Dynamic evaluation method for dangerous rock stability based on multi-source kinetic parameter fusion

Abstract

The present application provides a dynamic evaluation method for dangerous rock mass stability based on multi-source kinetic parameter fusion, comprising: synchronously collecting multi-source monitoring signals of dangerous rock mass and extracting multi-scale kinetic parameters; calculating the change trend value of the parameters relative to the initial value; using a nonlinear conversion function to convert the change trend value into the damage contribution of multi-source kinetic parameters to rock mass stability; introducing a stage self-adaptive weight mechanism, dynamically allocating weights according to different mechanical evolution stages of the rock mass, and introducing a transition factor for smooth transition calculation when the stage is switched; finally, combining the damage contribution and the dynamic weight to calculate the comprehensive dynamic stability index of the dangerous rock mass, and making current state judgment and graded early warning. The present application can solve the technical problems in the prior art, such as lagging response of early warning, lack of fracture mechanism fusion of multi-source data, fixed weight of evaluation model or hard switching in the transition section between different mechanical stages, resulting in evaluation index jump, easy to cause system false alarm or false alarm.

Description

Technical Field

[0001] This invention relates to the field of geological disaster monitoring technology, specifically to a dynamic evaluation method for the stability of unstable rock masses based on the fusion of multi-source dynamic parameters. Background Technology

[0002] As large-scale projects such as transportation, water conservancy, and mining continue to extend into complex mountainous areas, disasters caused by rockfalls are becoming increasingly frequent, posing a serious threat to project safety and people's lives and property. Currently, conventional monitoring methods for rockfalls mainly rely on equipment such as global navigation satellite systems, crack gauges, and total stations. Their core logic lies in capturing the surface displacement and macroscopic crack distribution of the rockfall.

[0003] However, during the incubation period of microcrack initiation and propagation within a rock mass, its external macroscopic displacement is often extremely weak. By the time the monitoring system detects significant apparent deformation, the rock mass is often already in a state of imminent collapse or in the process of crumbling, leaving a very short window for on-site personnel evacuation and engineering intervention. Furthermore, while some existing monitoring systems incorporate acoustic emission, microseismic, or environmental vibration equipment, the multi-source data is often isolated, merely a simple data compilation, lacking a comprehensive evaluation model capable of revealing the entire failure mechanism within the rock mass from stiffness degradation to crack-driven microfracture. This results in the inability to effectively identify the early, hidden damage stage of a rock mass transitioning from stability to separation.

[0004] Furthermore, the sensitivity and indicative significance of the same dynamic parameter to rock mass stability dynamically changes at different stages of rock mass deformation and failure. Traditional evaluation methods typically employ static weight allocation or directly switch weights when stages are manually changed. This rigid mechanism easily leads to drastic jumps in evaluation indices at stage boundaries, resulting in frequent false alarms or missed alarms, making it difficult to meet the requirements for all-weather, highly robust automatic early warning systems in the field.

[0005] Therefore, how to establish a dynamic evaluation model that can integrate micro and macro mechanisms and whose weights can be adaptively adjusted and smoothly transitioned with the evolution of rock mass damage is a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention proposes a dynamic evaluation method for the stability of unstable rock masses based on the fusion of multi-source dynamic parameters. This method aims to solve the technical problems in existing technologies, such as delayed early warning response, lack of mechanism fusion of multi-source data, and the tendency of fixed evaluation model weights to cause false alarms or missed alarms.

[0007] The technical solution adopted in this invention is as follows: simultaneously acquiring multi-source monitoring signals of the unstable rock mass and extracting multi-scale dynamic parameters; calculating the trend values ​​of parameter changes relative to the initial values; using a nonlinear transformation function to convert the trend values ​​into the damage contribution of multi-source dynamic parameters to the stability of the rock mass; introducing a stage-adaptive weighting mechanism to dynamically allocate weights according to different mechanical evolution stages of the rock mass, and introducing a transition factor for smooth transition calculation when switching stages; finally, combining the damage contribution and dynamic weights to calculate the comprehensive dynamic stability index of the unstable rock mass, and performing current state determination and graded early warning.

[0008] The specific steps of this invention include: Collect multi-source monitoring signals of unstable rock masses and extract dynamic parameters; Using the initial values ​​of the dynamic parameters of the unstable rock mass under stable conditions as a benchmark, calculate the trend value of the change of the dynamic parameters relative to the initial values ​​of the dynamic parameters at the current moment; According to a preset nonlinear transformation function, the changing trend value is converted into the damage contribution of the dynamic parameter to the stability of the unstable rock mass; A stage-adaptive weighting mechanism is introduced to assign corresponding dynamic weights to the dynamic parameters based on the current mechanical evolution stage of the unstable rock mass. The damage contribution of the dynamic parameters is integrated with their corresponding dynamic weights to calculate the comprehensive dynamic stability index of the unstable rock mass at the current moment.

[0009] Furthermore, the dynamic parameters include natural frequency, damping ratio, acoustic emission cumulative energy rate, signal spectral entropy, and equivalent fracture strength factor; Among them, environmental vibration signals are collected by a high-sensitivity accelerometer to extract the natural frequency and damping ratio; The rupture signal is acquired by acoustic emission or microseismic sensors to extract the cumulative acoustic emission energy and calculate the signal spectral entropy. The equivalent fracture strength factor is obtained by inversion based on vibration response or acoustic emission source mechanism.

[0010] Furthermore, the method for obtaining the equivalent fracture strength factor specifically includes: The equivalent crack depth of the unstable rock mass is inverted based on the natural frequency of the current cracked rock mass: in, Indicates the equivalent crack depth. This indicates the characteristic size of the unstable rock mass in the direction of the crack. This represents the measured natural frequency of the current cracked rock mass. The reference natural frequency of the intact rock mass is represented by γ, and the geometric shape factors are represented by n. Substituting the equivalent crack depth into the fracture mechanics formula, the equivalent fracture strength factor is calculated: in, Represents the equivalent fracture strength factor. Indicates the fracture toughness of the rock mass. This indicates the predicted frequency corresponding to the critical moment of rock mass instability.

[0011] Furthermore, the Sigmoid nonlinear function is used as the nonlinear transformation function to convert the changing trend value into the damage contribution of the dynamic parameter to the stability of the unstable rock mass: in, Indicates the degree of damage contribution. Indicates the value of the changing trend. Indicates the slope factor. This represents the trend value corresponding to the midpoint of the transformation.

[0012] Furthermore, the step of determining the slope factor includes: Loading failure tests were conducted on similar rock samples of the target unstable rock mass, and all dynamic parameters were collected simultaneously. The entire test process was divided into multiple mechanical stages, and the actual degree of damage in each mechanical stage was calibrated. A scatter plot was created with the trend values ​​of the experimentally measured parameters on the x-axis and the calibrated actual damage level on the y-axis. The plot was then fitted using the Sigmoid function to obtain the experimental values. ; Using physical correction formulas to adjust experimental values After making corrections, the physical correction value is obtained: in, Indicates experimental values. Indicates the physical correction value. Indicates the volume of the laboratory sample. This indicates the measured volume of the unstable rock mass at the site. This represents the initial damage coefficient of the unstable rock mass relative to the sample.

[0013] Furthermore, a stage-adaptive weighting mechanism is introduced to assign dynamic weights to the dynamic parameters based on the current mechanical evolution stage of the unstable rock mass. Specific steps include: The mechanical evolution stage S i Including the stable phase Separation stage Destruction phase ; When a rock mass is in the separation or failure stage, its corresponding dynamic weight is... The calculation steps include: Calculate the average value of the dynamic parameter across all data points during its mechanical evolution phase, and calculate its average relative rate of change with respect to the initial value in the steady-state phase: in, Indicates the average relative rate of change. Representing the stages of mechanical evolution The average of all data points within the range. Indicates the initial value during the stable phase; The stage-to-stage differentiation of the dynamic parameters between two adjacent mechanical evolution stages is calculated using the following formula: in, Let be the inter-stage discrimination between the i-th and j-th target stages. and These represent the stages of mechanical evolution. or The average of all data points within the range. This represents the pooled standard deviation; Combining the average relative rate of change and the inter-stage discrimination, calculate the comprehensive sensitivity coefficient of the kinetic parameters in the failure stage and the separation stage: in, , These represent the combined sensitivity coefficients for the separation and destruction stages, respectively. , and This represents the adjustment coefficient. , These represent the average relative rates of change during the separation and destruction phases, respectively. Normalize the overall sensitivity coefficients of all dynamic parameters to obtain the dynamic weights corresponding to each dynamic parameter in this stage: in, This represents the dynamic weight of the i-th dynamic parameter. This represents the comprehensive sensitivity coefficient of the dynamic parameters at the target stage; When the unstable rock mass is in a stable stage, its corresponding dynamic weight The calculation steps include: The monitoring data of the unstable rock mass during the period when it was in an absolutely stable state were obtained to form a stable state training dataset; The coefficients of variation for each dynamic parameter are calculated using the following formula: in, Represents the coefficient of variation. This represents the mean of the data. Indicates standard deviation; The dynamic weight of each dynamic parameter in the steady-state phase is obtained by taking the reciprocal of its coefficient of variation and normalizing it. : in, Represents dynamic parameters During the stable phase The dynamic weights, where N represents the total number of dynamic parameters. This represents the coefficient of variation.

[0014] Furthermore, in step S5, the damage contribution of the dynamic parameters is fused with their corresponding dynamic weights to calculate the comprehensive dynamic stability index of the unstable rock mass at the current moment. Specific steps include: Based on the following DSI calculation formula, the damage contribution of the dynamic parameters and their corresponding dynamic weights are integrated to calculate the comprehensive dynamic stability index of the unstable rock mass at the current moment: in, Cp(t) represents the comprehensive dynamic stability index at time t, and Cp(t) represents the dynamic parameters. The damage contribution obtained at time t through the Sigmoid function transformation. Represents dynamic parameters The real-time dynamic weights that actually participate in the calculation at time t.

[0015] Furthermore, when determining that the unstable rock mass meets the conditions and a stage transition occurs, a smooth transition calculation is performed on the dynamic weights of each dynamic parameter. The specific steps include: The moment when the recording phase switching criterion is first met is the transition start point t. s And preset the transition time length t t The transition endpoint is determined as follows: in, Indicates the end point of the transition. Indicates the starting point of the transition. Indicates the length of the transition time; The transition window period for phase switching is determined as follows: ; During the transition window period, the transition factor is calculated according to the following formula: in, Indicates the transition factor. Indicates the slope factor; Using the aforementioned transition factor, the real-time transition weight at the time of phase switching is calculated according to the following formula. in, Indicates the real-time transition weight. This represents the dynamic weights before the phase transition occurs. This indicates the dynamic weight after a phase transition occurs; When it is determined that the unstable rock mass meets the conditions and a phase transition occurs, within a preset transition window period, the real-time weight is substituted into the DSI calculation formula to replace the dynamic weight for the comprehensive dynamic stability index calculation.

[0016] Furthermore, the adjustment coefficient is dynamically adjusted and determined through objective function optimization based on classification robustness, specifically including the following steps: Construct a robust objective function to measure the effectiveness of phase division: in, Let represent the objective function for strong robustness, and let i represent the index number of the mechanical evolution stage, with values ​​1, 2, and 3 corresponding to the stable stage, separation stage, and failure stage, respectively. This represents the i-th stage of mechanical evolution. This indicates that it belongs to the stage of mechanical evolution. Sample size This represents the mean of the dynamic stability index for all samples within stage S. This represents the global mean of the dynamic stability index for all samples across all three phases. Indicating the stage of mechanical evolution The specific data sample index number, with values ​​1, 2, ... , This indicates that it belongs to the stage of mechanical evolution. The The actual calculated value of the dynamic stability index for each sample; Establish constraints based on the aforementioned adjustment coefficient: Under the above constraints, an optimization calculation is carried out with the objective of maximizing the strong robustness objective function, and the optimal weight coefficient combination A1*, A2*, A3* that maximizes the difference between stages and the concentration within stages is obtained through solution. Substitute the optimal weight coefficient combination into the calculation formula of the stage comprehensive sensitivity coefficient corresponding to the separation stage and the destruction stage, and calculate the optimal comprehensive sensitivity coefficient of each parameter in the corresponding stage. Normalize the optimal comprehensive sensitivity coefficient to obtain the final objective weight set of each dynamic parameter in the corresponding stage.

[0017] Further, after obtaining the dynamic stability index, it further includes comparing the dynamic stability index with a preset threshold range to determine the current stable state and the stage of the dangerous rock mass. The specific steps are as follows: Through numerical simulation and indoor mechanical tests, preset the critical threshold D1, warning threshold D2 and disaster threshold D3 of the dangerous rock mass. In response to DSI(t) ≤ D1, it is determined that the dangerous rock mass is in a stable stage, and a blue safety signal is issued. In response to D1 < DSI(t) ≤ D2, it is determined that the dangerous rock mass enters the initial stage of the separation stage, and a yellow warning is issued. In response to D2 < DSI(t) ≤ D3, it is determined that the dangerous rock mass enters the middle and late stages of the separation stage, and an orange warning is issued. In response to DSI(t) > D3, it is determined that the dangerous rock mass enters the critical period from separation to destruction, and is in a critically unstable state, and a red alarm is issued.

[0018] It can be seen from the above technical solutions that the beneficial technical effects of the present invention are as follows: 1. By deeply integrating micro-dynamic parameters such as acoustic emission and environmental vibration with the fracture mechanics model, the micro-evolution process of the initiation of micro-cracks and the degradation of the overall stiffness inside the rock mass is perceived, breaking through the limitation of the lag of traditional macroscopic displacement monitoring.

[0019] 2. By constructing the comprehensive sensitivity coefficient and the strong robustness objective function, the weight distribution of each parameter in different evolution stages is more in line with the mechanical mechanism. At the same time, a smoothing transition mechanism based on a time window and real-time transition weights are introduced, realizing the adaptive adjustment of the evaluation weight and the anti-abrupt change smoothing transition. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for the description of the specific embodiments or the prior art. In all the drawings, similar elements or parts are generally marked with similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to scale.

[0021] Figure 1 This is a flowchart of a method according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating the weight vector replacement process according to an embodiment of the present invention. Figure 3 This is a flowchart illustrating the update process for failures within the window period according to an embodiment of the present invention. Detailed Implementation

[0022] The embodiments of the technical solution of the present invention will now be described in detail with reference to the accompanying drawings. These embodiments are merely illustrative of the technical solution of the present invention and are therefore intended to limit the scope of protection of the present invention.

[0023] It should be noted that, unless otherwise stated, the technical or scientific terms used in this application should have the ordinary meaning as understood by those skilled in the art to which this invention pertains.

[0024] This embodiment provides a dynamic evaluation method for the stability of unstable rock masses based on the fusion of multi-source dynamic parameters. The working principle of this embodiment is explained in detail below: The method flowchart of this embodiment is as follows: Figure 1 As shown, it includes: Simultaneously and continuously acquire multi-source monitoring signals of unstable rock masses and extract dynamic parameters from multiple dimensions; Using the initial values ​​of the dynamic parameters of the unstable rock mass under stable conditions as a reference, calculate the relative rate of change of the dynamic parameters at the current moment relative to the initial values ​​of the dynamic parameters; In the formula, Indicates the value of the changing trend. This represents the parameter value at time t. Indicates the initial value of the parameter; According to the preset nonlinear transformation function, the relative rate of change is converted into the damage contribution of the dynamic parameter to the stability of the unstable rock mass, which is used to quantify the nonlinear physical influence of parameter changes on rock mass stability. An adaptive weighting mechanism is introduced to assign dynamic weights to the dynamic parameters based on the current mechanical evolution stage of the unstable rock mass. The damage contribution of the dynamic parameters is integrated with their corresponding dynamic weights to calculate the comprehensive dynamic stability index of the unstable rock mass at the current moment.

[0025] In this embodiment, the dynamic parameters further include natural frequency, damping ratio, acoustic emission cumulative energy rate, signal spectral entropy, and equivalent fracture strength factor. Specifically, the natural frequency is extracted by collecting environmental vibration signals using a high-sensitivity accelerometer. Damping ratio ; The cumulative acoustic emission energy rate is extracted by acquiring rupture signals using acoustic emission or microseismic sensors. And calculate the signal spectral entropy. ; The acoustic emission cumulative energy rate The extraction method is as follows: 1) Signal acquisition: A wideband acoustic emission sensor with a resonant frequency of 30-150kHz is used, coupled to the surface of the unstable rock mass or inside the borehole. The sampling frequency is controlled to be ≥1MHz to ensure complete recording of the micro-fracture waveform. A preamplifier (gain 40dB) and an anti-aliasing filter are set for signal acquisition.

[0026] 2) Raw Signal Denoising and Event Recognition: First, the Short-Time Energy to Long-Time Energy Ratio (STA / LTA) method is used for event detection. The short-time window length TSTA is defined as 100 μs, and the long-time window length TLTA as 1 ms. The STA / LTA ratio is calculated at each moment, and when the ratio exceeds a preset threshold, it is marked as the start point of an acoustic emission event. Then, an adaptive threshold zero-crossing count is used to remove electrical noise: only segments with a zero-crossing rate in the range of 20-200 kHz are retained.

[0027] 3) Single-event energy calculation: For the i-th detected acoustic emission event, its time-domain waveform is defined as xi[n], the number of sampling points is N, and the sampling interval is Δt. According to the formula... Calculate the energy of the event.

[0028] 4) Cumulative energy rate calculation: Set the sliding time window T w And step size ΔT. Within the window [t, t+Tw], the accumulated energy is ,in, This represents all events within the window.

[0029] When there are no events Given the background noise level, the final output normalized cumulative energy (dimensionless) is: ,in Within 24 hours of the stabilization phase of the median.

[0030] The method for calculating the signal spectral entropy is as follows: 1) Data acquisition: Continuous recording is performed using an accelerometer to obtain initial data.

[0031] 2) Signal Segmentation and Windowing: The continuously monitored signal y(t) is segmented into fixed time windows TH, with adjacent windows overlapping by 50%. A Hanning window is applied to the signal within each window to suppress spectral leakage.

[0032] 3) Power spectral density estimation: The Welch method is used to calculate the power spectral density P(f). Each window is further subdivided into multiple segments with 50% overlap between adjacent segments. The power spectrum of each segment is obtained by fast Fourier transform, and the average is taken to obtain the P(f) of the window. The frequency range covers the main natural frequencies and micro-fracture frequencies of the unstable rock mass.

[0033] 4) Normalized probability density calculation: The power spectral density is calculated according to... Convert to a probability distribution, where M represents the number of discrete frequency points, f k This represents the k-th frequency component.

[0034] 5) Spectral entropy calculation: Calculated according to the Shannon entropy formula. When the signal is a pure single-frequency sine wave, Hs→0; when the signal is white noise, Hs→log2M. To avoid numerical instability, when... When =0, set plog2p=0.

[0035] The equivalent fracture strength factor is obtained by inversion based on vibration response or acoustic emission source mechanism. .

[0036] In this embodiment, the method for obtaining the equivalent fracture strength factor specifically includes: Establishing the frequency-crack size relationship: The rock mass containing fractures is simplified into an equivalent model containing cracks. Based on elastic dynamics and damage mechanics, the square of the first natural frequency f 2 It is proportional to the equivalent bending stiffness EI of the structure. The presence of cracks will reduce the equivalent stiffness, especially for a length of... The edge cracks exhibit an approximate relationship: in, This indicates the natural frequency of the current cracked rock mass. The reference natural frequency representing a complete, crack-free rock mass. This represents the equivalent crack depth to be inverted. The characteristic dimension of the unstable rock mass in the crack direction is represented by γ and n, which are geometric shape coefficients related to crack type, location and boundary conditions. They can be calibrated for specific unstable rock mass models through finite element numerical simulation. In the formula, the calibration methods and typical values ​​of the geometric shape coefficients γ and n are as follows: A rock mass can usually be simplified as a cantilever beam or simply supported beam model. Taking a cantilever beam rock mass with a fixed root as an example, the typical values ​​are γ∈[1.0,1.5] and n=3. The specific calibration steps are: establish a three-dimensional finite element model of the specific rock mass, pre-determine cracks of different depths at the main control section, and extract the corresponding first-order natural frequencies through modal analysis. By fitting the extracted multiple sets of (a, f(a)) data pairs using the nonlinear least squares method, γ and n can be obtained in reverse.

[0037] The above approximate equation applies to the crack depth ratio In the case of a single controlling fracture, for ease of engineering application in real-time inversion calculations, extremely small higher-order infinitesimals are ignored, and the approximate equality sign in the above relationship is directly replaced with an absolute equality sign. Verification is performed through finite element calibration. Within the main monitoring range, the absolute value of the model truncation error introduced by this direct equivalent replacement is less than 0.1, which meets the engineering accuracy requirements for early warning of dangerous rock masses.

[0038] Based on the frequency-crack size relationship established and equivalently replaced above: Through algebraic transformation, the formula for calculating the current equivalent crack depth a(t) of the unstable rock mass can be derived as follows: in, Indicates the equivalent crack depth. This indicates the characteristic size of the unstable rock mass in the direction of the crack. This represents the measured first-order natural frequency of the cracked rock mass at the current moment. The reference natural frequency represents the intact, crack-free rock mass, while γ and n represent geometric shape coefficients related to crack type, location, and boundary conditions. These coefficients can be calibrated for specific unstable rock mass models through finite element numerical simulation. In this embodiment, the calibration methods for the geometric shape coefficients γ and n are as follows: 1) Establish a finite element model of the target unstable rock mass. The material parameters and boundary conditions are determined based on the actual exploration data.

[0039] 2) At the preset crack location, establish a series of models with crack depth ratios a / Hi = 0, 0.05, 0.10, 0.15, 0.20, 0.25, and 0.30 respectively. Perform modal analysis on each model and extract the first natural frequency f(a).

[0040] 3) Using [f(a) / f0]2 as the ordinate and a / Hi as the abscissa, the nonlinear least squares method is used to fit the formula y=1-γ·xn to obtain the optimal parameters γ and n.

[0041] 4) The fitting results must meet the requirements of a determination coefficient R2 ≥ 0.95 and a root mean square error RMSE ≤ 0.02. Otherwise, the mesh near the crack should be refined or the number of sample points should be increased.

[0042] Furthermore, the equivalent crack depth is substituted into the fracture mechanics formula to calculate the equivalent fracture strength factor. The specific derivation process is as follows: According to classical linear elastic fracture mechanics, the basic formula for the stress intensity factor of a member containing an edge crack is: in, Indicates the stress intensity factor. This represents the far-field nominal stress experienced by the unstable rock mass. Indicates the crack depth. This indicates the characteristic size of the unstable rock mass in the direction of the crack. This represents the shape correction factor related to the relative depth of the crack. In engineering practice, the instability of unstable rock masses is mainly caused by their own weight, and the nominal stress caused by the weight of the unstable rock mass is... It can be considered to remain constant. Therefore, when the crack depth extends to the critical state that causes macroscopic through-fracture instability of the rock mass, the crack depth reaches the critical value, and the stress intensity factor at this time reaches the fracture toughness of the rock mass. in, This represents the fracture toughness at the critical state. The critical value representing the crack depth; When the crack depth extends to the critical instability state When the stress intensity factor reaches the rock mass fracture toughness To eliminate unknown nominal stress and complex shape coefficients The effect will affect the stress intensity factor of the current state. Fracture toughness at critical state By analogy and using the aforementioned relationship between frequency and crack depth for algebraic substitution, the formula for calculating the equivalent fracture strength factor is finally derived as follows: in, Represents the equivalent fracture strength factor. Indicates the fracture toughness of the rock mass, ( Specifically, this includes drilling in-situ rock cores near the main structural plane of the unstable rock mass, processing them into standard samples for experimental testing, or utilizing the uniaxial compressive strength (UCS) of the rock using widely validated formulas. (Calculated using ≈0.1×UCS) This indicates the predicted frequency corresponding to the critical moment of rock mass instability (crack penetration) determined through numerical simulation or engineering judgment.

[0043] Predicted frequency The quantitative determination process is as follows: 1) Numerical simulation prediction: Establish an extended finite element model of the target unstable rock mass to simulate the entire process of crack propagation from its initial length to penetration, track the natural frequency at each step, and use the frequency corresponding to the crack penetration moment as... The grid convergence verification requires a frequency variation of less than 2%.

[0044] 2) Formula calculation: When there is no condition to perform a detailed simulation, the following formula is used: Calculation, where Take a value of 0.7 to 0.9 (determined based on the proportion of rock bridge), and indicate an error range of ±15%.

[0045] 3) On-site inversion correction: If the unstable rock mass has undergone macroscopic failure, the measured frequency at the last moment before failure will be used as the reference frequency. And use it to replace or correct the numerical simulation results.

[0046] 4) Final output: The confidence interval should be provided, and the applicable conditions should be clearly stated. When the measured frequency is less than 1.1 times... At that time, the system automatically increases the monitoring frequency and enters an early warning state. The specific methods for obtaining the damping ratio include: At least four three-component accelerometers are deployed at different locations on the unstable rock mass and on the stable bedrock to form a monitoring array and collect multi-channel mixed signals; After system installation, pure noise data was collected during a period without abnormal events. FastICA blind source separation was performed on the pure noise data to obtain a set of independent pure noise components. The average value of the kurtosis and spectral entropy of all independent pure noise components was calculated as the initial value. and And calculate their respective standard deviations. and ; An improved FastICA algorithm is used to separate the real-time acquired multi-channel mixed signal matrix X, obtaining the independent component matrix Y and the separation matrix W (satisfying Y=WX). The specific steps for separating the real-time acquired multi-channel mixed signal matrix X using the improved FastICA algorithm include: 1) Preprocessing: The multi-channel mixed signal matrix X is centered and whitened to obtain the whitened signal Z.

[0047] 2) Noise baseline initialization: During the initial installation of the acquisition equipment, a pure background noise signal is acquired, and a set of independent noise components is obtained by performing conventional FastICA separation. The mean kurtosis Kb, mean spectral entropy Hb, and standard deviations σK and σH are calculated. These are used for the standardization determination of subsequent components.

[0048] 3) Component-by-component separation and feature calculation: The FastICA fixed-point iterative algorithm is used to separate independent components one by one. After each component Ci is separated, the kurtosis Kurt(Ci), spectral Ent(Ci), and normalized offset relative to the noise baseline of that component are immediately calculated. , 4) Component classification and processing: Class A: Potentially useful pulse events, with the corresponding judgment criteria being: >3 and <1, mark and retain, store in the event library, and do not participate in subsequent separation boots.

[0049] Type B: Potential structural resonance response, corresponding to the criterion 1 < ≤3 and <-2, marked as structural response. Key improvement: Extract the column vector of the mixing matrix (spatial mode shape) corresponding to this component, use its orthogonal complement space as a constraint, and guide the algorithm to preferentially search for other modal components that are of the same origin as the current component in the remaining mixed signal.

[0050] Class C: Background noise or interference is judged if none of the above conditions are met, and the object is directly discarded and not proceeded to the next separation stage.

[0051] 5) Subsequent separation guided by orthogonal complement space: For the B-type component, let its corresponding mixing matrix column vector be aj, and calculate its orthogonal complement projection matrix. The remaining mixed signal is projected into this orthogonal complement space, and then FastICA iterations are continued in the projection space. This ensures that subsequently separated components are orthogonal to the already identified structural response components, avoiding repeated separation of the same mode.

[0052] 6) Iterate until all components are separated: Repeat steps 3-5 until the number of extracted independent components equals the number of sensor channels, or the kurtosis-entropy characteristics of the remaining signal all fall into Class C (noise) range.

[0053] For each isolated independent component in the independent component matrix Calculate its normalized offset relative to the initial noise value: kurtosis normalization offset: Spectral entropy normalization shift: in, and Representing components respectively kurtosis and spectral entropy; The 3σ rule is used to perform initial screening and adaptive control of the separated independent components based on joint criteria. 1) If both conditions are met If so, it is marked as a potentially useful pulse event and reserved; 2) If both conditions are met The initial screening labels are the set of potential structural resonance response components. Based on the orthogonal complement space of this component, the algorithm is guided to search for other modal responses of the same structure in subsequent separations; 3) If the above conditions are not met and the component exhibits low kurtosis and high spectral entropy, it is identified as diffuse background noise and removed. The algorithm actively suppresses subsequent separations with similar statistical characteristics to this component. When the characteristics of the components separated by continuous iterations are all stable within the noise range, the algorithm terminates early. The pseudo-inverse of the separation matrix W is calculated to obtain the mixing matrix A, which represents the potential structural resonance response component for each initial screening marker. Extract the j-th column vector from the mixing matrix A. This component serves as the measured spatial mode shape vector; Calculate the vector of each measured spatial mode shape one by one. Compared with the theoretical low-order vibration mode vector of the unstable rock mass obtained in advance through numerical simulation Modal confidence criterion (MAC) values ​​between: in, This represents the modal confidence criterion value. Let represent the transpose of the vector in the j-th column of the mixing matrix A. This represents the low-order vibration mode vector of the unstable rock mass theory. This represents the j-th column vector in the mixing matrix A. The transpose of the low-order mode vectors in the theory of unstable rock masses; If the MAC is greater than or equal to the preset threshold, the component is determined to have passed the spatial physical rationality test and is retained as a valid potential resonance component; if it is less than the preset threshold, it is determined to be a pseudo-mode or noise and is removed. Resonance sparse decomposition is performed on the potential resonance components that pass the test. The quality factor range of the decomposition is adaptively set according to the spectral peak of the separated components. Fine decomposition is performed around the main frequency band of the unstable rock mass. The high resonance components decomposed must satisfy that the coherence function of the adjacent sensor signals is greater than 0.9 at the resonance frequency. A pure resonance response signal with non-resonance interference removed is obtained. Based on the enhanced pure resonance response signal, the damping ratio was initially calculated using the random subspace identification method as the initial value. With minimizing the error between the theoretical frequency response function and the actual enhanced signal frequency response function as the objective function, the following three physical constraints are established: 1) Where C is the coefficient of internal friction of the material. The absolute temperature of the environment. The frequency of the vibration signal; 2) The statistically permissible error between the single-cycle energy dissipation ratio calculated from the damping ratio and the envelope attenuation rate obtained by demodulating the enhanced signal through the Hilbert transform is less than 15%; 3) When the unstable rock mass is in a stable structural state, the rate of change of the damping ratio of the same order mode within adjacent time windows is less than the set threshold. Substituting the initial value of the damping ratio into the equation, the objective function is solved using a sequential quadratic programming algorithm to obtain the optimal solution that satisfies the above constraints, which is the true damping ratio after removing the disturbance. .

[0054] In this embodiment, the Sigmoid nonlinear function is further used as the nonlinear transformation function to convert the relative rate of change into the damage contribution of the dynamic parameter to the stability of the unstable rock mass. in, Indicates the degree of damage contribution. Indicates the value of the changing trend. This represents the slope factor, used to control the steepness of the curve. This represents the trend value corresponding to the midpoint of the transformation.

[0055] The slope factor k is determined as follows: 1) Through indoor rock mechanics tests, paired data of the trend value of the dynamic parameter RP and the actual damage degree D were obtained, and the results were fitted using the Sigmoid function. .

[0056] 2) Using the scale correction formula: Calculate the initial values ​​at the site, where The initial damage coefficient is denoted as .

[0057] 3) In the initial stage of monitoring, with Using the initial values, inversion is performed using data from the stable phase to keep CP(t) close to 0 and without abrupt changes, ultimately obtaining the specific values ​​for this unstable rock mass. , which is the slope factor k. The points representing the significant impact of changes in this parameter are determined based on historical data: Convert midpoint The method for determining it is as follows: 1) Obtain paired data of the dynamic parameter variation trend value RP(t) and the corresponding actual damage degree D(t) through indoor rock mechanics tests or high-fidelity numerical simulations.

[0058] 2) Using RP as the x-axis and D as the y-axis, apply the Sigmoid function. Perform nonlinear least squares fitting, and obtain the fitted result That is, the midpoint of the transformation. .

[0059] 3) The fitting results must meet the requirement that the coefficient of determination R2 ≥ 0.90; otherwise, data points should be added or the experimental process should be checked.

[0060] In this embodiment, the step of determining the slope factor further includes: Loading failure tests were conducted on similar rock samples of the target unstable rock mass, and all dynamic parameters were acquired simultaneously at high frequency. The entire test process, from loading to failure, is divided into multiple mechanical stages (elastic, crack initiation, stable propagation, and unstable propagation), and the actual damage level of each mechanical stage is calibrated (for example, the damage in the elastic stage is defined as 0, and the damage at macroscopic fracture is defined as 1). A scatter plot was created with the trend values ​​of the experimentally measured parameters on the x-axis and the calibrated actual damage level on the y-axis. The plot was then fitted using the Sigmoid function to obtain the experimental values. ; Using physical correction formulas to adjust experimental values After making corrections, the physical correction value is obtained: in, Indicates experimental values. Indicates the physical correction value. Indicates the volume of the laboratory sample. This indicates the measured volume of the unstable rock mass at the site. This represents the initial damage coefficient of the unstable rock mass relative to the sample. (Value greater than 1, obtained through on-site investigation).

[0061] The initial damage coefficient reflects the degree of damage deterioration of natural joints and fissures within the unstable rock mass at the site compared to intact small-sized laboratory samples. The specific method for obtaining its value through field investigation is as follows: The method for determining Cd is as follows: Based on the "Engineering Rock Mass Classification Standard", the basic quality index BQsite and integrity coefficient Kv of the rock mass were calculated through on-site acoustic wave testing and point load testing.

[0062] 2) Arrange survey lines on the surface of the unstable rock mass, count the number of joints and fissures per unit volume Jv, and calculate the joint influence coefficient Fj=1+Jv / 10.

[0063] 3) Based on the degree of weathering at the site, select the weathering influence coefficient Fw according to the following table: Table 1. Values ​​of Weathering Influence Coefficient 4) Comprehensive calculation of Cd: in, As a benchmark quality indicator for intact, fresh rock, These are the basic quality indicators measured on-site for the rock mass.

[0064] After the monitoring system is installed, during the initial stable phase of the unstable rock mass (1-3 months), the time series of fluctuations in various dynamic parameters under natural conditions are acquired, with physical correction values. To optimize the variables, an inversion calculation was performed with the goal of ensuring that "the calculated damage contribution Cp(t) remains close to 0 throughout the stable phase without producing spurious mutations or trends." Once the rock mass enters the separation stage, the changes in significant parameters actually monitored are used to verify whether the Cp(t) curve calculated by the model is smooth and grows reasonably. The k value corresponding to the smoothness and reasonable growth of the Cp(t) curve is the final value.

[0065] In this embodiment, a stage-adaptive weighting mechanism is further introduced. Based on the current mechanical evolution stage of the unstable rock mass, dynamic weights are assigned to the dynamic parameters. Specific steps include: The mechanical evolution stage S i Including the stable phase Separation stage Destruction phase ; When a rock mass is in the separation or failure stage, its corresponding dynamic weight is... The calculation steps include: (1) Calculate the average relative rate of change in different stages. Calculate the dynamic parameters P(f, ξ, EAE, H, K) eff S in its mechanical evolution stage i (S1 represents stability, S2 represents separation, S3 represents disruption) The average value of all data points within the range. And calculate its initial value relative to the steady-state phase. average relative rate of change , The larger the value, the higher the dynamic parameter is in stage S. i The greater the range of change: in, Indicates the average relative rate of change. Representing the stages of mechanical evolution The average of all data points within the range. Indicates the initial value during the stable phase; (2) Calculation of parameter “inter-stage discrimination” The ability of parameter P to distinguish between two adjacent stages is calculated using Cohen's d effect size. Taking the distinction between the stable stage S1 and the separation stage S2 as an example, the discrimination is: The stage-to-stage differentiation of the dynamic parameters between two adjacent mechanical evolution stages is calculated using the following formula: in, Let be the inter-stage discrimination between the i-th and j-th target stages. Representing the stages of mechanical evolution The average value of all data points within the range, where i and j represent the index numbers of two adjacent mechanical evolution stages. Representing the stages of mechanical evolution The average of all data points within the range. This represents the pooled standard deviation. The greater the inter-stage discrimination, the greater the contribution of this parameter to distinguishing between the two stages.

[0066] Combining the average relative rate of change and the inter-stage discrimination, calculate the comprehensive sensitivity coefficient of the kinetic parameters in the failure stage and the separation stage: For the "separation" phase, its sensitivity coefficient simultaneously reflects its indicative role in both "de-stability" and "movement towards destruction": For the "destruction" stage, its sensitivity coefficient mainly reflects its indicative role in "final instability": in, , These represent the combined sensitivity coefficients for the separation and destruction stages, respectively. , and This represents the adjustment factor, initially set to 1 / 3, and subsequently fine-tuned based on the emphasis placed on amplitude and inter-stage differentiation. , These represent the average relative rates of change during the separation and destruction phases, respectively. For the aforementioned mechanical evolution stage, the comprehensive sensitivity coefficients of all dynamic parameters are normalized. For the separation stage... and destruction phase After calculating the comprehensive sensitivity coefficient of each parameter, the comprehensive sensitivity coefficients of all extracted dynamic parameters are normalized to obtain the objective dynamic weights corresponding to each dynamic parameter within the target mechanical stage. in, Represents dynamic parameters In the target mechanics evolution stage S i The corresponding objective dynamic weights, Represents dynamic parameters In the mechanical evolution stage S of this target i The overall sensitivity coefficient, This means summing up the comprehensive sensitivity coefficients of all extracted dynamic parameters in the fusion model at this target stage.

[0067] When the unstable rock mass is in a stable stage At that time, the dynamic weight of its corresponding stage The calculation steps include: 1) Collect initial data during the stable phase Acquire monitoring data when the unstable rock mass is in an absolutely stable state, either in the initial stage after system installation or during a risk-free period after long-term verification, continuously collect data for a period of no less than 7 days, and obtain the coefficient of variation (CV) for each parameter. P The 95% confidence interval width is less than 0.05 times its mean, forming a steady-state training dataset, and m data points were collected for each dynamic parameter.

[0068] 2) Calculate each dynamic parameter according to the following formula. Coefficient of variation: in, Represents the coefficient of variation. This represents the mean of the data. Indicates standard deviation; 3) Normalization determines the final weights Since a smaller CV value indicates greater stability, its reciprocal, 1 / CVP, is used as the basis for weight allocation. A larger CV value indicates greater parameter stability. The reciprocals of the coefficients of variation for each kinetic parameter are taken and normalized to obtain the kinetic parameters. Dynamic weights during the stable phase : in, Represents dynamic parameters During the stable phase The dynamic weights, where N represents the total number of dynamic parameters. Represents the coefficient of variation. The inverses of the coefficients of variation of all extracted kinetic parameters in the fusion model are summed.

[0069] In this embodiment, further, in step S5, the damage contribution of the dynamic parameters is fused with their corresponding dynamic weights to calculate the comprehensive dynamic stability index of the unstable rock mass at the current moment. Specific steps include: Based on the following DSI calculation formula, the damage contribution of the dynamic parameters and their corresponding dynamic weights are integrated to calculate the comprehensive dynamic stability index of the unstable rock mass at the current moment: in, Cp(t) represents the comprehensive dynamic stability index at time t, and Cp(t) represents the dynamic parameters. The damage contribution, obtained at time t through the Sigmoid function transformation, is used to describe the nonlinear effect of parameter changes on the stability of the unstable rock mass. Represents dynamic parameters The dynamic weights that actually participate in the calculation at time t.

[0070] Specifically, when the unstable rock mass is in a defined single stage, the aforementioned Directly take the objective dynamic weight corresponding to this stage, that is, take the weight for the stable stage. Separation stage Destruction phase When a rock mass undergoes a phase transition and occurs within a smooth transition window, the aforementioned To pass through the transition factor The calculated real-time transition weights.

[0071] In this embodiment, furthermore, when it is determined that the unstable rock mass meets the conditions and a stage transition occurs, a smooth transition calculation is performed on the dynamic weights of each dynamic parameter. The specific steps include: The moment when the recording phase switching criterion is first met is the transition start point t. s And preset the transition time length t t The transition endpoint is determined as follows: in, Indicates the end point of the transition. Indicates the starting point of the transition. Indicates the length of the transition time; The transition window period for phase switching is determined as follows: ; During the transition window period, the transition factor is calculated according to the following formula: in, Indicates the transition factor. This represents the slope factor, used to control the steepness of the curve; Using the aforementioned transition factor, for each dynamic parameter P, the real-time transition weight at the time of phase switching is calculated according to the following formula: in, Indicates the real-time transition weight. This represents the dynamic weights before the phase transition occurs. The dynamic weights after the phase transitions occur, including the stable phase, the separation phase, and the destruction phase.

[0072] Within the preset transition window period for determining the stage of unstable rock mass occurrence, the real-time transition weights calculated above will be applied. Substituting this into the calculation formula for the comprehensive dynamic stability index, replacing the original fixed-stage weights, i.e. .

[0073] Taking the calculation of transition weights from the stabilization to the separation phase as an example, the flowchart of weight vector replacement in this embodiment is as follows: Figure 2 As shown, it includes: Preset stable phase weight set Separation phase weight set and destruction phase weight set Each weight set includes the dynamic weights for its corresponding stable phase. Dynamic weights in the separation phase Dynamic weights during the destruction phase .

[0074] Based on the triggering of the stage criterion, the real-time weight vector W(t) is calculated according to the following rules: 1) If no stage criterion is triggered, the system is considered to be in a stable stage. In this case, there is no transition, and W(t) is directly applied: ; 2) If the separation phase criterion is triggered for the first time, it is determined that the transition period from stability to separation has begun, and the first transition factor is activated. A smooth incrementing process from 0 to 1, and according to... Calculate the weights; 3) When If the process remains in the separation phase, then the transition is complete, and the real-time weights are locked as: W(t) = ; 4) If the destruction stage criteria are further triggered during the separation stage, the transition period from separation to destruction is determined, and the second transition factor is activated. The smooth increment process from 0 to 1, with real-time weights replaced by: ; 5) When Then, the weights are locked as follows: = .

[0075] In the above rules, the transition factor and The pattern of change is determined according to the steps described above for "calculating the transition factor".

[0076] During the transition window period, the system continuously verifies the phase switching determination conditions. If the determination conditions are continuously met, the transition factor... The weights increase normally over time until a smooth transition to the target mechanical stage is completed; if the criterion fails within the window period, an adaptive rollback mechanism is triggered. In this embodiment, the update process for failure within the window period is as follows: Figure 3 As shown, the specific steps include: 1) Contribution to damage C P (t) Set the damage confirmation threshold C fir During the monitoring process, the maximum value ever reached by the damage contribution of each parameter was recorded in real time; 2) When the criterion becomes invalid within the window period, the transition factor shall be updated according to the following rules. : ①If there was any parameter C in history P (t) is greater than the damage confirmation threshold C fir If so, then it is determined that actual damage has occurred. At this point, Only rollback to a historical maximum value is allowed. The lower bound of the decision is: ,in This is the preset fluctuation buffer amount; ②If C of all parameters in history P (t) has never exceeded the damage confirmation threshold C. fir If so, this trigger is determined to be an unverified interference. At this point, the over-factor... It can decrease back to zero without restriction.

[0077] In all cases, the historical maximum value is always recorded and updated during the real-time update process: .

[0078] In this embodiment, the adjustment coefficient is further determined by dynamically adjusting the objective function based on classification robustness. Specific steps include: A robust objective function is constructed to measure the effectiveness of stage division, quantifying the discriminative power and concentration of the Dynamic Stability Index (DSI) calculated for different mechanical stages of the unstable rock mass. in, Let represent the objective function for strong robustness, and let i represent the index number of the mechanical evolution stage, with values ​​1, 2, and 3 corresponding to the stable stage, separation stage, and failure stage, respectively. This represents the i-th stage of mechanical evolution. This indicates that it belongs to the stage of mechanical evolution. Sample size This represents the mean of the dynamic stability index for all samples within stage S. This represents the global mean of the dynamic stability index for all samples across all three phases. Indicating the stage of mechanical evolution The specific data sample index number, with values ​​1, 2, ... , This indicates that it belongs to the stage of mechanical evolution. The The actual calculated value of the dynamic stability index for each sample.

[0079] In the above function, the numerator is the sum of squared deviations between the DSI mean of each stage and the global DSI mean, calculated based on the sample size of that stage. The larger the weighted value, the more significant the differences between stages. The denominator is the sum of squared deviations of all samples from the mean DSI of their respective stages. The smaller the value, the more concentrated the samples are within the same stage.

[0080] Establish constraints based on the aforementioned adjustment coefficient: Under the constraints, optimization calculations are performed with the goal of maximizing the robust objective function, and the optimal weight coefficient combinations A1*, A2*, and A3* that maximize the inter-stage differences and the intra-stage concentration are obtained. Substitute the optimal weight coefficient combination A1*, A2*, A3* into the calculation formula of the comprehensive sensitivity coefficient of the separation stage and the destruction stage to calculate the optimal comprehensive sensitivity coefficient of each parameter in the corresponding stage. The optimal comprehensive sensitivity coefficient is normalized to obtain the final objective weight set of each dynamic parameter at the corresponding stage.

[0081] In this embodiment, further, after obtaining the dynamic stability index, the method includes comparing the dynamic stability index with a preset threshold range to determine the current stability state and stage of the unstable rock mass. Specific steps include: Through numerical simulation and indoor mechanical tests, the critical threshold D1, warning threshold D2 and disaster threshold D3 of the unstable rock mass are preset; In response to DSI(t) ≤ D1, it is determined that the dangerous rock mass is in a stable stage, the rock mass is in an elastic state, the internal micro-fractures (acoustic emissions) are random and low-energy background activities, the dynamic parameters fluctuate normally near the initial values, and a blue safety signal is issued; In response to D1 < DSI(t) ≤ D2, it is determined that the dangerous rock mass enters the initial stage of the separation phase. Systematic damage begins to occur in the rock bridge or key structural plane. The micro-fracture activity is significantly enhanced (CE / CH > 0.5), the natural frequency shows a continuous and irreversible decrease (Rf < -0.05), the overall stiffness begins to degrade, and a yellow warning is issued; In response to D2 < DSI(t) ≤ D3, it is determined that the dangerous rock mass enters the middle and late stages of the separation phase, the damage accumulates rapidly, and the main control crack expands rapidly. High-energy events appear in the acoustic emission, the spectrum is highly disordered, the natural frequency decreases rapidly, and there may be a critical sign of the damping characteristic changing from positive to negative, and an orange warning is issued; In response to DSI(t) > D3, it is determined that the dangerous rock mass enters the critical period from separation to failure. The rock bridge is close to penetration or the locking section of the main control plane is about to fail. There may be a "relatively quiet period" of acoustic emission (precursor of fracture), the damping ratio turns negative, and the equivalent fracture strength factor approaches the fracture toughness of the rock mass and is in a critical instability state, and a red alarm is issued.

[0082] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered by the scope of the claims and the description of the present invention.

Claims

1. A dynamic evaluation method for the stability of unstable rock masses based on the fusion of multi-source dynamic parameters, characterized in that, include: Collect multi-source monitoring signals of unstable rock masses and extract dynamic parameters; Using the initial values ​​of the dynamic parameters of the unstable rock mass under stable conditions as a benchmark, calculate the trend value of the change of the dynamic parameters relative to the initial values ​​of the dynamic parameters at the current moment; According to a preset nonlinear transformation function, the changing trend value is converted into the damage contribution of the dynamic parameter to the stability of the unstable rock mass; A stage-adaptive weighting mechanism is introduced to assign corresponding dynamic weights to the dynamic parameters based on the current mechanical evolution stage of the unstable rock mass. The damage contribution of the dynamic parameters is integrated with their corresponding dynamic weights to calculate the comprehensive dynamic stability index of the unstable rock mass at the current moment.

2. The dynamic evaluation method for the stability of unstable rock masses based on the fusion of multi-source dynamic parameters according to claim 1, characterized in that, The dynamic parameters include natural frequency, damping ratio, acoustic emission cumulative energy rate, signal spectral entropy, and equivalent fracture strength factor. Among them, environmental vibration signals are collected by a high-sensitivity accelerometer to extract the natural frequency and damping ratio; The rupture signal is acquired by acoustic emission or microseismic sensors to extract the cumulative acoustic emission energy and calculate the signal spectral entropy. The equivalent fracture strength factor is obtained by inversion based on vibration response or acoustic emission source mechanism.

3. The dynamic evaluation method for the stability of unstable rock masses based on the fusion of multi-source dynamic parameters according to claim 2, characterized in that, The methods for obtaining the equivalent fracture strength factor specifically include: The equivalent crack depth of the unstable rock mass is inverted based on the natural frequency of the current cracked rock mass: in, Indicates the equivalent crack depth. This indicates the characteristic size of the unstable rock mass in the direction of the crack. This represents the measured natural frequency of the current cracked rock mass. The reference natural frequency of the intact rock mass is represented by γ, and the geometric shape factors are represented by n. Substituting the equivalent crack depth into the fracture mechanics formula, the equivalent fracture strength factor is calculated: in, Represents the equivalent fracture strength factor. Indicates the fracture toughness of the rock mass. This indicates the predicted frequency corresponding to the critical moment of rock mass instability.

4. The dynamic evaluation method for the stability of unstable rock masses based on the fusion of multi-source dynamic parameters according to claim 1, characterized in that, The Sigmoid nonlinear function is used as the nonlinear transformation function to convert the changing trend value into the damage contribution of the dynamic parameter to the stability of the unstable rock mass. in, Indicates the degree of damage contribution. Indicates the value of the changing trend. Indicates the slope factor. This represents the trend value corresponding to the midpoint of the transformation.

5. The dynamic evaluation method for the stability of unstable rock masses based on the fusion of multi-source dynamic parameters according to claim 4, characterized in that, The steps for determining the slope factor include: Loading failure tests were conducted on similar rock samples of the target unstable rock mass, and all dynamic parameters were collected simultaneously. The entire test process was divided into multiple mechanical stages, and the actual degree of damage in each mechanical stage was calibrated. A scatter plot was created with the trend values ​​of the experimentally measured parameters on the x-axis and the calibrated actual damage level on the y-axis. The plot was then fitted using the Sigmoid function to obtain the experimental values. ; Using physical correction formulas to adjust experimental values After making corrections, the physical correction value is obtained: in, Indicates experimental values, Indicates the physical correction value. Indicates the volume of the laboratory sample. This indicates the measured volume of the unstable rock mass at the site. This represents the initial damage coefficient of the unstable rock mass relative to the sample.

6. The dynamic evaluation method for the stability of unstable rock masses based on the fusion of multi-source dynamic parameters according to claim 1, characterized in that, An adaptive weighting mechanism is introduced to assign dynamic weights to the dynamic parameters based on the current mechanical evolution stage of the unstable rock mass. The specific steps include: The mechanical evolution stage S i Including the stable phase Separation stage Destruction phase ; When a rock mass is in the separation or failure stage, its corresponding dynamic weight is... The calculation steps include: Calculate the average value of the dynamic parameter across all data points during its mechanical evolution phase, and calculate its average relative rate of change with respect to the initial value in the steady-state phase: in, Indicates the average relative rate of change. Representing the stages of mechanical evolution The average of all data points within the range. Indicates the initial value during the stable phase; The stage-to-stage differentiation of the dynamic parameters between two adjacent mechanical evolution stages is calculated using the following formula: in, Let be the inter-stage discrimination between the i-th and j-th target stages. and These represent the stages of mechanical evolution. or The average of all data points within the range. This represents the pooled standard deviation; Combining the average relative rate of change and the inter-stage discrimination, calculate the comprehensive sensitivity coefficient of the kinetic parameters in the failure stage and the separation stage: in, , These represent the combined sensitivity coefficients for the separation and destruction stages, respectively. , and This represents the adjustment coefficient. , These represent the average relative rates of change during the separation and destruction phases, respectively. Normalize the overall sensitivity coefficients of all dynamic parameters to obtain the dynamic weights corresponding to each dynamic parameter in this stage: in, This represents the dynamic weight of the i-th dynamic parameter. This represents the comprehensive sensitivity coefficient of the dynamic parameters at the target stage; When the unstable rock mass is in a stable stage, its corresponding dynamic weight The calculation steps include: The monitoring data of the unstable rock mass during the period when it was in an absolutely stable state were obtained to form a stable state training dataset; The coefficients of variation for each dynamic parameter are calculated using the following formula: in, Represents the coefficient of variation. This represents the mean of the data. Indicates standard deviation; The dynamic weight of each dynamic parameter in the steady-state phase is obtained by taking the reciprocal of its coefficient of variation and normalizing it. : in, Represents dynamic parameters During the stable phase The dynamic weights, where N represents the total number of dynamic parameters. This represents the coefficient of variation.

7. The dynamic evaluation method for the stability of unstable rock masses based on the fusion of multi-source dynamic parameters according to claim 6, characterized in that, In step S5, the damage contribution of the dynamic parameters is fused with their corresponding dynamic weights to calculate the comprehensive dynamic stability index of the unstable rock mass at the current moment. Specific steps include: Based on the following DSI calculation formula, the damage contribution of the dynamic parameters and their corresponding dynamic weights are integrated to calculate the comprehensive dynamic stability index of the unstable rock mass at the current moment: in, Cp(t) represents the comprehensive dynamic stability index at time t, and Cp(t) represents the dynamic parameters. The damage contribution obtained at time t through the Sigmoid function transformation. Represents dynamic parameters The real-time dynamic weights that actually participate in the calculation at time t.

8. The dynamic evaluation method for the stability of unstable rock masses based on the fusion of multi-source dynamic parameters according to claim 7, characterized in that, When it is determined that the dangerous rock mass meets the conditions and undergoes a stage transition, a smooth transition calculation is performed on the dynamic weights of each dynamic parameter. The specific steps include: The moment when the recording phase switching criterion is first met is the transition start point t. s And preset the transition time length t t The transition endpoint is determined as follows: in, Indicates the end point of the transition. Indicates the starting point of the transition. Indicates the length of the transition time; The transition window period for phase switching is determined as follows: ; Within the transition window period, calculate the transition factor according to the following formula: in, Indicates the transition factor. Indicates the slope factor; Use the transition factor to calculate the real-time transition weight at the time of stage transition according to the following calculation formula in, Indicates the real-time transition weight. This represents the dynamic weights before the phase transition occurs. This indicates the dynamic weight after a phase transition occurs; When it is determined that the dangerous rock mass meets the conditions and undergoes a stage transition, within the preset transition window period, substitute the real-time weight into the DSI calculation formula to replace the dynamic weight for calculating the comprehensive dynamic stability index.

9. The dynamic evaluation method for the stability of unstable rock masses based on the fusion of multi-source dynamic parameters according to claim 6, characterized in that, The adjustment coefficient is dynamically adjusted and determined by optimizing the objective function based on classification robustness. The specific steps include: Construct a strong robustness objective function for measuring the stage division effect: in, Let represent the objective function for strong robustness, and let i represent the index number of the mechanical evolution stage, with values ​​1, 2, and 3 corresponding to the stable stage, separation stage, and failure stage, respectively. This represents the i-th stage of mechanical evolution. This indicates that it belongs to the stage of mechanical evolution. Sample size This represents the mean of the dynamic stability index for all samples within stage S. This represents the global mean of the dynamic stability index for all samples across all three phases. Indicating the stage of mechanical evolution The specific data sample index number, with values ​​1, 2, ... , This indicates that it belongs to the stage of mechanical evolution. The The actual calculated value of the dynamic stability index for each sample; Establish a constraint condition based on the adjustment coefficient: Under the constraint condition, perform optimization calculation with maximizing the strong robustness objective function as the solution target, and solve to obtain the optimal weight coefficient combination A1*, A2*, A3* that maximizes the difference between stages and the concentration within stages; Substitute the optimal weight coefficient combination into the stage comprehensive sensitivity coefficient calculation formula corresponding to the separation stage and the failure stage, and calculate the optimal comprehensive sensitivity coefficient of each parameter in the corresponding stage; Normalize the optimal comprehensive sensitivity coefficient to obtain the final objective weight set of each dynamic parameter in the corresponding stage.

10. The dynamic evaluation method for the stability of unstable rock masses based on the fusion of multi-source dynamic parameters according to claim 1, characterized in that, After obtaining the dynamic stability index, it further includes comparing the dynamic stability index with a preset threshold range to determine the current stable state and the stage where the dangerous rock mass is located. The specific steps include: Preset the critical threshold D1, warning threshold D2, and disaster threshold D3 of the dangerous rock mass through numerical simulation and indoor mechanical tests; In response to DSI(t) ≤ D1, determine that the dangerous rock mass is in the stable stage and issue a blue safety signal; In response to D1 < DSI(t) ≤ D2, determine that the dangerous rock mass enters the initial stage of the separation stage and issue a yellow warning; In response to D2 < DSI(t) ≤ D3, determine that the dangerous rock mass enters the middle and late stages of the separation stage and issue an orange warning; In response to DSI(t) > D3, determine that the dangerous rock mass enters the critical period from separation to failure and is in a critically unstable state, and issue a red alarm.

Images