Study On Energy Conversion Mechanism And State Identification During Rock Deformation And Instability

With the rapid development of China’s economy and accelerated process of urbanization,a series of mining and geotechnical engineering continuously emerge,such as deep resources exploitation engineering,underground space engineering,tunneling,et al.Due to the disturbance of rock mass used to construct engineering body,the frequency and intensity of disaster accidents markedly increase,and the difficulty of state identification,monitoring and early warning,accurate prevention and control of rock in mining and geotechnical engineering also increase significantly.In nature,the process of rock deformation and instability is the micro-crack closure,initiation,growth,and coalescence,the nonlinear evolution and conversion of energy contain the state characteristics of rock deformation and instability,and accompany various forms of precursor information,which is an important breakthrough of the disasters identification,monitoring and warning.In this study,the research methods,combining experimental research,theoretical analysis,numerical simulation,and machine learning,were used to study the energy conversion mechanism and precursor characteristics during rock deformation and instability.Triaxial tests of pre-peak monotonic loading & post-peak cyclic loading,and pre-peak & post-peak cyclic loading were conducted on sandstone,respectively.The stage-calculation method for energy was proposed,and the characteristic energy was used to quantify the confining pressure effect at pre-peak stage.At post-peak stage,the frictional dissipation energy per unit strain was proposed to characterize the structural effect of fractured rock,and the influence of confining pressure and pre-peak loading mode on post-peak energy were studied.Then,the relation of macroscopic energy conversion quantified by energy ratio was established,thereby the determination method of characteristic stresses and brittleness evaluation index were proposed,and the damage evolution equation and constitutive model were derived.Further,based on the microscopic energy conversion of crystal materials,the microscopic energy conversion law was analyzed form the microscopic morphology of rock,and then the mesoscopic energy evolution and conversion and the influencing factors were analyzed using PFC.On this basis,the mutation characteristics of energy conversion were analyzed in time series,and then an early warning model of energy conversion was established based on cusp mutation theory.Finally,the stage characteristics of the deformation and instability of rock with pre-existing cracks under uniaxial compression were analyzed using the proposed method for determining the characteristic stresses;the energy precursor of rock with pre-existing cracks were investigated using the catastrophic warning model of energy conversion;the numerical difference and spatial concentration of the deformation field were quantified by two statistical indexes,and the two statistical indexes relating with rock deformation and instability was found out,and then the deformation field precursor of rock with pre-existing crack were investigated,the precursor of acoustic emission also was extracted.Moreover,a mult-precursor identification system,which reflected the temporal characteristics of energy precursor,deformation field precursor and acoustic emission precursor,was created,and then an identification model of rock deformation instability state based on machine learning was established,and its identification capability was evaluated from algorithm and feature combination.The results show that:(1)All forms of energy had a nonlinear evolution during rock deformation and instability.At the pre-peak stage,all forms of energy increased with nonlinear growth rate,and the import energy reached the maximum value at the peak stress.The elastic strain energy reached the maximum value before the peak stress,and dropped suddenly at the peak stress,while the dissipated energy increased suddenly before the peak stress,and reached the maximum value at the peak stress,and the relationship between energy and confining pressure at each characteristic stress was different.At the post-peak stage,all forms of energy decreased with the stress dropping,and then changed stably with the stress stability;all forms of energy increased with the increasing of confining pressure;the frictional dissipation energy per unit strain not only revealed the confining pressure effect of the structure of the fractured rock,but also reflected the influence of pre-peak loading mode on the post-peak mechanical properties.(2)The nonlinearity of energy conversion showed a multi-stage evolution,thereof the obtained characteristic stresses were lower than the results determined by lateral strain method and volumetric strain method,but they had the same evolution trend with confining pressure.The proposed brittleness index varied monotonously in(-∞,1)and had a simple linear relationship with peak stress,elastic modulus,and post-peak modulus,and it was suitable for brittleness evaluation of rock under different rock types,confining pressures,and inclination angle of bedding planes.The damage evolution equation and constitutive model agreed with experimental data,and could reflect the nature of rock damage before and after the damage stress.(3)The essence of microscopic energy conversion of rock deformation and instability was the accumulation,dissipation,and release of energy among mineral particles,cementation,cracks and micropores.The evolution and conversion characteristics of microscopic energy were approximately consistent with the experimental results,and were greatly affected by confining pressure,loading rate and particle diameter.(4)The energy conversion in time series had obvious mutation characteristics,and the early warning of energy conversion mutation warning model was consistent with the development of new crack and the quiet period of AE ring count rate.(5)The damage evolution of rock with pre-existing crack had five stages,and the energy conversion showed mutation characteristics.The second less than zero of the mutation index for energy conversion appeared before the damage stress,and it could be used as the energy precursor of rock deformation and instability.The spatial index and degree index of heterogeneity deformation changed in three stages,which was closely related to the state of rock deformation and instability,and the first mutation of the two indexes evolution rate occurred before and after the damage stress,which was the precursor of rock deformation field.The first mutation of AE ring count rate and the frequency band(80 ～ 120 k Hz)were AE precursors.The sequence process of the three precursors was the energy precursors → the heterogeneous deformation precursors →the AE precursors,and the difference in sequence could be used as the warning of rock deformation and instability in different stages.Rock state identification model established by different machine learning algorithms had different effects on rock state identification under different feature combinations.From the three evaluation indexes of accuracy,accuracy and recall rate,the effect of multi-type feature combination on rock state identification was higher than that of single type feature combination.Among the multi-type feature combinations,the combination of energy feature and heterogeneous deformation field feature had the highest accuracy in identifying rock state.The rock state identification model based on RF algorithm had the best identification ability.