Multi-Parameter Precursor Characteristics Analysis In The Process Of Coal And Rock Failure

Rock burst was a kind of dynamic phenomenon of rock mass induced by engineering activities such as mining and excavation,which was a kind of dynamical disasters caused by mining.Due to the complexity of geological structure,geological movement and perturbations caused by various mining activities,and the dynamic process had strong nonlinearity,it was difficult to accurately predict the phenomenon of rock burst using single signal characteristics.Therefore,it was significant to guide the coal mine safety production by analyzing the multi-parameter precursor characteristics in the process of coal-rock rupture to improve the prediction accuracy.After analyzing acoustic emission signal,electromagnetic radiation signal and infrared radiation temperature,We found that the acoustic emission signal increased with the increase of load at the initial stage of loading,but there was a local low value before the disaster,After the instability and then significantly increased.Electromagnetic radiation experienced increased amplitude-maximum-decrease-minimum-the process of increasing again.When the rock and coal are destabilized and destroyed,the temperature of the infrared radiation increases.Afterwards,the temperature of the infrared radiation gradually decreases,and the average surface temperature will firstly increase and then decrease.Then,using a multi-parameter comprehensive index forecasting method of coal and rock disaster,the single index of danger of three indexes were calculated and their respective weights were calculated according to the three signal amplitudes,and then the comprehensive index of risk was calculated to evaluate the hazard level of disaster.Finally,optimized the GRNN method based on the fruit fly algorithm(FOA),the fruit flies randomly foraged for olfaction to determine the distance between the Drosophila individuals’ coordinates and the origin,calculated the reciprocal,and found the value(S),then substituted it into the SPREAD parameter of GRNN so that the root mean square error(RMSE)between the GRNN output and the measured value was as small as possible,the optimal S value was retained as the SPREAD value of the GRNN,the method was iteratively optimized.Then,the acoustic emission amplitude,electromagnetic radiation amplitude and infrared radiation temperature were taken as the input variables of GRNN.Based on the comprehensive risk index obtained from the fusion of the three signals,established rock burst prediction model based on FOA-GRNN.Prediction results show no omission,which can accurately reflect the phenomenon of rock burst.