Study On The Prediction Of Rock Burst Hazard Level Based On RS-MPSO-LSSVM

Rock burst has been parallel to the mine gas, coal mine fire, coal mine get the most serious dynamic disaster of the world's mine. Rock burst occurs when the elastic properties of rock mass is greater than the plastic properties,the remaining part propagates around in the form of shockwave propagates around. The mechanism of the impact is not only complex, but also has many influent factors. it has considerable much difficulties with the applying of conventional mathematics, mechanics theory to build rock burst prediction model,and has become an urgent problem to solve and research in the field of international mining engineering and rock soil mechanics areas. In order to establish close contact of the actual engineering and get convictional method, Researching testing technology is of great importance and must be carried out, and the key research of the field testing technology is how to take measures to use test data predict the rock burst class.This paper takes the observation dynamic phenomena of ZhangJi mine in Xuzhou Mining Bureau observation and experience of impact accident data in recent years. Puting forward the algorithm of rough set theory and support vector machine improved by modified least square particle swarm (RS-MPSO-LSSVM) which combines the study of historical data to forecast and research.By comparing of BP neural network and RS-PSO-LSSVM, the improved least square particle swarm support vector machine based on rough set and (RS-MPSO-LSSVM) the rock burst risk rating prediction model has faster convergence speed and better prediction accuracy characteristics. The research of this paper can provide a reasonable reference for Rock burst monitoring staff, it will be helpful to avoid and reduce the occurrence of rock burst accident.