ESTABLISHING A NEW ROCKBURST CLASSIFICATION SYSTEM USING BAYESIAN NETWORK

Abstract

Rockburst is a phenomenon commonly described as a sudden and violent event in underground mines that result in significant damages to underground excavations, equipment. It also threatens the safety of the mine workers and the profitability of the operations. This clearly demonstrates the importance of understanding the rockburst mechanisms. The literature reveals that despite the advance in predicting rockburst, a reliable prediction of the phenomenon is till posing problems. Therefore, this study aims to establish a new rockburst classification system that can be used as a reliable tool for rockburst intensity evaluation. The research methodology relies on the Bayesian Network (BN) approach and actual rockburst records from various mines and tunneling projects across the world. Two main databases were considered: first contains rock mass parameters and seismic magnitudes while the third contains only rock mass parameters. In addition, a second database generated using Monte Carlo simulation technique, was used in order to increase the data size of the first database. The input parameters of the first and second databases included the stress conditions (E1), ground support capacity (E2), span (E3), geology (E4), peak particle velocity (PPV) and the output is defined as Rockburst Damage Scale (RDS). For the third database, the input parameters included the tangential stress (𝛿𝜃), compressive stress (𝛿𝑐), tensile stress (𝛿𝑡), strain energy (𝑊𝑒𝑡), ratio of normal stress to compressive stress (𝛿𝜃/𝛿𝑐) and ratio of compressive stress to tangential stress (𝛿𝑐/𝛿𝑡). Three BN structures were constructed through Netica Software and consequently were evaluated with independent dataset. The results indicate that the classification accuracies vary between 70% and 82% depending on the database. Based on the obtained results, thresholds for rockburst intensity are proposed. The final results suggest that the newly proposed rockburst classification could contribute to a better rockburst management in mining and tunneling projects.