Xiao Li²Xuewei Xu²Yuze He²

• ¹Shandong Energy Group CO.,Ltd., Jinan, Shandong Province, China
• ²Chinese Institute of Coal Science, Beijing, China

The prediction and classification of rock burst risk based on microseismic data is the premise of preventing rock burst during deep mine excavation. By reviewing previous studies, this paper finds two problems that hinder the prediction of rock burst: 1) there is a lack of research on the distribution features of monitoring data on the main controlling factors of rockburst; 2) there is no research on the intra-class variance and inter-class gap of microseismic data. Based on the typical rockburst risk events, the quantitative information model of geology and mining is constructed. The relationship between the spatial-temporal distribution characteristics of microseismic data before rockburst, the main controlling factors of rockburst is studied. The results show that the distribution features may be different for the same type MS and rockburst events; and different type events may show similar distribution features. Therefore, based on the quantitative study of the relationship between the performance of deep learning prediction algorithm and rock burst prediction vector, a rockburst risk and type prediction algorithm based on CNN-GRU model with prototype based prediction is proposed. The CNN-GRU model can produce the prediction vectors by fusing implicit and explicit information extracted from original MS data and early warning indicators. The cross-entropy loss, vector-prototype contrastive loss and vector-prototype contrastive loss are proposed proposed to automatically control the intra-class variance and inter-class gap of prediction vectors belong to different rockburst risks and types. A large number of experiments show that the performance of the proposed CNN-GRU model with prototype based prediction is superior to other algorithms in the prediction of rockburst risks and types based on MS data.