Research On Early Warning Model Of Roof Instability Under Repeated Mining Based On RBF Neural Networ

In view of the problem that the lower coal seam is affected by repeated mining in the mining process of close distance coal seams,the stope roof is prone to collapse,resulting in end face roof fall accident,which seriously affects the safety production of working face.In this paper,the mining of close distance coal seams in Panjiang mining area of Guizhou Province is taken as the research background.By means of field observation,theoretical analysis,numerical simulation and similar simulation,the characteristics of end face roof caving under repeated mining of close distance coal seams are analyzed,the instability conditions of end face roof under different roof structures are studied,and the instability mechanism of end face roof caving under repeated mining is revealed.Through the analysis of influencing factors on the stability of end face roof under repeated mining,the characteristic parameters of end face roof fall under repeated mining of close-distance coal seams are determined.Based on the collected data of end-face roof instability under repeated mining,the RBF neural network was used for construction and training,and the early warning model of endface roof instability under repeated mining was established.Then,the monitoring and early warning system of end-face roof instability under repeated mining was developed,and the dynamic early warning of end-face roof instability under repeated mining of close distance coal seams was realized.The results show that:1.The roof movement under repeated mining is divided into three stages: normal mining,roof deterioration and end face roof fall,and the tip-to-face distance is the main influencing factor in the end face roof control.The roof collapse forms a cantilever beam structure,and the roof above the support is squeezed again due to the lifting and falling of the support.The cantilever beam structure makes tensile failure on the direct roof,and the cantilever beam structure is unstable.The dynamic load formed makes the broken roof collapse,resulting in the end face roof fall accident.2.Broken roof instability will form “loose arch” structure under repeated mining.The influencing factors of “loose arch” structure instability are tip-to-face distance,loose arch structure strength,support working resistance,support column inclination and friction coefficient between support top beam and direct roof;there are two instability modes of the cantilever beam structure,namely,the direct collapse mode of the cantilever beam structure and the hinged collapse mode of the cantilever beam structure.When the working resistance of the support is insufficient,the cantilever beam structure undergoes rotary deformation under the action of the upper coating load,resulting in the expansion and coalescence of the internal cracks in the rock stratum,and the direct collapse of the cantilever beam structure.The collapse of the cantilever beam hinged structure is divided into sliding instability and rotary instability.The sliding instability is related to the fracture degree i and the rotation angle θ of the key block B,and the rotary instability is related to the mining height and the immediate roof thickness.3.The stability of the end face roof under different advancing distances is obtained by numerical simulation.The roof is broken under repeated mining,the pressure is frequent,and the frame is not moved in time,which leads to the occurrence of the end face roof fall accident.By simulating the end face roof leakage under different influencing factors,it can be seen that improving the working resistance of hydraulic support,accelerating the propulsion speed,reasonable tip-to-face distance,suitable support column angle and strengthening the strength of surrounding rock can effectively prevent the occurrence of end face roof falling accidents.4.The input layer and output layer of RBF neural network are determined by analyzing the collected data of end face roof fall under repeated mining.The MATLAB software is used for analysis,and the prediction model of end face roof instability based on RBF neural network is established.The monitoring index of stope roof disaster under repeated mining is established.The monitoring and early warning system is designed by using Python programming language and MYSQL database,so as to realize the early warning and prediction of end face roof instability under repeated mining of close distance coal seams.