Roof fall hazard assessment with the use of artificial neural network

Abstract

The paper presents the results of roof fall hazard analysis for the copper mines in Poland. At first, RMR and RFRI systems have been checked as a tool for roof fall hazard assessment. It was proved that rock mass deteriorates in time because of mining operation at the site, hence the roof fall phenomena analysis needs a new approach. In the next step of the analysis the key factors of roof deterioration were determined, and divided into four groups: geological, mining, technical and monitoring. It was demonstrated that only regular roof monitoring can ensure the proper rock fall hazard assessment. Subsequently, the artificial neural network was created and several dozen simulations were conducted. As a result, two dimensionless indices were developed. The first one shows predisposition of a part of the rock mass to destruction, dislocation and deformation - CRFP - Coefficient of Roof Fall – Predisposition and the second one represents predisposition and possibility of maintaining of the working - CRFM- Coefficient of Roof Fall - Maintenance. The research on roof fall details and the running workings’ roof observations allowed for categorization of the values of both indices, and providing that way the information about roof stability CRFP, and the information about necessary measures and solutions for the supervisory staff in regard of monitoring of the roof rocks CRFM. Both indices allow for a reliable roof fall hazard assessment by comprehensively combining information from monitoring, observations and investigations.