Combined Prediction Model Of Broken Rock Zone And Its Application

The size of the broken rock zone directly affects the design of the roadway support.In addition to limited actual measurements,it is of great significance to accurately predict broken rock zone.In this paper,the GSO-LSSVM model based on the influence factors of broken rock zone was established,which based on the collected data of broken rock zone and their influencing factors.And an improved GM(2,1)model based on the development law of broken rock zone was established,which based on the field measurement of suburban mines.On the basis of the two prediction models,the entropy weight combination prediction model based on the data fluctuation of influencing factors is established.On this basis,the numerical calculation model of synergy support of bolts and cables is established,and the synergy matching effect of prestress field parameters during the combined support of bolts and cables is analyzed.The main conclusions of this paper are as follows:(1)According to the measurement of the broken rock zone of the geological radar and the borehole camera,it is considered that both of them are more than 2.0 m,which belongs to the surrounding rock of the large broken rock zone.The actual measurement results show that even in the apparently the same lithological stratum,the broken rock zone will change within the range of ±(5.0-20.0)cm.Therefore,it is suggested that the absolute value of the error between the predicted value and the actual value of the deep roadway broken rock zone should be controlled within 20.0cm.(2)The results of GSO-LSSVM broken rock zone prediction model on the Chengjiao Mine is smaller than the measured value.The average percentage error is within 6%,and the root mean square error is within 15%.For other mines,this model is not recommended for predicting loose circles.(3)The absolute value of the error between the predicted value of the improved GM(2,1)broken rock zone prediction model and the measured value is generally 0 ~15.0 cm in Chengjiao Mine.Its is generally within 10.0 cm in lithology change section,which can guide the support design of the second level roadway in Chengjiao Mine.(4)Based on the entropy weight theory,the combination prediction model of broken rock zone has a better prediction effect than the improved GM(2,1)model in lithological plateau,and it is better than the overall prediction effect of theGSO-LSSVM model.However,it cannot solve the problem that the absolute value of the prediction point error may exceed 20.0 cm due to data fluctuation and trend development.Therefore,it must be admitted that any broken rock zone prediction model cannot accurately predict the broken rock zone of all measuring points.However,it should be ensured that the predicted value is within the fluctuation range of the actual broken rock zone when the data fluctuates.The fundamental method to solve the problem is to increase the actual measurement,and monitor its changes,to provide more and better basic data for the prediction model.(5)For different types of roadways and their different positions,the correlation between each influencing factor and broken rock zone is different,and the overall correlation is that the vertical ground stress and the uniaxial compressive strength have the greatest correlation.The correlation of the remaining four items is not much different,but the correlation between the side stress coefficient and span-to-height ratio of the rectangular roadway is slightly larger than that of the arch roadway.(6)Aiming at the support of the surrounding rock roadway with large broken rock zone in the second level of Chengjiao Mine,the synergy of parameters such as the length of bolts and cables,the distance between rows,the length of anchoring,and the pretensioning force was studied.On-site engineering tests show that the optimized plan avoids the bolt or cable being broken and broken by each.The synergistic effect of the bolt and the cable effectively controls the deformation of the surrounding rock.There are 47 figures,26 tables and 121 references in this paper.