Deep rock mass fracture mechanism determination method based on wave information

Abstract

The invention belongs to the technical field of deep rock mass engineering, and discloses a method for determining a deep rock mass fracture mechanism based on wave information, which comprises the following steps of: firstly, acquiring an acoustic emission signal in a deep rock mass fracture process; effective acoustic emission signals corresponding to the acoustic emission events are screened out, and waveform kinetic parameters, including average frequency AF, rise time / amplitude wave ratio RA, waveform kurtosis Kt and information entropy density IED, of each effective acoustic emission signal are calculated; dividing the effective acoustic emission signals into a tension signal set, a shear signal set and a mixed signal set based on the waveform kinetic parameters; and finally, according to the tension signal set and the shear signal set, fitting a tension signal boundary line and a shear signal boundary line in the AF-RA map, and determining a fracture mechanism of a position corresponding to each acoustic emission event of the deep rock mass. According to the method, the waveform in the deep rock mass damage process is analyzed based on the acoustic emission waveform dynamic characteristics, the deep rock mass fracture mechanism is accurately and efficiently judged, and the method has definite physical significance and definite judgment basis.

Description

Technical Field

[0001] This invention belongs to the field of deep rock mass engineering technology, and particularly relates to a method for determining the fracture mechanism of deep rock masses based on wave information. Background Technology

[0002] Research on rock damage mechanisms is of great strategic significance for assessing the structural stability of deeply buried underground chambers, optimizing hydraulic fracturing in the efficient development of shale gas, and ensuring the long-term airtightness of nuclear waste geological storage facilities. A deep understanding of the damage evolution laws governing the initiation, propagation, and connection of microfractures within rocks is the core foundation for accurately predicting rock mass disasters, optimizing engineering parameters, and ensuring barrier function under extreme environments. However, due to the heterogeneity within rocks and the invisibility, complexity, and multi-mechanism coupling characteristics (such as tension, shear, and particle crushing) of damage evolution, traditional theories and observational methods face significant challenges in dynamically and quantitatively defining damage mechanisms.

[0003] Current methods for identifying rock damage mechanisms mainly include microscopic imaging observations (such as SEM and CT), stress-strain relationship inversion, active ultrasonic testing, and moment tensor inversion methods. However, these methods all have certain limitations in practice. For example, while high-resolution CT scans and scanning electron microscopes in microscopic imaging observations can provide static characterization of crack morphology, the destructive sampling process hinders the real-time capture of the dynamic evolution of damage, and their spatial resolution is difficult to match the nanosecond-level timescale of micro-fracture events. Furthermore, the high cost of high-resolution CT scans limits their widespread use. In SEM scans, researchers can determine the damage mechanism of rocks based on cross-sectional morphology, but the identification of steps, river patterns, tears, and micro-erosion pits is somewhat subjective. Stress-strain relationship inversion methods extrapolate damage evolution through macroscopic mechanical parameters; however, insufficient parameter sensitivity leads to limited accuracy in distinguishing tensile and shear mechanisms, especially under multiaxial stress environments where errors are significantly amplified. While active ultrasonic technology offers the advantage of being non-destructive, its low-frequency excitation characteristics (typically <100 kHz) and the anisotropic effects of rock masses severely limit its ability to capture transient responses to microfractures. Furthermore, these methods can only be used in the laboratory to determine damage mechanisms in small rock samples, and cannot be applied to the identification of damage mechanisms in deep engineering rock masses, such as those used in deep shale gas extraction, deep gas storage facilities, and deep nuclear waste geological storage projects. The moment tensor inversion method is the most widely accepted method due to its theoretical completeness; however, it requires at least six acoustic emission probes to detect the same acoustic emission event for discrimination, and the identification of the first arrival and phase is a key factor in its accuracy, which greatly limits its applicability.

[0004] Acoustic emission (AE) or microseismic monitoring technology, with its unique advantages of being non-contact, highly sensitive, and having high spatiotemporal resolution, provides a breakthrough in overcoming this challenge. It can capture transient elastic wave signals released from internal micro-fractures during the entire rock loading process in situ and in real time. By deeply mining the nonlinear complexity parameters and fine morphological parameters contained in the AE or microseismic waveform signals, the essential differences in the scale of the fracture source region, fracture mode, and energy release pattern under different damage mechanisms can be effectively revealed, thus constructing a strong correlation mapping between the waveform parameters of AE or microseismic signals and the physical processes of damage. Therefore, accurate identification of rock damage and failure mechanisms based on waveform information can not only deepen the understanding of the evolution of rock mass damage but also provide a scientific basis for intelligent early warning of precursors to deep engineering disasters, active control of hydraulic fracturing networks, and integrity assessment of nuclear waste storage barriers, strongly supporting the sustainable development of energy security and major environmental engineering projects. Furthermore, AE and microseismic equipment can be directly deployed at the engineering site, effectively monitoring vibration signals generated during deep rock mass failure. However, this method also has shortcomings: First, the damage mechanism discrimination method based on acoustic emission mainly constructs a physical driving framework based on the AF / RA ratio method. However, the determination of the AF / RA threshold lacks a rigorous physical benchmark. In many studies, the AF / RA threshold has been determined to be 1:30, 1:66.7, 1:200, 1:240.77, etc., which leads to differences in the interpretation of the same signal in different studies. Second, the binary classification system used by this method cannot characterize the tensile-shear mixed mode that accounts for as much as 30%-65% of rock mass failure.

[0005] The above analysis shows that current methods for identifying rock damage mechanisms cannot meet the goals of real-time identification, accurate classification, and convenient and rapid identification of damage mechanisms. Therefore, it is necessary to propose an effective method for identifying rock damage mechanisms based on waveform dynamics characteristics, thereby achieving accurate, real-time, and efficient identification of rock damage mechanisms. This will provide important evidence for the safe exploitation of deep energy resources, long-term stability control of underground energy storage facilities, and prevention and control of rock mass disasters in major engineering projects. Summary of the Invention

[0006] The purpose of this invention is to address the technical problems existing in the prior art by providing a method for determining the fracture mechanism of deep rock masses based on wave information, which can accurately and efficiently identify the damage and failure mechanism of deep rocks.

[0007] Considering that acoustic emission and microseismic waveform signals contain information related to rock damage mechanisms—for example, the dynamic characteristics (complexity and waveform morphology parameters) of acoustic emission waveforms are directly physically related to damage mechanisms—and that acoustic emission monitoring technology can be used for real-time monitoring not only in laboratories but also in engineering sites, this invention analyzes waveforms during deep rock mass damage based on the dynamic characteristics of acoustic emission waveforms, thereby accurately and efficiently identifying the fracture mechanisms of deep rock masses.

[0008] Based on this, the technical solution adopted in this invention is a method for determining the fracture mechanism of deep rock masses based on wave information, comprising the following steps: S1, collects acoustic emission signals during the fracturing process of deep rock masses; S2, Filter out the valid acoustic emission signals corresponding to the acoustic emission events from the acoustic emission signals in step S1; S3, calculate the waveform dynamic parameters of each valid acoustic emission signal in step S2; the waveform dynamic parameters include average frequency AF, rise time / amplitude wave ratio RA, and waveform kurtosis K. t and information entropy density (IED); S4, based on waveform dynamic parameters, divides the effective acoustic emission signal into a tension signal set T, a shear signal set S, and a mixed signal set M; S5. Based on the tension signal set T and the shear signal set S, the tension signal boundary line and the shear signal boundary line are fitted in the AF-RA diagram to determine the fracture mechanism at the corresponding location of each acoustic emission event in the deep rock mass.

[0009] In step S1 above, in order to obtain more acoustic emission waveform information and accurately locate the fracture point, it is preferable to use an acoustic emission probe to collect acoustic emission information, and the acoustic emission probe is arranged in a spatial array on the rock.

[0010] In step S2 above, it is difficult to avoid collecting acoustic emission signals not generated by rock fracture, such as electrical and mechanical noise, during the acquisition of acoustic emission signals. If these signals are included in subsequent signal analysis, they will fail to reflect the true fracture mechanism of the rock. Therefore, according to acoustic emission theory, an effective acoustic emission signal is preferably defined as a waveform signal of an acoustic emission event that generates at least four acoustic emission impact signals. This effectively avoids interference from noise signals and thus more accurately reflects the physical mechanism of rock fracture.

[0011] In step S3 above, the waveform kurtosis K t It is a fourth-order statistic describing the steepness of the probability distribution, and can comprehensively reflect the strength of outliers in each waveform data. Therefore, waveform kurtosis K is preferred. t As waveform morphology parameters of this invention. According to the Griffitz energy criterion, tensile failure is an instantaneous release of strain energy, and its acoustic emission waveform has a steep rising edge (high kurtosis); while shear failure is accompanied by interfacial friction, and the waveform exhibits multiple superimposed microslips (low kurtosis but high continuity). A larger kurtosis value reflects a sharp single peak in the waveform, corresponding to instantaneous energy release (such as tensile crack propagation), while a lower kurtosis reflects a low, multi-peaked waveform, corresponding to continuous frictional sliding (such as shear failure).

[0012] The physical meaning of waveform complexity parameters lies in their quantification of the nonlinear dynamic characteristics of acoustic emission signals, reflecting the geometric morphology of crack propagation paths, energy release patterns, and the non-uniformity of stress field distribution. Among the many complexity parameters characterizing waveforms, Information Entropy Density (IED) can calculate the coupling relationship between signal energy and information entropy per unit time, quantifying the spatiotemporal aggregation characteristics of fracture information in the waveform. According to the entropy generation rate theory, shear failure involves a large number of irreversible particle frictions, resulting in high-frequency acoustic emission signals (high entropy); while tensile failure entropy production is mainly concentrated at the moment of crack tip separation (low entropy density). In other words, a high IED indicates that the signal is concentrated in high-entropy fluctuations within a specific time period, characterizing the competitive propagation of multi-source micro-fractures (such as the interaction of secondary cracks in shear failure). A lower IED indicates that energy is concentrated in the low-entropy segment, reflecting the rapid propagation of a single main crack (typical tensile failure).

[0013] Based on this, the preferred waveform kurtosis K of the present invention t Information entropy density (IED) is used as an important parameter for determining the fracturing mechanism of deep rock masses.

[0014] The waveform kurtosis K t The calculation formula is: (1); In the formula, x i For the i-th sampling point, The average of all sampled amplitudes, N is the number of sampling points, and the spike characteristics of the acoustic emission waveform reflect the suddenness of crack propagation. K t There are two critical values ​​K t-t K t-s When K t >K t-t At that time, the acoustic emission waveform is generated by a type I crack (tensile crack), when K t <K t-s At that time, shear failure is the primary cause, while when K t-s ≤K t ≤K t-t At that time, the rock failure mode was a mixed tensile-shear failure mode.

[0015] The formula for calculating the information entropy density (IED) is as follows: (2); In the formula: M is the number of signal frames, which is empirically taken as 16; E m H represents the energy of the m-th frame. m The Shannon entropy of the m-th frame is calculated according to the following formula; (3); In the formula: K is the number of boxes, which is empirically taken as 16; p m,kLet be the probability distribution of the m-th frame within the discretized amplitude range.

[0016] It should be noted that the calculation methods for the average frequency AF and rise time / amplitude wave ratio RA are existing technologies, as shown in the following formulas: AF = Ring count / Duration; In the formula, the ring count refers to the number of times the signal pulse exceeds the threshold value (i.e., the trigger threshold); the duration refers to the total time from the first time the signal pulse exceeds the threshold value to the last time it falls below the threshold value.

[0017] RA = Rise time / Peak amplitude; In the formula, rise time refers to the time required for the signal to rise from the threshold value to the peak amplitude, and peak amplitude refers to the maximum amplitude of the signal pulse.

[0018] The purpose of step S4 above is to divide the effective acoustic emission signal into a tension signal set T, a shear signal set S, and a mixed signal set M, which includes the following sub-steps: S41, Determine the initial K t Threshold and IED threshold; S42, based on the initial K t The threshold and IED threshold initially divide the effective acoustic emission signal into a tension candidate set S. T Cut candidate set S S and the hybrid candidate set S M ; S43, for K t Optimize thresholds and IED thresholds; S44, based on the optimized K t The threshold and IED threshold are used to reclassify the effective acoustic emission signal into a tensile signal set T, a shear signal set S, and a mixed signal set M.

[0019] In step S41 above, K t Thresholds include tension signal K t-t Threshold and shear signal K t-s Threshold; IED threshold includes tension signal IED t Threshold and shear signal IED s Threshold.

[0020] Furthermore, the initial K t The steps for determining the threshold and IED threshold are as follows: S411, based on parameters AF and RA, the effective acoustic emission signals are initially divided into tension signal clusters and shear signal clusters; S412, based on the tension signal cluster and the shear signal cluster, the acoustic emission signal K in the tension signal cluster t The median value of the distribution is defined as the initial K. t-tThe threshold, the median value of the IED distribution is defined as the initial IED. t Threshold; to shear the acoustic emission signal K in the signal cluster. t The median value of the distribution is defined as the initial K. t-s The threshold, the median value of the IED distribution is defined as the initial IED. s Threshold.

[0021] In step S42 above, based on the initial K t The threshold and IED threshold initially divide the effective acoustic emission signal into a tension candidate set S. T Cut candidate set S S and the hybrid candidate set S M The rules are as follows: ; In the formula, , The initial tension signal K is respectively t-t Threshold and shear signal K t-s Threshold; , These are the initial tension signals IED. t Threshold and shear signal IED s Threshold.

[0022] In step S43 above, based on the classification purity index, given K... t Within the threshold and IED threshold range, a grid search is used to determine the optimal K. t Threshold and IED threshold; The classification purity index is expressed as: (4); In the formula, P T For the purity of tension-type signals; P S For shearing signal purity; P M The purity of mixed-type signals; the purity of each type is represented by K. t The ratio of the number of samples classified into corresponding signal classes after threshold and IED threshold iterations to the number of samples in the candidate set of the same signal class before iteration; N T N represents the total number of samples in the candidate set of tension signals before iteration; S N represents the total number of samples in the candidate set of the sheared signal before iteration. M The total number of samples in the candidate set of the mixed signal.

[0023] Specifically, based on the classification purity index, in a given first K... t Within the threshold range and the first IED threshold range, a first grid search is performed to obtain the optimized K. t Threshold and IED threshold, i.e., K t, t_1 K t,s_1 IEDt_1 IED s_1 ; and with the optimized K t The threshold and IED threshold are used to re-divide the effective acoustic emission signal and the tension candidate set S. T Cut candidate set S S and the hybrid candidate set S M Update;

[0024] With (K) t, t_1 , K t,s_1 IED t_1 IED s_1 Construct the second K around the center. t The threshold range and the second IED threshold range are based on the classification purity index, combined with the updated tension candidate set S. T Cut candidate set S S and the hybrid candidate set S M A second grid search is performed to obtain the optimal K. t Threshold and IED threshold, i.e., K t,t-opt , K t,s-opt IED t-opt IED s-opt .

[0025] Furthermore, the step size of the second grid search is half the step size of the first grid search.

[0026] In step S44 above, the tension distance d is defined. T and shear distance d S And based on the defined tension and shear distances, the acoustic emission signals in the mixed signal set M are fed back: if d T >d S , and d T If d > 0.3, then the acoustic emission signal is transferred to the tension signal set T; if d S >d T , and d S If the value is greater than 0.3, the acoustic emission signal is transferred to the shear signal set S.

[0027] Tension distance d T and shear distance d S The definition is as follows: (5); (6).

[0028] In this step, the expressions for the tension signal set T, the shear signal set S, and the mixed signal set M are as follows: .

[0029] Furthermore, after any acoustic emission signal in the mixed signal set M undergoes feedback processing, the classification purity index after feedback processing is calculated: if the classification purity index increases by more than 1%, the acoustic emission signal is transferred to the corresponding signal set; if the classification purity index increases by no more than 1%, the acoustic emission signal remains in the mixed signal set.

[0030] The purpose of step S5 is to fit the boundary line between the tension signal set T and the shear signal set S in step S4, obtain the optimal boundary line between the fitted tension signal set T and the shear signal set S, and determine the fracture mechanism of deep rock mass.

[0031] Specifically, the boundary between the tension signal set T and the shear signal set S is fitted using a combination of mathematical optimization and physical constraints; this includes the following operations:

[0032] Linear regression fitting is performed on the tension signal set T and the shear signal set S according to the following formula. For the tension signal set T, we have: (7); For the shearing signal set S, we have: (8); Where: AF i Let RA be the average frequency of the i-th sampling point in the tension signal set T. i To express the rise time / amplitude ripple ratio of the i-th sampling point in the signal set T, AF j Let RA be the average frequency of the j-th sampling point in the sheared signal set T. j Let be the rise time / amplitude ripple ratio of the j-th sampling point in the shear signal set T, and let a1 and b1 be the slope and intercept of the tension signal boundary line, respectively; a2 and b2 be the slope and intercept of the shear signal boundary line, respectively.

[0033] Tension signal boundary line L T and shear signal boundary line L S Represented as: Tension signal boundary line L T : (9); Shear signal boundary line L S : (10);

[0034] The physical constraint is that at least 80% of the signals in the tension signal set T satisfy AF. i ≥ a1·RA i +b1, at least 80% of the signals in the sheared signal set S satisfy AF. j ≤ a2·RAj +b2.

[0035] Furthermore, the error distance D of the quantization boundary partitioning error is defined. T D S D M And construct the objective function f: (11); (12); (13); (14); In the formula, D T To increase the distance beyond the boundary; D S D represents the shearing cross-boundary distance. M As an evaluation index for the rationality of the mixed zone; The total number of mixed signal sets; To stretch the signal boundary line L T and shear signal boundary line L S λ represents the actual signal count between the two boundary lines; λ is the weighting coefficient, used to balance the rationality of the mixing zone with the importance of tension and shear overshoot, and its value is 0.5.

[0036] Based on the objective function, the parameters a1, b1, a2, and b2 are optimized using the gradient descent algorithm until the objective function value is lower than a set threshold.

[0037] In this step, gradient descent calculates the partial derivatives (gradients) of the objective function with respect to each parameter and updates the parameters in the opposite direction of the gradient to reduce the objective function value. The iteration terminates when the change in the objective function f is less than 0.01 over 10 consecutive iterations (the function converges, and further iterations offer no significant optimization), and the optimal double-boundary parameters (a) are output. 1,opt , b 1,opt ,a 2,opt , b 2,opt This allows for the precise division of tension and shear signals.

[0038] This invention also provides the application of the tension signal boundary line and shear signal boundary line determined by the above method in the determination of the fracture mechanism of deep rock mass, for the purpose of determining the fracture mechanism of deep rock mass.

[0039] Specifically, the steps include: First, acquiring acoustic emission signals during the rock mass fracturing process; then, plotting the tensile signal boundary line and shear signal boundary line of the acoustic emission event of the rock mass on the AF-RA diagram according to the above method; finally, projecting the acoustic emission signals obtained by similar rock masses during fracturing under the same stress environment onto the AF-RA diagram to determine the fracturing mechanism of the rock mass.

[0040] Compared with the prior art, the present invention has the following beneficial effects: (1) The method proposed in this invention is based on the direct physical relationship between the acoustic emission waveform dynamic characteristics (complexity and waveform morphology parameters) and the damage mechanism. Compared with the traditional AF-RA discrimination method, this method has clear physical meaning and clear discrimination basis. (2) The method proposed in this invention can classify rock damage and failure mechanisms into three categories, which makes up for the shortcomings of the traditional AF-RA classification method in the tensile and shear binary classification method; (3) The method proposed in this invention overcomes the dependence of the moment tensor inversion method on acoustic emission events and first arrival identification. Attached Figure Description

[0041] Figure 1 A flowchart of the method for determining the fracture mechanism of deep rock mass based on wave information provided by the present invention;

[0042] Figure 2 The flowchart for effective acoustic emission signal screening in the embodiment is as follows: a) is a schematic diagram of the acoustic emission three-dimensional positioning system installed inside the rock sample, in which sensor 1#, sensor 2#, sensor 3#, and sensor n# are acoustic emission probes; b) is a schematic diagram of all waveform signals for locating acoustic emission events, in which numbers ①- To locate all waveform signals of the event; c) is the acoustic emission signal diagram in b), where 1#-8# are the waveform signal diagrams of eight waves;

[0043] Figure 3 The diagrams show the division between tension and shear signal clusters: a) AF-RA scatter plots corresponding to different k-value boundaries; b) tension acoustic emission signal diagram; c) shear acoustic emission signal diagram.

[0044] Figure 4 For K t A schematic diagram of the IED optimal threshold determination process, a) showing the preliminary classification results of tension and shear signals; b) showing K t Box plot; c) is the IED box plot; d) is the initial K t Schematic diagram of tensile, shear, and mixed signals divided under the IED threshold; e) is the first grid search stage K. t f) is a schematic diagram of the IED threshold iteration process; g) is a schematic diagram of the signal purity iteration process in the first grid search stage; h) is a schematic diagram of the 3C-ICP evolution process in the first grid search stage; and k is a schematic diagram of the optimization search stage. t A schematic diagram of the IED threshold and classification purity iteration process; i) is the optimal K. tA schematic diagram showing the tensile signal, shear signal, and mixed signal corresponding to the IED threshold;

[0045] Figure 5 The diagrams show the iterative process of the optimal boundary parameters a and b, with a) showing the evolution of the evaluation index and b) showing the evolution of the boundary parameters a and b.

[0046] Figure 6 A schematic diagram showing the classification results of tension signals, shear signals, and mixed signals;

[0047] Figure 7 The diagram shows the classification of acoustic emission signal types and the categorization of signal types derived from moment tensor inversion. a) is uniaxial granite; b) is split granite; c) is uniaxial sandstone; d) is split sandstone; e) is uniaxial marble; and f) is split marble. Detailed Implementation

[0048] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are part of the present invention.

[0049] Example

[0050] This embodiment provides a method for determining the fracture mechanism of deep rock masses based on wave information, and its flowchart is as follows: Figure 1 As shown, it includes the following steps:

[0051] S1, collects acoustic emission signals during the fracturing process of deep rock masses; In this step, such as Figure 2 As shown in Figure a, n acoustic emission probes are arranged in a spatial array on the rock to collect acoustic emission signals.

[0052] S2, filter out the acoustic emission signals corresponding to valid acoustic emission events from the acoustic emission signals in step S1; The flowchart for filtering valid acoustic emission signals in this step is as follows: Figure 2 As shown.

[0053] Specifically, first according to Figure 2 The acoustic emission three-dimensional localization system shown in figure a filters out acoustic emission events inside the rock sample based on the localization results, and then determines the location based on the acoustic emission events. Figure 2 As shown in b, locate all waveform signals of the event and obtain them as follows: Figure 2 The acoustic emission signal shown in c.

[0054] It should be noted that, in order to more accurately reflect the physical mechanism of rock fracture, after selecting the effective acoustic emission waveforms, wavelet denoising processing is also required on the effective acoustic emission waveforms.

[0055] S3, calculate the waveform dynamic parameters of each effective acoustic emission signal in step S2, including the average frequency AF, rise time / amplitude wave ratio RA, and waveform kurtosis K. t And information entropy density (IED).

[0056] Specifically, waveform kurtosis K t The calculation formula is: (1); In the formula, x i For the i-th sampling point, The average amplitude of all sampling points is given by K, where N is the number of sampling points. The spike characteristics of the acoustic emission waveform reflect the suddenness of crack propagation. t There are two critical values ​​K t-t K t-s When K t >K t-t At that time, the acoustic emission waveform is generated by a type I crack (tensile crack), when K t <K t-s At that time, shear failure is the primary cause, while when K t-s ≤K t ≤K t-t At that time, the rock failure mode was a mixed tensile-shear failure mode.

[0057] The formula for calculating Information Entropy Density (IED) is: (2); In the formula: M is the number of signal frames, which is empirically taken as 16; E m H represents the energy of the m-th frame. m The Shannon entropy of the m-th frame is calculated according to the following formula; (3); In the formula: K is the number of boxes, which is empirically taken as 16; p m,k Let be the probability distribution of the m-th frame within the discretized amplitude range.

[0058] Plot an AF-RA scatter plot using AF and RA as coordinates. For example... Figure 3 As shown in Figure a, the acoustic emission signal when AF / RA>500 is considered as a tension signal cluster (RA). i ,AF i Acoustic emission signals with AF / RA < 10.41 are considered as sheared signal clusters (RA). j ,AF j ). Figure 3 b is Figure 3 The tension acoustic signal diagram in a Figure 3 c is Figure 3 The shearing signal diagram in a.

[0059] S4, based on waveform dynamic parameters, divides the effective acoustic emission signal into a tension signal set T, a shear signal set S, and a mixed signal set M, specifically including the following sub-steps:

[0060] S41, Determine the initial K t Threshold and IED threshold.

[0061] It should be noted that: the initial K t The threshold includes the initial tension signal K. t-t Threshold (i.e., K) t,t_init ) and initial shear signal K t-s Threshold (i.e., K) t,s_init The initial IED threshold includes the initial tension signal IED. t Threshold (i.e., IED) t_init ) and initial shear signal IED s Threshold (i.e., IED) s_init ).

[0062] Specifically, it includes the following steps:

[0063] S411, based on parameters AF and RA, the effective acoustic emission signals are initially divided into tension signal clusters and shear signal clusters;

[0064] Specifically, based on the AF-RA scatter plot and referring to the research results of Zhang et al. (Zhang ZH, Deng JH. A new method for determining the crack classification criterion in acousticemission parameter analysis. International Journal of Rock Mechanics and Mining Sciences. 2020;130:104323. doi:10.1016 / j.ijrmms.2020.104323), the acoustic emission signal when AF / RA>500 is regarded as a tensile signal cluster (RA). i ,AF i Acoustic emission signals with AF / RA < 10.41 are considered as sheared signal clusters (RA). j ,AF j The effective acoustic emission signals are initially divided into tension signal clusters and shear signal clusters.

[0065] S412, based on the tension signal cluster and the shear signal cluster, the acoustic emission signal K in the tension signal cluster t The median value of the distribution is defined as the initial K. t-t The threshold, the median value of the IED distribution is defined as the initial IED. t Threshold; to shear the acoustic emission signal K in the signal cluster. t The median value of the distribution is defined as the initial K. t-s The threshold, the median value of the IED distribution is defined as the initial IED. s Threshold.

[0066] Specifically, based on equations (1)-(3), several K values ​​from the tension signal cluster and the shear signal cluster can be obtained. t And IED values, plot these values ​​as K t And the IED box-type distribution diagram, respectively, yielded the K of the tension signal cluster. t And the IED box-type distribution diagram, and the K of the sheared signal cluster. t And the IED box-type distribution diagram. The K of the tension signal cluster... t The median value of the box plot is defined as the initial K. t-t Threshold (i.e., K) t,t_init The median value of the IED box plot of the tension signal cluster is defined as the initial IED. t Threshold (i.e., IED) t_init ); cut the K of the signal cluster t The median value of the box plot is defined as the initial K. t-s Threshold (i.e., K) t,s_init The median value of the IED box plot of the sheared signal cluster is defined as the initial IED. s Threshold (i.e., IED) s_init ).

[0067] S42, based on the initial K t The threshold and IED threshold initially divide the effective acoustic emission signal into a tension candidate set S. T Cut candidate set S S and the hybrid candidate set S M The rules are as follows: ; In the formula, , The initial tension signal K is respectively t-t Threshold and shear signal K t-s Threshold; , These are the initial tension signals IED. t Threshold and shear signal IED s Threshold.

[0068] S43, for K t Optimize thresholds and IED thresholds;

[0069] Specifically, based on the classification purity index, given K t Within the threshold and IED threshold range, the optimal K is determined by two grid searches. t Threshold and IED threshold;

[0070] The classification purity index is expressed as: (4); In the formula, P T For the purity of tension-type signals; P S For shearing signal purity; P M The purity of mixed-type signals; the purity of each type is represented by K. t The ratio of the number of samples classified into corresponding signal classes after threshold and IED threshold iterations to the number of samples in the candidate set of the same signal class before iteration; N T N represents the total number of samples in the candidate set of tension signals before iteration; S N represents the total number of samples in the candidate set of the sheared signal before iteration. M The total number of samples in the candidate set of the mixed signal.

[0071] More specifically, in the first grid search, given the first K... t Within the threshold range and the first IED threshold range, set K. t The search interval is (K) t,s_init ×0.8, K t,t_init ×1.2), with a step size of 0.1, the search range for the IED is set to (IED). t_init ×0.8, IED s_init ×1.2), with a step size of 0.05, iterate through all threshold combinations (K t,t_th , K t,s_th IED t_th IED s_th ), must satisfy K t,s_th < K t,t_th And IED t_th < IED s_th Different threshold combinations correspond to different candidate sets. Based on the classification purity index 3C-ICP shown in equation (4), K is adjusted... t Maximizing the 3C-ICP value by adjusting the IED threshold yields a better K. t Combined with IED threshold (K) t, t_1 , K t,s_1 IED t_1 IED s_1 ); and with the optimized K tThe threshold and IED threshold are used to re-divide the effective acoustic emission signal and the tension candidate set S. T Cut candidate set S S and the hybrid candidate set S M Update;

[0072] In the second grid search, the threshold combination (K) t, t_1 , K t,s_1 IED t_1 IED s_1 Centered on ) and combined with the updated tension candidate set S T Cut candidate set S S and the hybrid candidate set S M Set K t The search interval is (K) t, s_1 ×0.8, K t, t_1 ×1.2), with a step size of 0.05, the search range for the IED is set to (IED). t_1 ×0.8, IED s_1 ×1.2), with a step size of 0.0025, iterates through all threshold combinations (K t,t_th , K t,s_th IED t_th IED s_th ), must satisfy K t,s_th < K t,t_th And IED t_th < IED s_th Different threshold combinations correspond to different candidate sets. Based on the classification purity index 3C-ICP shown in equation (4), K is adjusted... t To maximize the 3C-ICP value using the IED threshold, during the grid search optimization process, each 3C-ICP corresponds to a set of K values. t and IED threshold, when K t and the rate of change ζ of the IED threshold Kt and ζ IED When all values ​​are below 0.01, the corresponding optimal K is... t Combined with IED threshold (K) t,t-opt , K t,s-opt IED t-opt IED s-opt ).

[0073] S44, based on the optimized K t The threshold and IED threshold are used to reclassify the effective acoustic emission signal into a tensile signal set T, a shear signal set S, and a mixed signal set M.

[0074] The expressions for the tension signal set T, the shear signal set S, and the mixed signal set M are as follows: .

[0075] Define the tension distance d T and shear distance d S The acoustic emission signals in the mixed signal set M are fed back into the system based on the defined tension and shear distances to avoid interference from "pseudo-mixed signals" caused by static threshold division, thereby further refining the boundaries of the three types of signals.

[0076] If d T >d S , and d T If d > 0.3, then the acoustic emission signal is transferred to the tension signal set T; if d S >d T , and d S If the value is greater than 0.3, the acoustic emission signal is transferred to the shear signal set S.

[0077] Tension distance d T and shear distance d S The definition is as follows: (5); (6); In the formula: K t The waveform kurtosis of the mixed signal is represented by IED; the information entropy density of the mixed signal is represented by K. t,s-opt The optimal threshold for kurtosis of sheared waveforms determined after iterative optimization; K t,t-opt The optimal kurtosis threshold for tension-type waveforms determined after iterative optimization; IED t-opt The optimal threshold for the information entropy density of the tension class, determined after iterative optimization; IED s-opt This is the optimal threshold for the entropy density of the shearing class determined after iterative optimization.

[0078] In equation (5), the molecular response waveform kurtosis K t The kurtosis threshold K for sheared waveforms is higher than the optimal threshold. t,s-opt The degree and information entropy density (IED) are lower than the optimal threshold for information entropy density in shearing-type applications. s-opt The degree of kurtosis, the denominator reflects the optimal threshold K for tensile waveforms. t,t-opt The kurtosis threshold K for sheared waveforms is higher than the optimal threshold. t,s-opt The degree of and the optimal threshold for entropy density of shearing-type information ED s-opt Information entropy density above the optimal threshold (IED) for tension-type information t-opt The larger the value, the closer it is to the tensile characteristics. In equation (6), the molecule reflects the waveform kurtosis K. tBelow the optimal threshold K for kurtosis of tension-type waveforms t,t-opt The degree and IED are higher than the optimal threshold for information entropy density of tensile classes. t-opt The degree of kurtosis, the denominator reflects the optimal threshold K for tensile waveforms. t,t-opt The kurtosis threshold K for sheared waveforms is higher than the optimal threshold. t,s-opt The degree of and the optimal threshold for entropy density of shearing-type information ED s-opt Information entropy density above the optimal threshold (IED) for tension-type information t-opt The higher the value, the closer it is to the shearing feature.

[0079] Furthermore, after feeding back any acoustic emission signal in the mixed signal set M, the classification purity index after feeding back is recalculated: if the classification purity index increases by more than 1%, the acoustic emission signal is transferred to the corresponding signal set; if the classification purity index increases by no more than 1%, the acoustic emission signal remains in the mixed signal set.

[0080] S5. Based on the tension signal set T and the shear signal set S, the tension signal boundary line and the shear signal boundary line are fitted in the AF-RA diagram to determine the fracture mechanism at the corresponding location of each acoustic emission event in the deep rock mass.

[0081] Specifically, a combination of mathematical optimization and physical constraints is used to fit the boundary line between the tension signal set T and the shear signal set S.

[0082] More specifically, firstly, the tension signal set T and cut set signal S Linear regression fitting is performed, and the regression objectives are shown in Equations (7) and (8), which are to minimize the sum of squares of the vertical distances between the tension signal and the initial line (Equation 7) and to minimize the sum of squares of the vertical distances between the shear signal and the initial line (Equation 8), respectively. For the tension signal set T, we have: (7); For the shearing signal set S, we have: (8); Where: AF i Let RA be the average frequency of the i-th acoustic emission signal in the tension signal set T. i To express the rise time / amplitude ratio of the i-th acoustic emission signal in the tensile signal set T, AF j Let RA be the average frequency of the j-th acoustic emission signal in the shear signal set S. j Let be the rise time / amplitude ratio of the j-th acoustic emission signal in the shear signal set S, and let a1 and b1 be the slope and intercept of the tension signal boundary line, respectively; a2 and b2 be the slope and intercept of the shear signal boundary line, respectively.

[0083] Tension signal boundary line L Tand shear signal boundary line L S Represented as: Tension signal boundary line L T : (9); Shear signal boundary line L S : (10);

[0084] The physical constraint is that at least 80% of the signals in the tension signal set T satisfy AF. i ≥ a1·RA i +b1, at least 80% of the signals in the cut signal set S satisfy AF. j ≤ a2·RA j +b2, thus ensuring that the initial upper boundary line is approximately below the tension signal and the initial lower boundary line is approximately above the shear signal.

[0085] After the above operations are completed, define the error distance D of the quantization boundary partitioning error. T D S D M And construct an objective function f so that the final dividing line can accurately distinguish between tension and shear signals, and reasonably define the middle region of the mixed signal.

[0086] Error distance D T D S D M The objective function f is shown in the following equation: (11); (12); (13); (14); In the formula, D T To increase the distance beyond the boundary; D S D represents the shearing cross-boundary distance. M As an evaluation index for the rationality of mixed signal regions; The total number of mixed signal sets; To stretch the signal boundary line L T and shear signal boundary line L S The actual number of acoustic emission signals between the two boundary lines; λ is a weighting coefficient used to balance the rationality of the mixing zone with the importance of tension and shear overshoot, and its value is 0.5.

[0087] Based on the objective function, the parameters a1, b1, a2, and b2 are optimized using the gradient descent algorithm until the objective function value is lower than the set threshold, ultimately obtaining the optimal boundary line that satisfies the constraints.

[0088] Understandingly, gradient descent works by calculating the partial derivatives (gradients) of the objective function with respect to each parameter, and then updating the parameters in the opposite direction of the gradient to reduce the objective function value. When the change in the objective function J is less than 0.01 over 10 consecutive iterations (the function converges, and further iterations offer no significant optimization), the iteration terminates, and the optimal double-boundary parameters (a) are output. 1,opt , b 1,opt ,a 2,opt , b 2,opt This allows for the precise division of tension and shear signals.

[0089] Experimental Example

[0090] This experimental example uses deeply buried granite as an example, and adopts the method for determining the deep rock mass fracture mechanism based on wave information provided by the present invention to explain the foregoing content in detail.

[0091] First, mechanical tests were conducted on the rock to obtain all acoustic emission signals and failure point location information during the tests. Then, waveforms of acoustic emission events that generated at least four acoustic emission impact signals were used as valid acoustic emission signals. The time-domain parameters (average frequency AF and rise time / amplitude wave ratio RA) of the corresponding waveforms were calculated based on these valid acoustic emission signals. Next, based on the AF-RA scatter plot, a tensile signal cluster consisting of 3814 tensile signals and a shear signal cluster consisting of 6606 shear signals were selected using AF / RA ratios of 10.41 and 500, respectively. Figure 4 As shown in a. Then, based on the selected tensile and shear signal clusters, the K values ​​of these signals are plotted. t (like Figure 4 (as shown in b) and IED (such as Figure 4 (See diagram c) Box-shaped distribution diagram, based on K t The median value of the IED box plot was used to determine the K value in the tensile and shear signal clusters. t and the initial threshold of IED, such as Figure 4 As shown in b, K represents the tension signal and the shear signal. t The initial thresholds are K t,t_init =18.32、K t,s_init =9.23; as Figure 4 As shown in Figure c, the initial thresholds for IED of the tension signal and the shear signal are respectively IED t_init =0.45 and IED s_init =0.60.

[0092] Based on the determined K t and the initial threshold of IED, such as Figure 4 As shown in d, all acoustic emission signals are initially divided into tension candidate sets S. T Cut candidate set SS and the hybrid candidate set S M Define the classification purity index 3C-ICP, and use two grid searches to evaluate K. t The initial threshold of IED is coarsely adjusted and locally finely adjusted to obtain the optimal K. t and IED threshold; specifically, Figure 4 g shows that during the first grid search iteration, the classification purity index 3C-IDP varies with IED and K. t The evolution of indicators. For example... Figure 4 As shown in f, after 196 iterations, the classification purity index 3C-ICP reaches its maximum value of 0.867, at which point K is obtained. t, t_1 K t,s_1 IED t_1 IED s_1 The figures are 19.17, 11.12, 0.4, and 0.62 respectively (e.g., ...). Figure 4 (As shown in e). After the first coarse-tuning grid search is completed, a second grid search is performed on the threshold obtained in the first search using a local fine-tuning method; that is, local fine-tuning. For example... Figure 4 As shown in h, the classification purity index 3C-ICP reaches its maximum value of 0.8904 when the number of iterations is 9000. At this time, K t,th and IED th The rate of change of all satisfy the iteration termination condition, resulting in the following: Figure 4 The K shown in i t The optimal threshold for IED: K t,t-opt K t,s-opt IED t-opt IED s-opt The values ​​are 20.8, 11.12, 0.46, and 0.55, respectively. After the second grid search is completed, the mixed signal feedback process is performed according to equations (5) and (6) to further purify the boundaries of the three types of signals.

[0093] It should be noted that in this experimental example, when dealing with K... t After feedback processing using the IED threshold, the recalculated classification purity index improvement was -70.70%, far below 1%. Therefore, based on the K value after the second grid search... t The IED threshold is the optimal threshold.

[0094] Based on the optimal threshold, all effective acoustic emission signals are reclassified into a tension signal set T, a shear signal set S, and a mixed signal M (e.g., ...). Figure 4 As shown in i).

[0095] Once the three signal sets are determined, such as Figure 5 As shown, linear regression fitting is performed on the tension signal set and the shear signal set. Figure 5Figure 'a' reflects the evolution of the "error distance" of the three types of signals with the number of iteration steps. As can be seen from the figure, when the number of iteration steps is 85, the stretching over-limit distance D... T Shearing cross-boundary distance D S and the rationality evaluation index D of the mixed zone M All reached their minimum values, which were 1182 × 10⁻⁶. 4 18675×10 4 and 1.052×10 4 At this point, if the objective function J satisfies the condition that its change is less than 0.01 in 10 consecutive iterations, the iteration terminates. Figure 5 As can be seen from b, when the number of iterations reaches 50, the parameter values ​​a and b of the two boundary lines have reached a state of equilibrium, resulting in a1 values ​​of 14.79, b1 values ​​of 41.3, a2 values ​​of 6.63, and b2 values ​​of 16.4. Therefore, the final boundary line equation is as follows: ; ;

[0096] Based on the two dividing lines obtained, we finally got Figure 6 The three types of signals shown are then determined according to K. t Based on the IED threshold, the blue signal in the figure represents the tensile signal set T, the red signal represents the shear signal set S, and the green signal represents the mixed signal set M. Distributed in AF... i Signals above the dividing line are the final defined tension signals, accounting for 64%, and are distributed in AF. j The signal below the dividing line is a sheared signal, accounting for 15.8%, while the signal between the two dividing lines is a mixed signal, accounting for 20.2%.

[0097] Verification Example

[0098] This verification example verifies the method provided in the embodiments to illustrate the accuracy and reliability of the method described in this invention.

[0099] Specifically, the source mechanism based on moment tensor inversion is used to determine the damage mechanism of rocks during loading.

[0100] It should be noted that the method of using moment tensor inversion to determine the damage mechanism is specific to a particular acoustic emission event, and requires that the same event be received by at least six acoustic emission probes. Therefore, in a single experiment, the number of events that can be used for inversion is relatively small. This invention first selects acoustic emission events that can be used for inversion calculations, and then uses the moment tensor inversion method to determine the damage mechanism of these events, assuming that the acoustic emission event and the acoustic emission signal locating that event indicate the same damage mechanism. It is important to emphasize that the acoustic emission signals used for moment tensor inversion must be valid acoustic emission signals. Simultaneously, the proposed method is used to classify the valid acoustic emission signals, and the damage mechanism signal determined by moment tensor inversion is compared with the acoustic emission signal type determined by this invention to verify the reliability of the proposed method.

[0101] More specifically, such as Figure 7 As shown, taking granite, sandstone, and marble as examples, firstly, effective acoustic emission signals of the above rocks are collected under two different stress environments: uniaxial compression and splitting. Then, all effective acoustic emission signals are divided according to the boundary equations (Equations 9 and 10), and the boundary lines between the tension signal set and the shear signal set are obtained, as shown below. Figure 7 The classification diagram within the large rectangle shown; simultaneously, the acoustic emission signals corresponding to various acoustic emission events obtained from moment tensor inversion are extracted, and the waveform set corresponding to the shearing event is denoted as S. M-j The set of waveforms corresponding to the tension event is denoted as S. M-t The set of waveforms corresponding to the mixed events is denoted as S. M-m ,get Figure 7 Signal classification diagram within small and medium-sized rectangles, and simultaneously, S M,j S M,i and S M,m The signals located in the shear, tension, and mixed failure mechanism regions defined by the boundary line within the set were statistically analyzed, and the statistical results are as follows: Figure 7 The bar chart in the middle.

[0102] from Figure 7 It can be seen that even for the same type of specimen, under different stress environments of uniaxial compression and splitting, K t The optimal threshold for IED is also inconsistent. However, overall, under uniaxial compressive stress conditions, Figure 7 K of a uniaxial granite t The optimal threshold is the largest, at 11.52. Figure 7 K in uniaxial marble of e t The optimal threshold is the smallest, at 9.32. Figure 7 K in uniaxial sandstone of c t The optimal threshold is located in the middle, at 10.58. The optimal threshold for IEDs shows the opposite trend. Figure 7The optimal threshold for IED in uniaxial granite of type a is the smallest, at 0.51. Figure 7 The optimal threshold for IED is the largest for uniaxial marble in region e, at 0.68. Figure 7 The optimal threshold for IED in uniaxial sandstone (c) is in the middle, at 0.58. Meanwhile, the experiments revealed inconsistencies in the parameter values ​​a1, a2, b1, and b2 determined for each type of sample. Specifically, regardless of whether uniaxial compression or splitting conditions are applied, the a1, a2, b1, and b2 values ​​for granite are the highest. Figure 7 As shown in 7a and 7b, marble has the smallest values ​​for parameters a and b, such as... Figure 7 As shown in e and 7f.

[0103] In addition, from Figure 7 The histogram clearly shows that even though the boundary equations obtained for each type of sample are inconsistent, the S values ​​for each type of sample... M-j S M-t and S M-m The quantities in the set are also inconsistent, but in each type of sample, S M-j S M-t and S M-m The proportion of regions located within the various failure mechanism zones defined by the boundary lines is high, exceeding 94%. This indicates a high degree of consistency between the damage mechanisms determined by the moment tensor inversion method and those determined by the method proposed in this invention, indirectly proving the reliability of the proposed method. Furthermore, the application of the tensile signal boundary lines and shear signal boundary lines determined by the method of this invention in the determination of fracture mechanisms in deep rock masses demonstrates high accuracy.

[0104] Those skilled in the art will recognize that the embodiments described herein are intended to help the reader understand the principles of the invention, and should be understood that the scope of protection of the invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific modifications and combinations based on the technical teachings disclosed in this invention without departing from the spirit of the invention, and these modifications and combinations are still within the scope of protection of this invention.

Claims

1. A method for determining the fracture mechanism of deep rock masses based on wave information, characterized in that, Includes the following steps: S1, collects acoustic emission signals during the fracturing process of deep rock masses; S2, Filter out the valid acoustic emission signals corresponding to the acoustic emission events from the acoustic emission signals in step S1; S3, calculate the waveform dynamic parameters of each valid acoustic emission signal in step S2; the waveform dynamic parameters include average frequency AF, rise time / amplitude wave ratio RA, and waveform kurtosis K. t and information entropy density (IED); S4, based on waveform dynamic parameters, divides the effective acoustic emission signal into a tension signal set T, a shear signal set S, and a mixed signal set M; S5. Based on the tension signal set T and the shear signal set S, the tension signal boundary line and the shear signal boundary line are fitted in the AF-RA diagram to determine the fracture mechanism at the corresponding location of each acoustic emission event in the deep rock mass.

2. The method for determining the fracture mechanism of deep rock mass based on wave information according to claim 1, characterized in that, The effective acoustic emission signal is a waveform signal of an acoustic emission event that generates at least four acoustic emission impact signals.

3. The method for determining the fracture mechanism of deep rock mass based on wave information according to claim 1, characterized in that, Step S4 includes the following sub-steps: S41, Determine the initial K t Threshold and IED threshold; S42, based on the initial K t The threshold and IED threshold initially divide the effective acoustic emission signal into a tension candidate set S. T Candidate set S for cutting S and the hybrid candidate set S M ; S43, for K t Optimize thresholds and IED thresholds; S44, based on the optimized K t The threshold and IED threshold are used to reclassify the effective acoustic emission signal into a tensile signal set T, a shear signal set S, and a mixed signal set M.

4. The method for determining the deep rock mass fracture mechanism based on wave information according to claim 3, characterized in that, In step S41, K t Thresholds include tension signal K t-t Threshold and shear signal K t-s Threshold; IED threshold includes tension signal IED t Threshold and shear signal IED s Threshold; In step S42, based on the initial K t The threshold and IED threshold initially divide the effective acoustic emission signal into a tension candidate set S. T Candidate set S for cutting S and the hybrid candidate set S M The rules are as follows: ； In the formula, , The initial tension signal K is respectively t-t Threshold and shear signal K t-s Threshold; , These are the initial tension signals IED. t Threshold and shear signal IED s Threshold.

5. The method for determining the deep rock mass fracture mechanism based on wave information according to claim 4, characterized in that, In step S41, the initial K t The steps for determining the threshold and IED threshold are as follows: S411, based on parameters AF and RA, the effective acoustic emission signals are initially divided into tension signal clusters and shear signal clusters; S412, based on the tension signal cluster and the shear signal cluster, the acoustic emission signal K in the tension signal cluster t The median value of the distribution is defined as the initial K. t-t The threshold, the median value of the IED distribution is defined as the initial IED. t Threshold; to shear the acoustic emission signal K in the signal cluster. t The median value of the distribution is defined as the initial K. t-s The threshold, the median value of the IED distribution is defined as the initial IED. s Threshold.

6. The method for determining the fracture mechanism of deep rock mass based on wave information according to any one of claims 3-5, characterized in that, In step S43, based on the classification purity index, given K... t Within the threshold and IED threshold range, a grid search is used to determine the optimal K. t Threshold and IED threshold; The classification purity index is expressed as: ; In the formula, P T For the purity of tension-type signals; P S For shearing signal purity; P M The purity of mixed-type signals; the purity of each type is represented by K. t The ratio of the number of samples classified into corresponding signal classes after threshold and IED threshold iterations to the number of samples in the candidate set of the same signal class before iteration; N T N represents the total number of samples in the candidate set of tension signals before iteration; S N represents the total number of samples in the candidate set of the sheared signal before iteration. M The total number of samples in the candidate set of the mixed signal.

7. The method for determining the deep rock mass fracture mechanism based on wave information according to claim 6, characterized in that, In step S43, based on the classification purity index, in the given first K... t Within the threshold range and the first IED threshold range, a first grid search is performed to obtain the optimized K. t Threshold and IED threshold, i.e., K t, t_1 K t,s_1 IED t_1 IED s_1 ; and with the optimized K t The threshold and IED threshold are used to re-divide the effective acoustic emission signal and the tension candidate set S. T Candidate set S for cutting S and the hybrid candidate set S M Update; With (K) t, t_1 , K t,s_1 IED t_1 IED s_1 Construct the second K around the center. t The threshold range and the second IED threshold range are based on the classification purity index, combined with the updated tension candidate set S. T Candidate set S for cutting S and the hybrid candidate set S M A second grid search is performed to obtain the optimal K. t Threshold and IED threshold, i.e., K t,t-opt , K t,s-opt IED t-opt IED s-opt .

8. The method for determining the fracture mechanism of deep rock mass based on wave information according to claim 3, characterized in that, In step S44, the tension distance d is defined. T and shear distance d S And based on the defined tension and shear distances, the acoustic emission signals in the mixed signal set M are fed back: if d T >d S , and d T If d > 0.3, then the acoustic emission signal is transferred to the tension signal set T; if d S >d T , and d S If the value is greater than 0.3, the acoustic emission signal is transferred to the shear signal set S. Tension distance d T and shear distance d S The definition is as follows: ； 。 9. The method for determining the fracture mechanism of deep rock mass based on wave information according to claim 8, characterized in that, After any acoustic emission signal in the mixed signal set M is fed back, the classification purity index after the feeding back process is recalculated: if the classification purity index is improved by more than 1%, the acoustic emission signal is transferred to the corresponding signal set. If the classification purity index increases by no more than 1%, the acoustic emission signal will remain in the mixed signal set.

10. The method for determining the fracture mechanism of deep rock mass based on wave information according to claim 1, characterized in that, In step S5, the fitting steps for the tension signal boundary line and the shear signal boundary line are as follows: Linear regression fitting is performed on the tension signal set T and the shear signal set S according to the following formula. For the tension signal set T, we have: ; For the shearing signal set S, we have: ; Where: AF i Let RA be the average frequency of the i-th acoustic emission signal in the tension signal set T. i To express the rise time / amplitude ratio of the i-th acoustic emission signal in the tensile signal set T, AF j Let RA be the average frequency of the j-th acoustic emission signal in the shear signal set S. j Let be the rise time / amplitude ratio of the j-th acoustic emission signal in the shear signal set S, and let a1 and b1 be the slope and intercept of the tension signal boundary line, respectively; a2 and b2 be the slope and intercept of the shear signal boundary line, respectively. Tension signal boundary line L T and shear signal boundary line L S Represented as: Tension signal boundary line L T : ; Shear signal boundary line L S : .

11. The method for determining the fracture mechanism of deep rock mass based on wave information according to claim 10, characterized in that, When performing linear regression fitting on the tension signal set T and the shear signal set S, the following physical constraint must also be met: at least 80% of the signals in the tension signal set T satisfy AF. i ≥ a1·RA i +b1, at least 80% of the signals in the sheared signal set S satisfy AF. j ≤ a2·RA j +b2.

12. The method for determining the deep rock mass fracture mechanism based on wave information according to claim 10 or 11, characterized in that, Define the error distance D of the quantization boundary partitioning error. T D S D M And construct the objective function J: ； ； ； ； In the formula, D T To increase the distance beyond the boundary; D S D represents the shearing cross-boundary distance. M As an evaluation index for the rationality of mixed signal regions; The total number of mixed signal sets; To stretch the signal boundary line L T and shear signal boundary line L S The actual number of acoustic emission signals between the two boundary lines; λ is the weighting coefficient. Based on the objective function, the parameters a1, b1, a2, and b2 are optimized using the gradient descent algorithm until the objective function value is lower than a set threshold.

13. The application of the tension signal boundary line and shear signal boundary line determined by the method of any one of claims 1 to 12 in the determination of the fracture mechanism of deep rock masses, characterized in that, Used for determining the fracture mechanism of similar rock masses under the same stress environment.

Images