Rock fracture prediction method based on cooperative monitoring of acoustic emission and surface strain

Abstract

The invention relates to the technical field of rock mechanical tests and safety monitoring, and discloses a rock fracture prediction method based on acoustic emission and surface strain cooperative monitoring, and the method comprises the steps: building a monitoring system comprising acoustic emission and an industrial camera array, and building a unified timestamp of each subsystem through a GPS time service module; through load triggering logic, a synchronous trigger is utilized to synchronously acquire an acoustic emission signal and a sample surface image sequence, and evolution characteristics of three-dimensional coordinates of an acoustic emission source and surface full-field strain data are solved. And on the basis, calculating a main strain field variation coefficient, and performing cross-correlation analysis on the resampled energy rate and strain rate. And finally, according to multi-parameter coupling criteria of acoustic emission time sequence parameters, positioning events and surface strain, identifying rock fracture precursor types, and predicting fracture moments and areas by combining a Voight model inverse velocity method and a seismic source projection technology.

Description

Technical Field

[0001] This invention relates to the field of rock mechanics testing and safety monitoring technology, specifically a rock fracture prediction method based on the synergistic monitoring of acoustic emission and surface strain. Background Technology

[0002] In deep underground engineering and geotechnical engineering, the instability and fracturing of rock materials are often accompanied by a violent release of energy. Accurate prediction of its failure characteristics is of great significance for disaster prevention and mitigation in engineering projects. Currently, indoor rock mechanics testing is the main method for studying the mechanism of rock damage evolution. Commonly used non-contact monitoring technologies mainly include acoustic emission monitoring technology and digital image correlation technology. Acoustic emission technology focuses on capturing elastic wave signals generated by the propagation of microcracks inside the rock, which can reflect the evolution process of internal damage; digital image correlation technology focuses on full-field non-contact deformation measurement, which can intuitively present the characteristics of strain localization zones on the rock surface.

[0003] However, existing technologies have certain limitations when using the two methods for joint monitoring. Acoustic emission systems and DIC optical systems typically operate as independent devices, each with its own independent data acquisition clock and triggering mechanism. In actual experiments, data alignment often relies on manually recorded start times or software-level post-fitting, lacking a unified microsecond-level time base at the hardware level. Due to the startup response delays and internal clock drift of different devices, high-frequency acoustic emission signals and low-frequency image data cannot be precisely correlated on the time axis, making it difficult to establish a reliable time-series correlation in the microsecond to millisecond-level transient rock fracturing process.

[0004] Furthermore, existing fracture prediction methods are mostly limited to threshold alarms for a single physical quantity. For example, they judge danger solely based on a sudden increase in acoustic emission ring count rate or an accelerated change in strain at a certain point, ignoring the physical mapping relationship between internal damage accumulation and surface macroscopic deformation. This single-perspective monitoring makes it difficult to quantify the driving process of internal micro-fractures propagating to the surface and is easily affected by environmental noise or local non-destructive deformation, leading to misjudgments. At the same time, traditional failure time prediction models are mostly based on regression analysis of single-point data, focusing on prediction in the time dimension, and failing to effectively perform geometric projection and superposition analysis of the spatial distribution of internal seismic sources and high-strain areas on the surface. Therefore, it is difficult to achieve simultaneous and accurate prediction of the location and specific time of macroscopic fracture occurrence. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a rock fracture prediction method based on the coordinated monitoring of acoustic emission and surface strain. This method solves the problems in existing rock mechanics tests where the lack of a unified hardware time base for heterogeneous monitoring equipment leads to time series deviations in multi-source data, and where a single physical criterion is difficult to effectively correlate internal damage with surface deformation, resulting in misjudgments.

[0006] To achieve the above objectives, the first aspect of the present invention provides a method for predicting rock fracture based on the coordinated monitoring of acoustic emission and surface strain, comprising the following steps: Constructing a collaborative monitoring system and preparing samples: A rectangular rock standard specimen was selected, and a random speckle field consisting of a matte white background layer and uniformly distributed matte black spots was constructed on the specimen surface to serve as a feature carrier for digital image correlation analysis. A hardware system including a loading device, an acoustic emission monitoring array, and an industrial camera array was constructed. Spatially non-coplanar and non-collinear acoustic emission sensors were arranged on the side of the specimen to acquire source data containing depth information; a high-resolution industrial camera with its optical axis orthogonal to the specimen surface was deployed to record the full-field deformation process of the specimen surface.

[0007] Establishing a unified time base for multi-source systems: Connect the GPS timing module to the control computers of each subset of the collaborative monitoring system via a 232 interface or a USB interface, install GpsTimer time synchronization software on each control computer, and establish a unified timestamp for the collaborative monitoring system.

[0008] Execute synchronous triggering and data retrieval: A continuous load is applied by a loading device, and a load triggering logic is used. When the monitored load first reaches a preset threshold, a synchronous trigger generates a pulse signal to trigger the acoustic emission monitoring system to start collecting acoustic emission signals at the sampling frequency. The industrial camera array then starts collecting surface images at the image acquisition frame rate, thereby realizing the synchronous acquisition and recording of multiple physical information.

[0009] Solving for acoustic emission and strain field characteristic parameters: Ringing counts of acoustic emission signals were collected, and the three-dimensional spatial coordinates of the acoustic emission source were iteratively solved using the damped least squares method. Simultaneously, a digital image correlation algorithm was used to track the pixel sub-region position changes of the random speckle field in the image sequence, calculate random measurement points and full-field displacement data on the sample surface, and generate full-field strain data and maximum shear strain field data through displacement field differentiation.

[0010] Precursor identification for multi-parameter coupling: Correlation analysis was performed on ring counts, source spatial coordinates, and full-field strain data. The principal strain field variation coefficient of the full-field strain data was calculated. This coefficient is based on the dispersion of the equivalent principal strains of the effective nodes in the full field and their mean values, and is used to characterize the degree of non-homogeneity of rock surface deformation.

[0011] Determining the strong spatiotemporal coupling characteristics: Cross-correlation analysis was performed on the acoustic emission energy rate sequence and the maximum principal strain rate sequence of the surface after resampling at a unified time step, and the cross-correlation coefficient and time lag were calculated. When the rate of change of the principal strain field coefficient of variation exceeded a set threshold, and the time lag was less than or equal to the unified time step, and the cross-correlation coefficient indicated strong coupling characteristics, a precursor to rupture was identified. Based on the abrupt change sequence of ringing counts and full-field strain data, precursors of internal crack activation, surface rupture, and critical instability were distinguished.

[0012] Predicting the region and timing of macroscopic rupture: The inverse velocity method of the Voight model was employed, using the total shear strain as the observed physical quantity. Regression analysis was performed on the linear relationship between the inverse strain rate and time to calculate the theoretical failure time. The spatial coordinates of the seismic source corresponding to the theoretical failure time were projected onto the specimen surface and spatially superimposed with the high-strain region in the total strain data. The overlapping region is the predicted macroscopic failure location.

[0013] A second aspect of the present invention provides a computer device, including a memory, a processor, and a communication module.

[0014] The memory stores a computer program, and when the processor executes the computer program, it implements the rock fracture prediction method based on the coordinated monitoring of acoustic emission and surface strain described in the first aspect above.

[0015] The technical solution provided by this invention eliminates the temporal deviation of multi-source data through a synchronous triggering mechanism of physical connections. It utilizes the coefficient of variation of the strain field to characterize the degree of deformation localization and quantifies the temporal correlation between internal damage release and surface deformation response through cross-correlation analysis. By combining the Voight model with three-dimensional positioning projection technology, it achieves dual prediction of the timing and spatial location of macroscopic rock fracture, improving the early warning accuracy of rock instability monitoring.

[0016] This invention provides a method for predicting rock fracture based on the coordinated monitoring of acoustic emission and surface strain, and a computer device thereof. It has the following beneficial effects: 1. This invention establishes a unified timestamp for different monitoring and acquisition devices through a GPS timing module. This mechanism places the load, acoustic emission signal, and image sequence on the same absolute time axis, ensuring the temporal relationship between internal micro-fractures and surface macro-deformation of rocks with microsecond-level precision, and solving the technical problem of accurately fusing multi-source monitoring data.

[0017] 2. This invention establishes a multi-parameter coupling criterion including the coefficient of variation of the principal strain field and the cross-correlation coefficient. By calculating the time lag between the acoustic emission energy rate and the surface strain rate after resampling, the driving characteristics of internal damage propagation to the surface are quantified. This method overcomes the limitations of monitoring a single physical quantity. By identifying the strong coupling stage between internal damage accumulation and surface deformation response, it effectively distinguishes different precursor types such as internal fracture activation, surface fracturing, and critical instability, thus improving the reliability of rock fracture prediction.

[0018] 3. This invention combines the Voight model inverse velocity method with source spatial projection technology. It uses total field cumulative shear strain data to estimate the theoretical failure time and projects the three-dimensional acoustic emission source coordinates onto the sample surface and high-strain regions for superposition analysis. This approach not only achieves linear regression prediction of the macroscopic fracture time of the rock but also pinpoints the specific fracture location through the spatial overlap between the internal source and the surface strain concentration area, providing a dual spatiotemporal early warning for the instability and failure of the rock sample. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating the overall process of an embodiment of the present invention. Figure 2 A schematic diagram of the ringing count time history curve for real-time acoustic emission of the present invention is shown. Figure 3 This is a schematic diagram of the spatial positioning configuration of the present invention; Figure 4 This is a schematic diagram illustrating the real-time calculation of strain data at measurement points based on images captured by an industrial camera, according to the present invention. Detailed Implementation

[0020] The technical solutions in 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 within the scope of protection of the present invention.

[0021] See attached document Figure 1 This invention provides a rock fracture prediction method based on the synergistic monitoring of acoustic emission and surface strain, comprising the following steps: S110, prepare rock samples with speckle fields, and construct a collaborative monitoring system including a loading device, an acoustic emission monitoring array, and an industrial camera array; S120 uses a GPS timing module to establish a unified timestamp for time synchronization configuration of the loading device, acoustic emission monitoring array and industrial camera array, and sets acoustic emission acquisition parameters and camera shooting parameters. S130 applies a continuous load to the rock sample and, based on the load triggering logic and synchronous trigger, synchronously triggers the acquisition of acoustic emission signals and sample surface image sequences. S140, collect and calculate the ring count and source spatial coordinates of the acoustic emission signal, and use digital image correlation algorithm to track the pixel sub-region position changes of the random speckle field in the sample surface image sequence, and calculate the random measurement points and full-field strain data of the sample surface. S150 uses correlation analysis of acoustic emission ringing counts, seismic source spatial location, and surface strain field evolution characteristics to identify rock fracture precursor types and predict fracture regions based on multi-parameter coupling criteria. The specific identification logic is as follows: Precursor to internal fracture activation: When a significant increase in acoustic emission ringing count rate is detected (e.g., the count rate exceeds 30% of the peak value within a unit time window), while there is no significant abrupt change in the overall strain data (the coefficient of variation of the principal strain field remains stable), it is determined that the damage mainly occurs inside the rock. At this point, the concentrated area of ​​internal damage is determined based on the three-dimensional kernel density estimation of the source spatial coordinates. Precursors to surface fracture: When significant localization of strain data is detected across the entire field (e.g., the maximum principal strain exceeds three times the average strain across the entire field, forming a shear band), and the acoustic emission signal response lags in time or has low energy, it is determined to be surface spalling or shallow crack propagation. In this case, the area of ​​concentrated surface deformation is determined based on the overall strain data. Critical instability precursor: When a burst of acoustic emission ringing count and the expansion of high-strain bands on the surface are detected to occur synchronously in time (cross-correlation coefficient greater than 0.85), it is determined to be a critical instability precursor to a through-crack.

[0022] Finally, by combining the inverse velocity method of the Voight model with the source projection technique, the overlapping area of ​​the source location cloud map and the surface strain cloud map is locked as the predicted location of the macroscopic rupture surface.

[0023] The following section will provide a detailed explanation of the implementation details of the above steps, using specific experimental parameters and calculation models.

[0024] In step S110, the specific implementation methods for constructing the collaborative monitoring system and preparing rock samples include the following sub-steps: S111, Preparation of rock specimens suitable for digital image correlation analysis. The rock mass is selected and machined into a cuboid geometry to provide four independent observation surfaces. The specimen size is set to length... ,Width ,high ,in and The value range is from 50mm to 100mm. The value range is from 100mm to 200mm, and satisfies and The upper and lower loading end faces and side surfaces of the sample were precision-machined using a grinding machine. The flatness tolerance of the upper and lower end faces was controlled to be less than 0.02 mm, and the parallelism error of the upper and lower end faces was controlled to be less than 0.05 mm, so as to ensure uniform stress distribution under axial loading. After machining, anhydrous ethanol was used to clean the surface dust and oil stains.

[0025] S112, a high-contrast artificial random speckle field was constructed on the sample surface. To address the issue that the natural grayscale differences in the rock surface texture are insufficient to support relevant calculations, a matte white aerosol paint was applied to the four observation surfaces of the sample to form a uniform and non-reflective white background layer. After the primer dried, matte black aerosol paint was randomly sprayed onto the white background layer. The distance between the nozzle and the sample surface, as well as the duration of nozzle pressing, were controlled to ensure that the diameter of the physical black speckles deposited on the sample surface was between 0.2 mm and 1.0 mm. This ensured that, under the subsequently selected camera field of view, the physical speckle diameter would map to an image sensor image size occupying 3 to 7 pixel units. The coverage of the black speckles on the white background was controlled between 40% and 60% to avoid large-area aggregation or voids.

[0026] S113, Deploy a three-dimensional acoustic emission sensor monitoring array. Select no fewer than eight resonant acoustic emission sensors, couple them to the sample surface using Vaseline or a dedicated acoustic coupling agent, and fix them with elastic clamps to maintain a constant contact pressure between the sensors and the sample surface. The sensors are arranged according to the principle of spatial non-coplanarity and non-collinearity, distributing them in the upper, middle, and lower regions of the sample's sides, with sensors on adjacent sides staggered to form a three-dimensional monitoring network surrounding the predicted fracture zone inside the sample. After installation, establish a spatial rectangular coordinate system with the geometric center of the sample as the origin. Use a vernier caliper with an accuracy of no less than 0.02 mm to measure or use a three-dimensional laser scanner to obtain the three-dimensional coordinates of the center point of each sensor in this coordinate system. ,in Indicates the sensor number.

[0027] S114, a multi-view industrial camera imaging array is installed. Four high-resolution industrial cameras are arranged at a 90-degree angle around the sample, each equipped with a fixed-focus, low-distortion lens. The camera mounts are adjusted so that the optical axes of each camera are orthogonal to the center of the corresponding sample surface. The object distance and focal length are adjusted to ensure that the camera's field of view completely covers the height and width of the sample, while retaining a 10% to 20% redundancy field of view along the upper and lower boundaries of the loading direction to prevent the target area from moving out of the frame during sample deformation. A DC-driven, flicker-free, high-brightness LED cold light source is used to supplement the illumination of the sample surface. The incident angle of the light source is adjusted to eliminate localized high-brightness reflections and shadows on the sample surface. The aperture and exposure time are corrected to ensure that the grayscale histogram of the acquired image follows a normal distribution and is located in the middle range of gray levels.

[0028] The S115 integrates a collaborative monitoring hardware system. The rock sample is placed at the center of the lower pressure plate of an electro-hydraulic servo rigid rock mechanics testing machine. An acoustic emission sensor is connected to a multi-channel acoustic emission acquisition instrument via a preamplifier, with the preamplifier's gain set to 40dB and its frequency response range set to 20kHz to 400kHz. Four industrial cameras are connected to the image acquisition workstation via gigabit Ethernet cables or high-speed data interfaces. A GPS timing module is connected to the loading device control computer, the acoustic emission acquisition instrument control computer, and the image acquisition workstation via RS232 or USB interfaces. Simultaneously, external synchronization triggers are installed, with their signal outputs connected to the trigger input ports of the loading device controller, the acoustic emission monitoring system, and the industrial camera array via BNC cables or I / O control lines, ensuring smooth physical connections between all control terminals.

[0029] In step S120, the specific operations for achieving time synchronization and parameter configuration of the multi-source system include the following sub-steps: S121, Establish a unified timestamp. Install GpsTimer synchronization software on each control computer of the loading device, acoustic emission monitoring system, and industrial camera array. Before the experiment begins, activate the GPS time synchronization module to receive satellite signals, open the GpsTimer synchronization software, and the software automatically parses the GPS signal and performs millisecond-level calibration on the system time of each control computer, thereby establishing a unified timestamp for the collaborative monitoring system. This operation ensures that all data files recorded by the devices are time-stamped based on Uniform Coordinated Universal Time (UTC).

[0030] S122, Determining the longitudinal wave propagation velocity in rock media. Two points with known coordinates are selected on the sample surface and marked as the source and receiver points, respectively. The straight-line Euclidean distance between the two points is denoted as . At the source point, a simulated sound source is excited using the Hsu-Nielsen lead-breaking method, i.e., a step pulse signal is generated by breaking a 0.5 mm diameter lead core.

[0031] Simultaneously, the acoustic emission sensor closest to the source point is set as the trigger sensor. When it receives a signal, the time is recorded and marked as relative zero in the software. The time when the sensor at the receiving point receives the first wave signal is also recorded. The lead-breaking test is repeated at least five times, and outliers are removed before taking the average value. The longitudinal wave velocity in the rock medium is obtained by calculating the ratio of the Euclidean distance between the source and receiving points to the average signal propagation time difference. During the calculation, the inherent time difference correction values ​​of the system, such as sensor response delay and cable transmission delay, need to be deducted.

[0032] S123, quantify ambient background noise and set the acoustic emission acquisition threshold. With the system powered on but in a silent state without any actual load applied, run the acoustic emission acquisition system to sample ambient noise. The system calculates the root mean square (RMS) voltage and peak voltage of each channel's analog signal in real time. Extract the maximum peak voltage during the monitoring period as the background noise reference. Set the sampling frequency of the acoustic emission signal (e.g., 1MHz-5MHz) and set the hardware trigger threshold (unit: dB). This threshold must be 6dB to 10dB higher than the background noise peak value to ensure that the signal from a small crack is acquired while shielding against electromagnetic interference from the mechanical hydraulic system.

[0033] S124, Configure the acquisition parameters of the industrial camera array. Based on the estimated uniaxial compressive strength of the rock sample and the preset loading strain rate, predict the maximum instantaneous surface deformation rate that the sample may experience before failure. Set the image acquisition frame rate parameters of the industrial camera array (e.g., 5fps to 30fps) to ensure that the number of effective image frames acquired is not less than the analysis requirements, typically set between 5fps and 30fps. After completing the parameter settings, keep the illumination, focus, and white balance constant.

[0034] In step S130, the detailed process of executing synchronous triggering and continuous data acquisition includes the following sub-steps: S131, switch between preloading and servo control modes. Start the hydraulic pump station of the rock mechanics testing machine and control the actuator to apply a preload to the specimen in force control mode. Set the loading rate to 0.5 kN / s until the axial load reaches the preset contact value. (This value is set to 0.5% to 1.0% of the estimated uniaxial compressive strength of the rock specimen; for example, for a standard specimen with an expected strength of 100 MPa, the contact value is set to 1 kN to 2 kN.) The load is held constant for 30 seconds to eliminate clamp clearance and establish a stable stress transfer path. After the holding phase, the controller performs a mode switch, seamlessly transitioning from force control mode to displacement control mode. Maintaining the current actuator position, the feedback variable is seamlessly switched from the load sensor signal to the displacement sensor signal, inheriting the current PID control parameters to prevent sudden load changes. The axial loading rate is set. To avoid the influence of inertial effects and meet the quasi-static loading conditions, The settings must strictly adhere to the following constitutive constraints: ; In the formula, The physical height of the specimen along the loading direction (mm); For quasi-static strain rate, the value range is limited to 10. 5 s 1 Up to 5×10 5 s 1 This ensures the internal stress balance of the specimen and the gradual nature of its failure process.

[0035] S132, execute the synchronous triggering logic. Connect the external synchronous trigger to the trigger ports of the loading device controller, the acoustic emission monitoring system, and the industrial camera array, respectively. After the loading program starts running, when the axial load detected by the loading device first reaches the preset trigger threshold (e.g., 1.2 times the contact value or a specific set value), the loading device sends a signal to the synchronous trigger, which then generates a synchronous pulse signal and distributes it to the acoustic emission monitoring system and the industrial camera array. After receiving the pulse, the acoustic emission monitoring system begins to continuously acquire the acoustic emission waveform stream according to the preset sampling frequency; after receiving the pulse, the industrial camera array begins to continuously record the image sequence of the sample surface according to the preset image acquisition frame rate.

[0036] See attached document Figure 2 S133, continuously acquire acoustic emission waveform stream and characteristic parameters. The acoustic emission system not only records the full waveform data stream but also extracts impact features in real time. Based on the signal attenuation curve obtained from the Hsu-Nielsen lead-breaking test in the previous steps, set the peak definition time (PDT), impact definition time (HDT), and impact latch-up time (HLT). Specifically, set HDT to be greater than the average time required for the signal to attenuate from the peak to the threshold, and set HLT to be greater than the attenuation time of the signal due to multipath effects and reciprocating reflections within the sample. Typically, HLT is set to ≥2ms to ensure that the oscillation waveform corresponding to a single micro-fracture is identified as a unique acoustic emission impact event (Hit). Based on the set threshold voltage, the system discretizes and statistically analyzes the continuous analog signal, and calculates the instantaneous acoustic emission impact rate and incremental count within a set differential time window (typically 0.5s to 1.0s) in real time.

[0037] S134, Acquire the time-series digital image sequence. After receiving the synchronization signal, the industrial camera array begins image acquisition. Raw image data is directly written to the RAID disk array via a high-speed bus. Each frame of the image is recorded with a precise system timestamp calibrated based on step S121, which serves as the basis for subsequent data alignment.

[0038] S135, Real-time Monitoring and Abnormal Interruption Protection. During loading and execution, the main control computer polls the status parameters of each subsystem in real time via shared memory. Dual protection logic is implemented: First, monitor the step change in the cumulative energy of acoustic emission. When the energy release rate increases exponentially within five consecutive time windows, it is marked as a precursor to critical instability. Secondly, the slope of the tangent line of the load-time curve is monitored. When the slope changes from positive to negative and the absolute value exceeds the stiffness reduction threshold, macroscopic fracture of the specimen is determined. Once any of the above conditions are met, the system automatically extends the acquisition time by 30 seconds to record the residual strength stage characteristics, and then synchronously stops loading and data acquisition via command. The design of the oil temperature cooling circuit and the overload relief valve structure of the hydraulic servo system are well-known technologies to those skilled in the art and will not be elaborated upon here.

[0039] See attached document Figure 3 In step S140, the technical means for calculating acoustic emission characteristic parameters and surface strain field data include the following sub-steps: S141, construct the travel time acquisition and localization equations for the acoustic emission signal. From the waveform data stream obtained in the previous steps, apply the AIC (Akaike Information Criterion) criterion to calculate the minimum value of the entropy function of the waveform sequence, which is used as the precise arrival time of the signal. (i.e., the first) The event reached the first (Time of each sensor). Based on the pre-calibrated three-dimensional spatial coordinates of each sensor. and the equivalent longitudinal wave velocity of the rock medium Assume the acoustic emission source is located at the coordinates to be determined. And the time of the earthquake is The following set of nonlinear positioning equations is established: ; In the formula, ,and This is to provide redundant observation data for error adjustment. The equations characterize the physical constraints between the geometric distance and time delay of sound waves propagating in an isotropic medium.

[0040] S142, source iterative solution based on nonlinear least squares method. Constructing the objective residual function. This function characterizes the weighted sum of squared errors between the theoretically calculated time and the actually observed time: ; In the formula, These are weighting coefficients, and their values ​​are related to the first... The peak amplitude of the channel signal is proportional to the signal strength to reduce the weight of timing errors caused by weak signals.

[0041] In the solution process, first select The sensor coordinates with the smallest value are used as the initial values ​​for iteration. Damped least squares (Levenberg-Marquardt algorithm) is used to evaluate the objective function. Perform iterative optimization until the residuals converge to obtain the spatiotemporal solution of the acoustic emission source. After the solution is completed, the covariance matrix is ​​calculated. If the semi-major axis of the positioning error ellipsoid exceeds the preset accuracy threshold (e.g., 5mm), the event is marked as a discrete invalid event and discarded.

[0042] S143, perform sub-region matching operation for digital image correlation (DIC). Read the acquired grayscale image sequence, define the first frame before loading as the reference image, and perform the sub-region matching operation during the loading process. The frame is a deformed image. A grid is divided within the computational region of the reference image, with the center defined as... The dimensions are A reference sub-region for each pixel is defined. A first-order shape function is introduced to describe the rigid body displacement, tension, and shear deformation of the sub-region. The grayscale distribution of the sub-pixel position is reconstructed using a bicubic B-spline interpolation algorithm. A correlation function is constructed using the Zero-mean Normalized Sum of Differences (ZNSSD) criterion. This function quantifies the grayscale similarity between the reference sub-region and the deformed sub-region by calculating the sum of squares of the differences between their grayscale average values ​​after normalization.

[0043] The shape function parameters are iteratively corrected using the inverse combination Gauss-Newton method (IC-GN) until the correlation function reaches its minimum value, thereby determining the precise displacement vector of the sub-region center point. and .

[0044] See attached document Figure 4 S144, calculate the global strain tensor and localization bandwidth. For the discrete displacement field data obtained in S143, a local least squares fitting smoothing algorithm with a window size of 5×5 to 9×9 nodes is used to suppress high-frequency noise. Central difference operations are performed on the smoothed displacement field, and the Green-Lagrange strain tensor is calculated based on finite deformation theory. Among these, the vertical normal strain along the loading axis is... The calculation is as follows: ; In the formula, Characterizes linear deformation terms; Characterizes nonlinear higher-order correction terms.

[0045] Extracting the maximum shear strain field The data matrix was processed using a binarization image processing algorithm to extract connected components with strain values ​​exceeding three times the overall average, which were identified as shear bands. Simultaneously, strain time-history data from each random measurement point were output to aid in the analysis of local deformation characteristics.

[0046] S145, Spatial mapping and damage projection of heterogeneous data. Based on a unified timestamp, retrieval and... (The sentence is incomplete and requires further context to be fully translated.) Frame Image Physical Time The matching set of acoustic emission localization events, with the matching time window set to The coordinates of the three-dimensional acoustic emission sphere within this time window. Orthogonally project onto the two-dimensional DIC observation plane on the front surface of the sample to obtain the projection point. Set a depth threshold. (e.g., 5mm), only retain Axis coordinates satisfy Shallow micro-fracture events were identified. The retained projection points were superimposed on the DIC maximum principal strain contour map, and the centroid coordinates of the projection point group and the normal distance to the skeleton line of the high-strain localization zone were calculated to characterize the spatiotemporal correlation of the evolution of microcrack damage inside the rock to the macroscopic fracture zone on the surface.

[0047] See attached document Figure 2 and reference appendix Figure 4 In step S150, the specific implementation method for multi-parameter coupling analysis and fracture prediction includes the following sub-steps: S151 quantifies the spatiotemporal nonuniformity of the surface deformation field. Based on the full-field strain data obtained in S144, correlation coefficients are first removed. For noise nodes greater than 0.02, only valid computational nodes are retained. The coefficient of variation of the principal strain field for each frame is calculated. As a dimensionless index characterizing the degree of deformation non-homogeneity: ; In the formula, The total number of valid nodes; For the first Frame number Equivalent principal strain at each node; This represents the average strain across the entire field for that frame.

[0048] Continuous system monitoring The evolutionary trend will The first derivative with respect to time is greater than a set threshold. The moment is determined as the nucleation start point of the strain localization band (Shear Band).

[0049] S152, establish the spatiotemporal coupling correlation between internal damage and surface deformation. Since acoustic emission data is based on event-triggered high-frequency discrete signals (with time precision in the microsecond range), while DIC data is based on frame rate-based low-frequency continuous signals (with time precision in the millisecond range), direct point-to-point statistical analysis is not possible. Therefore, the following resampling and sequence construction operations are required: Define a unified time grid: use the reciprocal of the DIC image acquisition frame rate as the unified time step (e.g., if the camera frame rate is 5fps), and construct a discrete time series.

[0050] Integration and downsampling of acoustic emission data: For acoustic emission signals, a time-window energy accumulation method is used for resampling. That is, within each time step interval, all acoustic emission impact events falling within that interval are retrieved, and the energy values ​​of these events are summed to obtain the acoustic emission energy value for that time step. In this way, a massive number of discrete acoustic emission events are transformed into an equal-length energy sequence that corresponds one-to-one with the number of image frames.

[0051] Interpolation and resampling of surface strain data: For the maximum principal strain rate data obtained by DIC calculation, if the original image acquisition time is within the same time grid... There are slight deviations, so linear interpolation is used for calculation. Precise strain rate value at time This ensures that the data points are strictly aligned on the timeline.

[0052] Cross-correlation analysis: Based on the above steps, equal-length sequences are constructed. and Calculate the cross-correlation function for the normalized sequences. : ; In the formula, Characterizes the amount of time lag; Characterization based on a unified time step The division of the first Discrete physical moments; Representation at time (i.e., time shift was taken into account) The maximum principal strain rate (after) The arithmetic mean of the maximum principal strain rate sequence over the entire analysis period; Characterizing the acoustic emission energy rate sequence over the entire analysis period The arithmetic mean.

[0053] Calculation makes The largest absolute value of time lag .like and The sample was determined to be in a strong damage-deformation coupling stage, meaning that the propagation of internal microcracks directly drove the macroscopic deformation of the surface.

[0054] S153, performs instability prediction using multi-precursor fusion. Configures comprehensive judgment logic; when the system simultaneously detects the following state reversal signals: Coefficient of variation in S152 Upon entering the exponential growth phase, the fracture time prediction module is triggered. Based on the Voight material failure differential equation, a linearized solution is obtained using the inverse velocity method. Observed physical quantities are used. To calculate the total shear strain across the entire field, the model parameters are set. Then the failure evolution equation degenerates into a linear form: ; In the formula, Characterizing the macroscopic cumulative deformation of rocks; Characterizes the instantaneous deformation rate or strain rate; The reciprocal of the inverse strain rate or deformation rate; Characterizes the tangent modulus parameter or attenuation coefficient; Characterizes the predicted macroscopic rupture moment.

[0055] The system has recently Inverse strain rate at each sampling point With time Perform linear regression analysis and calculate the intercept of the regression line with the time axis (i.e., time The intercept (value) is the predicted macroscopic rupture time. .

[0056] S154 generates an early warning signal and control feedback. It calculates the remaining time in real time. ,when When the time exceeds the preset equipment protection threshold (e.g., 500ms), the host computer sends a high-priority hardware interrupt signal to the testing machine controller. If the testing machine is in displacement control mode, the controller immediately freezes the servo valve opening to maintain actuator position hold; if it is in force control mode, the controller switches to the preset rapid unloading logic, reducing the oil pressure at a rate of 10%FS / s to prevent rebound impact from sample bursting from damaging the loading platen and sensor components. The specific circuit implementation of this control logic is well-known in the art and will not be described in detail here.

[0057] This invention also provides a computer device, which serves as the core control and data processing unit of the collaborative monitoring system. A typical implementation of this device can be a dedicated computing platform integrated into an industrial computer, server, or high-performance embedded system. Device structure: The computer device includes: Memory: Typically non-volatile memory, such as a solid-state drive (SSD), is used to store the computer program executed by the processor, as well as the raw data collected by the collaborative monitoring system and various characteristic parameter data generated. Specifically, the computer program stored in the memory contains instructions that, when executed by the processor, can control the collaborative monitoring system to complete all or part of the process from steps S110 to S150. In addition, the memory is also used to store preset model parameters, criterion thresholds, system configuration parameters, and intermediate and final result data processed in real time or post-processing during the experiment, such as acoustic emission waveform libraries, image sequences, ringing count time histories, source coordinate sets, full-field strain field matrices, rupture precursor type labels, predicted failure time, and rupture region coordinates.

[0058] Processor: Typically a Central Processing Unit (CPU), Graphics Processing Unit (GPU), Digital Signal Processor (DSP), or a combination thereof. The processor communicates with memory and is configured to execute computer programs stored in memory, thereby controlling the operation of the collaborative monitoring system and processing data. Its specific execution flow includes: In the initial stage, control commands are generated and sent, the operating parameters of the loading device, acoustic emission acquisition instrument, industrial camera and GPS timing module are configured through the communication module, and their readiness status feedback is received.

[0059] During the loading and acquisition phase, the system receives a synchronization trigger signal (TTL pulse rising edge) sent by the GPS timing module and sends a synchronization start command to each subsystem based on this signal, or confirms that each subsystem has started based on this signal. At the same time, it receives, verifies, and stores data streams from each subsystem in real time.

[0060] During the data processing phase, the corresponding algorithm modules are invoked to execute the calculation tasks in steps S140 and S150. For example, the source location program, which includes damped least squares iterative solution, is run; the digital image correlation algorithm is executed to calculate the full-field strain and strain field variation coefficient; the cross-correlation analysis of acoustic emission energy rate and strain rate sequence is performed; and the inverse velocity regression analysis and spatial projection superposition analysis of the Voight model are run.

[0061] During the early warning and feedback phase, based on the judgment result of the multi-parameter coupling criterion in step S150, if a critical instability precursor is identified, the control processor calls the Voight model to calculate and predict the rupture time. It generates early warning information including the predicted rupture time and spatial coordinates of the rupture area. Simultaneously, it can send control commands to the control unit of the loading device via the communication module to trigger preset protection logic (such as switching control modes, locking the position, or unloading).

[0062] Communication module: Includes various wired and / or wireless interfaces for data exchange and control command transmission between computer equipment and various hardware sub-devices of the collaborative monitoring system. Typically includes: Network interface: Typically a gigabit Ethernet interface, used to connect industrial camera arrays and receive high-resolution image data streams.

[0063] Data acquisition card interface or dedicated bus interface: used to connect to the acoustic emission acquisition instrument to receive high-speed acoustic emission waveform data and characteristic parameters.

[0064] Analog / digital input / output interface: used to connect to the controller of the loading device, receive sensor signals such as load and displacement, and send commands such as control mode switching and loading rate setting.

[0065] GPS timing interface: Usually an RS232 serial interface or a USB interface, used to receive time message information from the GPS timing module for the GpsTimer software to perform system time calibration.

[0066] Synchronous trigger interface: Used to connect to a synchronous trigger, receive or monitor the TTL pulse signal emitted by it, and confirm the system startup status.

[0067] Human-computer interaction interface: such as USB, HDMI, etc., used to connect monitors, keyboards, and mice to enable parameter setting, process monitoring and result display.

Claims

1. A rock fracture prediction method based on the synergistic monitoring of acoustic emission and surface strain, characterized in that, Includes the following steps: S110, Select a cuboid rock block as a rock sample, construct a random speckle field suitable for digital image correlation analysis on the surface of the rock sample, and construct a collaborative monitoring system including a loading device, an acoustic emission monitoring system and an industrial camera array. S120, Based on the GPS timing module, a unified timestamp is established for the loading device, the acoustic emission monitoring array and the industrial camera array, and the signal trigger threshold parameters of the loading device and the sampling frequency of the acoustic emission monitoring system and the image acquisition frame rate parameters of the industrial camera array are set. S130, a continuous load is applied to the rock sample by the loading device, and the collaborative monitoring system is triggered by the synchronous trigger and the load trigger threshold parameter. The acoustic emission monitoring system collects acoustic emission signals according to the sampling frequency, and the industrial camera array acquires sample surface image sequences according to the image acquisition frame rate parameter. S140, collect and calculate the ring count and source spatial coordinates of the acoustic emission signal, and use digital image correlation algorithm to track the pixel sub-region position changes of the random speckle field in the sample surface image sequence, and calculate the random measurement points and full-field strain data of the sample surface. S150, Correlation analysis is performed on the evolution characteristics of the ringing count, the spatial coordinates of the seismic source, and the strain data. Based on the multi-parameter coupling criterion, the rock fracture precursor type is identified and the fracture area is predicted.

2. The rock fracture prediction method based on the coordinated monitoring of acoustic emission and surface strain according to claim 1, characterized in that, In step S110, the acoustic emission monitoring array includes no less than eight acoustic emission sensors. The acoustic emission sensors are distributed in the upper, middle and lower regions of the side of the rock sample according to the principle of spatial non-coplanar and non-collinearity, and the positions of the acoustic emission sensors on adjacent sides are staggered. The industrial camera array comprises four high-resolution industrial cameras arranged at a 90-degree angle, with the optical axes of the high-resolution industrial cameras orthogonal to the center of the corresponding rock sample surface.

3. The rock fracture prediction method based on the coordinated monitoring of acoustic emission and surface strain according to claim 1, characterized in that, In step S110, the dimensions of the rock sample satisfy the geometric conditions that the ratio of height to length is greater than or equal to 2.0 and the ratio of height to width is greater than or equal to 2.

0. The random speckle field consists of a matte white background layer covering the surface of the rock sample and matte black spots sprayed onto the white background layer, with the matte black spots having a coverage rate of 40% to 60%.

4. The rock fracture prediction method based on the coordinated monitoring of acoustic emission and surface strain according to claim 3, characterized in that, In step S120, the specific method for establishing a unified timestamp is as follows: The GPS timing module is connected to the control computer of the loading device, the acoustic emission monitoring system, and the industrial camera array via a 232 interface or a USB interface. GpsTimer time synchronization software is installed on each control computer. Before the experiment begins, GpsTimer is turned on to synchronize the time and establish a unified timestamp for the loading device, the acoustic emission monitoring system, and the industrial camera array.

5. The rock fracture prediction method based on the coordinated monitoring of acoustic emission and surface strain according to claim 4, characterized in that, In step S130, the continuous load is applied in the following ways: After loading to a preset contact value using the loading device in force control mode, the system switches to displacement control mode. The specific logic for triggering the collaborative monitoring system is as follows: the synchronous trigger is connected to the loading device, the acoustic emission monitoring system, and the industrial camera array. When the axial load detected by the loading device reaches the preset trigger threshold for the first time, the synchronous trigger generates a pulse signal to trigger the acoustic emission monitoring system to start collecting acoustic emission signals at the sampling frequency, and the industrial camera array starts collecting surface images at the image acquisition frame rate.

6. The rock fracture prediction method based on the synergistic monitoring of acoustic emission and surface strain according to claim 1, characterized in that, The method for calculating the spatial coordinates of the seismic source is as follows: a three-dimensional positioning equation system is constructed based on the time difference of arrival data of the acoustic emission signal, and the three-dimensional spatial position of the acoustic emission source is obtained by iteratively solving the damped least squares method. The calculation method for the random measurement points and full-field strain data is as follows: by tracking the displacement vector of each pixel sub-region in the random speckle field before and after deformation, the maximum shear strain field data is generated by using displacement field differential operation.

7. The rock fracture prediction method based on the synergistic monitoring of acoustic emission and surface strain according to claim 1, characterized in that, The specific content of the multi-parameter coupling criterion includes: Calculate the coefficient of variation of the principal strain field of the full-field strain data, and monitor the rate of change of the coefficient of variation of the principal strain field with time. Cross-correlation analysis was performed on the acoustic emission energy rate sequence and the maximum principal strain rate sequence of the surface after resampling based on a unified time step, and the cross-correlation coefficient and time lag were calculated. When the rate of change of the coefficient of variation of the principal strain field exceeds a set threshold, the time lag is less than or equal to the uniform time step, and the cross-correlation coefficient is greater than 0.85 and indicates strong coupling characteristics, it is determined that a rupture precursor has occurred.

8. The rock fracture prediction method based on the synergistic monitoring of acoustic emission and surface strain according to claim 1, characterized in that, The types of rock fracture precursors identified include: When a significant increase in the ringing count is detected while there is no significant abrupt change in the overall strain data, it is determined to be a precursor to internal crack activation, and the area of ​​concentrated internal damage is determined based on the spatial coordinates of the seismic source. When the monitoring shows localized concentration of the full-field strain data and the acoustic emission signal response is lagging, it is determined to be a precursor to surface rupture, and the concentrated area of ​​surface deformation is determined based on the full-field strain data. When both the ringing count and the full-field strain data show abnormalities, it is determined to be a precursor to critical instability.

9. The rock fracture prediction method based on the coordinated monitoring of acoustic emission and surface strain according to claim 1, characterized in that: In step S150, the method for predicting the fracture area is as follows: The inverse velocity method of the Voight model is used, with the total shear strain in the whole field as the observed physical quantity, and regression analysis is performed on the linear relationship between the inverse strain rate and time to calculate the theoretical failure time. The projection point of the spatial coordinates of the seismic source corresponding to the theoretical failure time is spatially superimposed with the high-strain region in the full-field strain data, and the overlapping region is the predicted macroscopic rupture location.

10. A computer device, characterized in that, It includes a memory, a processor, and a communication module, wherein the processor, when executed, implements the rock fracture prediction method based on the coordinated monitoring of acoustic emission and surface strain as described in any one of claims 1 to 9.

Images