A method and system for testing the compressive strength of a bentonite mud

Abstract

The application discloses a bentonite mud compressive strength test method and system, and belongs to the technical field of bentonite mud performance test. The method comprises the following steps: obtaining a bentonite mud sample and placing the sample in a test cavity with a dynamic loading function; applying dynamic disturbance excitation to the bentonite mud sample, collecting response waveform data of the bentonite mud sample under the dynamic disturbance excitation; constructing a structure response matrix of the bentonite mud sample according to the response waveform data; iteratively calculating a nonlinear compressive factor of the bentonite mud sample based on the structure response matrix and a plurality of preset strength benchmark thresholds; and mapping the nonlinear compressive factor to a preset compressive strength grading system to generate a compressive strength test result. The system comprises a memory, a processor and a computer program, and the processor realizes the above method steps when executing the program.

Description

Technical Field

[0001] This invention relates to the field of bentonite slurry performance testing technology, specifically to a method and system for testing the compressive strength of bentonite slurry. Background Technology

[0002] Bentonite slurry is widely used in civil engineering, diaphragm wall construction, and tunnel boring machines. Its compressive strength directly affects the stability of the engineering structure and construction safety. Current methods for testing the compressive strength of bentonite slurry generally employ uniaxial or triaxial static loading failure tests. These tests gradually increase mechanical pressure until the sample structure collapses, recording the stress value at failure as a strength indicator. This type of method is destructive; the sample completely loses its original structure after the test, making repeated verification or continuous monitoring impossible. Furthermore, the static loading method only reflects the ultimate strength of the slurry under slowly increasing pressure conditions, neglecting the impact of dynamic disturbances such as vibration and impact on the gradual weakening of its internal structure in actual engineering environments. This leads to discrepancies between the test results and the actual bearing capacity under real-world conditions.

[0003] The existing technical solutions have the following main drawbacks. First, the destructive testing mode makes it impossible to evaluate the same batch of samples multiple times, and it is difficult to confirm the accuracy of the data through retesting when the sample representativeness is insufficient. Second, the single strength index based on static mechanical parameters lacks the characterization of the structural response characteristics of mud under dynamic excitation, and cannot reveal the nonlinear behavior such as stiffness degradation and phase shift of the internal particle skeleton of mud during disturbance, so that the strength evaluation remains at the level of apparent failure point rather than the evolution of structural state. Third, conventional data processing methods usually directly use peak stress or strain, without establishing a robust mapping from multi-dimensional dynamic response to a unified strength index, making it difficult to integrate the differentiated response information of samples at different depths and frequencies, resulting in large dispersion of test results.

[0004] For the problem of characterizing structural state under dynamic disturbance conditions, it is necessary to solve how to obtain a feature matrix that can sensitively reflect changes in the internal structure of mud through specific excitation methods and response analysis techniques, so as to extract mechanical labels that are intrinsically related to compressive strength under non-destructive premises. For the problem of generating strength indices, it is necessary to solve how to use multiple strength benchmark thresholds for hierarchical comparison and iterative parameter updates, converge complex matrix information into stable nonlinear factors, and automatically incorporate them into a grading system to output readable strength levels, avoiding the uncertainty of empirical interpretation. Summary of the Invention

[0005] A test method and system are provided to obtain the grading results of the compressive strength of bentonite mud under dynamic disturbance conditions without destroying the sample. The non-destructive and accurate evaluation of the compressive strength of the mud can be achieved through matrix analysis of response waveform data and calculation of nonlinear factors based on multi-threshold iteration.

[0006] The objective of this invention can be achieved through the following technical solutions: This invention provides a method and system for testing the compressive strength of bentonite slurry, which achieves high-precision evaluation of the compressive strength of bentonite slurry by combining dynamic disturbance excitation with structural response analysis.

[0007] In one technical solution of the present invention, the method includes: acquiring a bentonite slurry sample and placing it in a test chamber with dynamic loading function; applying dynamic perturbation excitation to the bentonite slurry sample and acquiring response waveform data of the bentonite slurry sample under dynamic perturbation excitation; constructing a structural response matrix of the bentonite slurry sample based on the response waveform data; iteratively calculating the nonlinear compressive strength factor of the bentonite slurry sample based on the structural response matrix and multiple preset strength benchmark thresholds; mapping the nonlinear compressive strength factor to a preset compressive strength grading system to generate the compressive strength test result of the bentonite slurry sample. This method avoids the destructive impact of traditional static testing on the sample structure and can more realistically reflect the dynamic bearing capacity of the slurry in actual engineering environments.

[0008] As a preferred technical solution of the present invention, when obtaining bentonite slurry samples, multiple sub-samples are extracted from multiple different depth locations of the bentonite slurry to be tested. Each sub-sample undergoes bubble removal and isothermal curing treatment. The treated sub-samples are then sequentially placed into the sample container of the test chamber, and the placement posture and sample volume information of each sub-sample are recorded. Through multi-point sampling and standardized pretreatment, the interference of internal bubbles and temperature differences on test accuracy is effectively eliminated, ensuring the consistency of sample state and the comparability of test results. Preferably, the bubble removal treatment uses a vacuum degassing device to continuously treat each sub-sample for 5 to 10 minutes under negative pressure to fully remove microbubbles from the sample.

[0009] As a technical solution of the present invention, when applying dynamic perturbation excitation to bentonite slurry samples and collecting response waveform data, a sinusoidal perturbation sequence with linearly increasing frequency is sequentially applied to each sub-sample in the test chamber. During the application process, multiple piezoelectric sensors attached to the outer wall of the sample container synchronously collect the vibration response signal of each sub-sample. The vibration response signal is segmented in the time domain to obtain response waveform segments corresponding to multiple time windows for each sub-sample. Then, all response waveform segments are arranged in frequency order to form response waveform data. By using frequency sweep excitation and synchronous acquisition by multiple sensors, the dynamic response characteristics of the sample in different frequency bands can be comprehensively captured. Preferably, the frequency range of the sinusoidal perturbation sequence is from 0.1Hz to 1000Hz, and the frequency interval between adjacent frequency points is distributed on a logarithmic scale. This can obtain higher frequency resolution in the low-frequency band and cover a wider frequency band in the high-frequency band, thereby improving the information richness of the response waveform data.

[0010] In the step of constructing the structural response matrix of this invention, the peak amplitude and phase delay of each subsample in the response waveform segment of each time window are extracted. Based on this, the dynamic stiffness coefficient of each subsample at each frequency point is calculated. The dynamic stiffness coefficients of all subsamples at all frequency points are arranged according to the subsample index and frequency index to form an initial stiffness matrix. Then, normalization and singular value decomposition are performed on the initial stiffness matrix to obtain the reduced-order structural response matrix. This processing method can extract the core characteristic parameters representing the internal structure of mud from complex response waveform data and eliminate redundancy and dimensional differences between data, providing a more stable and concise numerical basis for subsequent factor calculations. Preferably, the maximum value of the dynamic stiffness coefficients of all subsamples at all frequency points is used as the normalization benchmark during normalization.

[0011] In the iterative calculation of the nonlinear compressive strength factor, an initial weight coefficient is assigned to each element in the structural response matrix. The structural response matrix is ​​then compared with a first strength benchmark threshold. Based on the result, the weight coefficients are adjusted to obtain the first adjusted weight. The weighted structural response matrix is ​​then compared with a second strength benchmark threshold. Based on the result, intermediate regression parameters are generated. These intermediate regression parameters are substituted into the recursive update formula for the nonlinear compressive strength factor. The generation and recursive update steps are repeated until the rate of change of the intermediate regression parameters is less than a preset rate of change threshold, at which point the iteration stops and the current nonlinear compressive strength factor is output. Through dual threshold comparison and iterative regression, the factor calculation process can gradually converge to a stable value under the constraints of multiple strength benchmarks, improving the factor's ability to characterize the nonlinear mechanical behavior of mud. Preferably, a momentum factor is introduced into the recursive update formula to suppress numerical oscillations of the nonlinear compressive strength factor during the iteration process and accelerate the convergence process.

[0012] When generating compressive strength test results, a compressive strength grading mapping table containing multiple continuous intervals is pre-constructed. Each continuous interval corresponds to a strength level identifier. The target continuous interval into which the nonlinear compressive factor falls is queried, and the corresponding strength level identifier is extracted as the preliminary strength level of the current sub-sample. Then, the preliminary strength levels of all sub-samples are merged by majority voting, and the strength level with the most votes is taken as the compressive strength test result of the bentonite mud sample. This grading mapping and voting fusion mechanism can integrate the test information of multiple sub-samples, effectively reduce the impact of abnormal fluctuations in a single sample on the final result, and improve the robustness and accuracy of the test results.

[0013] As a further improvement of this invention, after generating the compressive strength test result, the result is compared with historical test results to calculate the test deviation value. When the test deviation value exceeds a preset deviation alarm threshold, a recalibration process is triggered. In the recalibration process, multiple preset strength benchmark thresholds are reacquired and the nonlinear compressive strength factor is recalculated iteratively. By introducing a self-calibration and deviation alarm mechanism, timely corrections can be made when the test system state drifts or the sample characteristics change significantly, ensuring the stability and reliability of long-term continuous testing.

[0014] In another preferred embodiment, when applying dynamic perturbation excitation to the bentonite slurry sample, at least three different waveform types are selected from a preset perturbation waveform library, including sine waves, triangular waves, and square waves. The selected perturbation waveforms are grouped according to waveform type to generate multiple perturbation excitation sequences, which are then applied to the bentonite slurry sample in a fixed order: first sine waves, then triangular waves, and finally square waves. A static time window without excitation is inserted each time the perturbation excitation sequence is switched. Multi-waveform excitation can stimulate the structural response of the slurry sample from different angles, and the insertion of the static time window effectively avoids crosstalk of residual vibrations caused by previous excitations to subsequent tests. Accordingly, during the acquisition of response waveform data, corresponding response waveform sub-data is independently acquired during the application of each set of perturbation excitation sequences. Each set of response waveform sub-data is then subjected to Fourier transform to obtain frequency domain response sub-data. The dominant frequency energy value and dominant frequency offset are extracted from each set of frequency domain response sub-data. The dominant frequency energy value and dominant frequency offset are then concatenated according to the waveform type to form a multi-dimensional response feature vector as the response waveform data. This method compresses the response information under multi-waveform excitation into a compact feature vector, facilitating the subsequent construction of the structural response matrix. Further, when constructing the structural response matrix, each dimension element of the multi-dimensional response feature vector is used as the initial column vector of the structural response matrix. The cross-correlation value between each initial column vector and the remaining initial column vectors is calculated. The initial column vectors are then reordered according to the magnitude of the cross-correlation value, with the column vectors having the largest cross-correlation value arranged as adjacent columns to obtain a reordered matrix. Principal component analysis is then performed on the reordered matrix to extract principal component components whose cumulative contribution rate reaches a preset contribution rate threshold. These principal component components constitute the dimensionality-reduced structural response matrix. This construction method enhances the structured feature representation within the matrix through relevant sorting and principal component extraction, further compressing the data dimensionality and highlighting key information closely related to compressive strength.

[0015] This invention also provides a bentonite slurry compressive strength testing system, including a memory, a processor, and a computer program stored in the memory and running on the processor. When executing the computer program, the processor implements the steps of the aforementioned bentonite slurry compressive strength testing method. This system integrates the above method into an automated testing platform, enabling fully automated processing from sample excitation, data acquisition, feature analysis to strength grade determination. It provides an efficient and repeatable technical means for large-scale slurry strength testing in engineering field and laboratory environments.

[0016] The beneficial effects of this invention are: By acquiring bentonite slurry samples and placing them in a test chamber with dynamic loading capabilities, dynamic perturbation excitation is applied to the samples, and response waveform data is collected to construct a structural response matrix. This approach replaces the traditional static pressure-to-failure method with non-destructive dynamic excitation, allowing the same sample to be tested repeatedly and eliminating the limitation of sample non-reusability caused by destructive testing. Parameters such as peak amplitude and phase delay extracted from the response waveform can capture the dynamic stiffness changes of the slurry under perturbation at different frequencies, reflecting the real-time evolution of the interparticle structural connections. The obtained structural response matrix contains multi-dimensional mechanical characteristic information, providing a more complete description of the structural state of the slurry under perturbation than a single static failure stress value.

[0017] Based on the structural response matrix and multiple preset strength benchmark thresholds, a nonlinear compressive strength factor is calculated iteratively. During the iteration process, the structural response matrix is ​​hierarchically compared with the first and second strength benchmark thresholds, the element weight coefficients are dynamically adjusted, and intermediate regression parameters are generated. Then, a recursive update formula incorporating a momentum factor is used to suppress iterative oscillations, ensuring the stable convergence of the nonlinear compressive strength factor. This process compresses complex matrix data into a compressive strength factor with a clear physical orientation, overcoming the problem of conventional methods relying solely on peak stress while ignoring the evolution of the structural nonlinear response. The introduction of the momentum factor effectively suppresses numerical jumps in the factor during iteration, ensuring consistency of results across multiple tests.

[0018] The nonlinear compressive strength factor is mapped to a pre-defined compressive strength grading system to generate compressive strength test results. A pre-constructed compressive strength grading mapping table containing multiple continuous intervals is used, allowing direct lookup of the corresponding strength grade identifier, eliminating the arbitrariness of manually set thresholds. For multiple sub-samples extracted from different depth locations, independent testing is performed, followed by majority voting to determine the final strength grade. This ensures that the overall result comprehensively reflects the strength performance of the mud at different layers, suppressing the interference of individual abnormal sub-samples on the overall judgment and improving the robustness and representativeness of the test results. Attached Figure Description

[0019] The invention will now be further described with reference to the accompanying drawings.

[0020] Figure 1 This is a flowchart of the method for testing the compressive strength of bentonite slurry; Figure 2 This is a flowchart of bentonite mud sample processing and parameter recording. Figure 3 This is a flowchart of the dynamic stiffness coefficient calculation and structural response matrix order reduction process; Figure 4 This is a schematic diagram of the bentonite slurry compressive strength testing system. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] See Figure 1 This invention provides a method for testing the compressive strength of bentonite slurry, comprising: acquiring a bentonite slurry sample and placing it in a test chamber with dynamic loading function; applying dynamic perturbation excitation to the bentonite slurry sample and acquiring response waveform data of the bentonite slurry sample under dynamic perturbation excitation; constructing a structural response matrix of the bentonite slurry sample based on the response waveform data; iteratively calculating the nonlinear compressive strength factor of the bentonite slurry sample based on the structural response matrix and multiple preset strength benchmark thresholds; mapping the nonlinear compressive strength factor to a preset compressive strength grading system to generate the compressive strength test result of the bentonite slurry sample.

[0023] In specific implementation, please refer to Figure 2 Multiple subsamples were extracted from different depths of the bentonite mud to be tested. Using a multi-point sampler with a depth scale, one subsample was extracted from the mud pit or mud tank at depths of 0.3 m, 1.0 m, and 2.0 m from the upper surface, respectively. Each subsample had a volume of 200 ml. The three subsamples were numbered separately and processed independently.

[0024] When performing bubble removal treatment on each subsample, the subsample is transferred to the sealed chamber of the vacuum degassing device, which is connected to a vacuum pump. The vacuum pump is started, reducing the absolute pressure inside the sealed chamber to a negative pressure range of 5 kPa to 10 kPa, and this negative pressure condition is maintained for 8 minutes. Under negative pressure, the bubbles inside the subsample expand, rise, and burst, thus achieving bubble removal. After treatment, the pressure is slowly restored to normal, and the subsample is removed.

[0025] The isothermal curing process was carried out in an isothermal chamber. The subsamples, after bubble removal treatment, were placed inside the chamber, which was set to 25 degrees Celsius for 2 hours. This isothermal treatment stabilized the internal structure of the subsamples, reducing the impact of temperature differences on subsequent dynamic perturbation responses.

[0026] Each subsample, after being filtered to remove air bubbles and cured at a constant temperature, was sequentially placed into the sample container of the test chamber. The sample container was a cylindrical rigid container with an inner diameter of 60 mm and a height of 120 mm, and a vibration platform was fixed at the bottom. After each subsample was injected into the sample container, the upper surface was leveled with a scraper to ensure a uniform filling height. During the injection process, the sample container was kept vertical to avoid introducing new air bubbles.

[0027] When recording the placement posture information of each subsample in the sample container, a six-axis attitude sensor installed at the center of the bottom of the sample container acquires the tilt angle and azimuth angle of the sample container in real time. The attitude sensor outputs three-axis acceleration values. , , And the three-axis gyroscope values. The pitch angle of the sample container is calculated using the three-axis acceleration values. and roll angle The specific calculation formula is as follows:

[0028] in, This represents the acceleration component along the x-axis of the attitude sensor, expressed in meters per second squared. This represents the acceleration component along the y-axis of the attitude sensor, expressed in meters per second squared. This represents the acceleration component along the z-axis of the attitude sensor, expressed in meters per second squared. This represents the pitch angle of the sample container relative to the horizontal plane, in radians, and its value ranges from [value missing]. Roll angle The calculation method uses a similar formula, which will not be elaborated here. Pitch and roll angles are recorded as placement attitude information.

[0029] When recording sample volume information, a laser rangefinder is used to measure the distance from the top surface of the sub-sample inside the sample container to the top of the container, combined with the known radius of the inner cylinder of the sample container. ,in The value is 30 mm, and the total height of the sample container's inner cavity is... ,in Calculate the filling volume of the subsample, given a thickness of 120 mm. The calculation method is as follows ,in The distance value is measured by the laser rangefinder, in millimeters. The calculated filling volume is recorded as the sample volume information. For each subsample, the above process of recording the placement posture information and sample volume information is performed independently, and the recorded data is stored in association with the subsample number.

[0030] In specific implementation, please refer to Figure 3A sinusoidal perturbation sequence with linearly increasing frequency is sequentially applied to each subsample in the test chamber. The sinusoidal perturbation sequence is generated by a signal generator and driven by a power amplifier to power an electromagnetic exciter located at the bottom of the sample container. The frequency range of the sinusoidal perturbation sequence is set from 0.1Hz to 1000Hz, with the frequency intervals between adjacent frequency points distributed on a logarithmic scale. The total number of frequency points is set to 50. Frequency value at each frequency point Determined based on the logarithmic scale distribution relationship, where For frequency index, Take values ​​from 1 to 50. It is 0.1Hz. For 1000Hz, Frequency points greater than 1, frequency values Depend on Calculation. The exciter continuously excites 10 complete sine wave cycles at each frequency point to ensure that the response enters a steady state. There is no time interval between switching between adjacent frequency points, forming a continuous sine wave perturbation sequence.

[0031] During the application of a sinusoidal perturbation sequence, vibration response signals of each subsample are synchronously acquired using multiple piezoelectric sensors attached to the outer wall of the sample container. Four piezoelectric sensors are evenly distributed along the circumference of the sample container, attached 20 mm from the bottom. The output signals of the piezoelectric sensors are conditioned by a charge amplifier and then synchronously sampled by a data acquisition card at a sampling frequency of 10000 Hz. The discrete signal acquired by each sensor channel is the vibration response signal.

[0032] The vibration response signal of each subsample is segmented in the time domain to obtain response waveform segments corresponding to multiple time windows for each subsample. Time-domain segmentation uses frequency points as index units. Within the duration of the sinusoidal perturbation sequence at each frequency point, the vibration response signal of the last three complete sinusoidal cycles is extracted as a response waveform segment for one time window. Therefore, each subsample obtains response waveform segments from four sensor channels at each frequency point, and the length of each response waveform segment corresponds to three complete cycles. The response waveform segments of each subsample at all 50 frequency points and all four sensor channels are arranged in ascending order of frequency index, i.e., the frequency index... The response waveform segment with a value of 1 is listed first, frequency index. The response waveform segment with a value of 50 is placed last. The response waveform segments of the four sensor channels under the same frequency index are arranged in order according to the sensor number. The arranged data set constitutes the response waveform data.

[0033] Digital phase-locked amplification (PLA) is used to extract the peak amplitude and phase delay of each subsample in the response waveform segment at each time window. The response waveform segment is demodulated with a zero-phase sine and cosine wave of a reference signal at the same frequency as the excitation frequency. Low-pass filtering is then used to obtain the in-phase and quadrature components, which are then used to calculate the peak amplitude. and phase delay ,in For subsample indexing, The values ​​are 1, 2, and 3, which correspond to subsamples extracted from three different depth locations. For frequency index, The value ranges from 1 to 50; For sensor channel index, Values ​​range from 1 to 4. Peak amplitude The phase delay is the amplitude value of the frequency component in the waveform segment that is the same as the excitation frequency. This is the phase lag value of the frequency component in the response waveform segment relative to the excitation reference signal.

[0034] Based on the peak amplitude and phase delay, the dynamic stiffness coefficient of each subsample at each frequency point is calculated. The dynamic stiffness coefficient represents the specimen's ability to resist deformation under dynamic disturbances, and its calculation is performed independently for each subsample, each frequency point, and each sensor channel. For the subsample index... Frequency Index and sensor channel index Dynamic stiffness coefficient Calculate using the following formula:

[0035] in, The peak amplitude is expressed in volts, representing the peak value of the first volt. The subsamples at the th At the frequency point, the first The amplitude value of the response waveform segment acquired by each sensor channel; The phase delay is expressed in radians, representing the phase lag angle of the component corresponding to the excitation frequency in the response waveform segment relative to the excitation reference signal. The displacement reference value is expressed in micrometers, representing the displacement amplitude output by the electromagnetic exciter at a frequency of 1 Hz under no-load calibration conditions in the test chamber. The value is set to 10 micrometers, which is based on the midpoint of the linear range between the input voltage and output displacement of the exciter at a frequency of 1 Hz. This is the cosine function value of the phase delay, dimensionless. The calculated dynamic stiffness coefficient... The unit is volts per micrometer.

[0036] The dynamic stiffness coefficients of all subsamples at all frequency points are arranged according to the subsample index and frequency index to form an initial stiffness matrix. The rows of the initial stiffness matrix correspond to the combined arrangement of subsamples and frequency points, and the columns correspond to the sensor channels. Specifically, the arrangement is as follows: [The text abruptly ends here, likely due to an incomplete sentence or missing information.] 1. Frequency Index The dynamic stiffness coefficients of the 50 groups of 4 sensor channels corresponding to numbers 1 to 50 are used as rows 1 to 50 of a matrix; then the subsample indexes are... 2. Frequency Index The 50 sets of dynamic stiffness coefficients corresponding to 1 to 50 are used as rows 51 to 100 of the matrix; finally, the subsample indexes are... 3. Frequency Index The 50 sets of dynamic stiffness coefficients corresponding to 1 to 50 are used as rows 101 to 150 of the matrix. Therefore, the initial stiffness matrix has 150 rows and 4 columns.

[0037] When normalizing the initial stiffness matrix, the maximum value of the dynamic stiffness coefficients of all subsamples at all frequency points is used as the normalization benchmark. The elements of the initial stiffness matrix (150 rows, 4 columns) are traversed, and the element with the largest absolute value is found and denoted as . Divide each element in the initial stiffness matrix by This yields a normalized matrix, where the value of each element ranges from -1 to 1.

[0038] Singular value decomposition (SVD) is performed on the normalized matrix to obtain the reduced-order structural response matrix. SVD decomposes the normalized matrix into the product of a left singular vector matrix, a singular value diagonal matrix, and a right singular vector matrix. Let the normalized matrix be... Its dimension is 150 rows and 4 columns, obtained from singular value decomposition. ,in It is a 150-row, 150-column left singular vector matrix. It is a 150-row, 4-column singular value diagonal matrix. The matrix is ​​a 4x4 right singular vector matrix. The singular value diagonal matrix contains four singular values, arranged in descending order. Extract the left singular vector matrix corresponding to the three largest singular values. The first three columns form a reduced-order matrix with dimensions of 150 rows and 3 columns. This reduced-order matrix is ​​the structural response matrix after the order reduction. The 150 rows of the structural response matrix still correspond to the combination of subsamples and frequency points, and the 3 columns correspond to the response features along the retained principal singular value directions.

[0039] In practice, an initial weight coefficient is assigned to each element in the structural response matrix. The structural response matrix has 150 rows and 3 columns. An initial weight coefficient of 1 is assigned to each element of the structural response matrix one by one, forming an initial weight matrix with the same dimension as the structural response matrix. The number of rows and columns of the initial weight matrix are the same as those of the structural response matrix.

[0040] The structural response matrix is ​​compared with a first strength benchmark threshold. The first strength benchmark threshold is a preset scalar value, set to 0.65. This threshold is determined by analyzing the statistical distribution of structural response matrix elements from a large number of bentonite slurry samples in historical tests, using the 50th percentile of the cumulative distribution function as the boundary between low and high responses. The specific method of the first comparison operation is as follows: Each element in the structural response matrix is ​​iterated over, and its value is compared with the first strength benchmark threshold. For any element, if the value is greater than the first strength benchmark threshold, a comparison marker value of 1 is generated; if the value is less than or equal to the first strength benchmark threshold, a comparison marker value of -1 is generated. All the comparison marker values ​​constitute a comparison marker matrix with the same dimension as the structural response matrix.

[0041] The initial weight coefficients are adjusted based on the results of the first comparison operation to obtain the first adjusted weight. The adjustment method is as follows: the initial weight coefficient corresponding to each element in the initial weight matrix is ​​linearly superimposed with the comparison mark value in the corresponding comparison mark matrix and a scaling factor multiplied by the comparison mark value. The scaling factor is set to 0.05. The basis for this value is that, in parameter update stability testing, this value ensures that the single adjustment amplitude does not exceed 5% of the initial weight coefficient value, avoiding excessive fluctuations in the weight coefficient in the early stages of iteration. The adjustment calculation is that the first adjusted weight equals the initial weight coefficient plus the product of the scaling factor and the comparison mark value. After adjusting all elements, the first adjusted weight matrix is ​​obtained, and the dimension of the first adjusted weight matrix is ​​the same as that of the structural response matrix.

[0042] The structural response matrix, after being weighted by the first adjustment weight, is compared with the second strength benchmark threshold in a second comparison operation. The structural response matrix after the first adjustment weight is obtained by multiplying each element of the first adjustment weight matrix by the corresponding element in the structural response matrix. The second strength benchmark threshold is set to 0.80, which is greater than the first strength benchmark threshold. The second strength benchmark threshold of 0.80 is based on the statistical distribution analysis of the structural response matrix element values ​​of a large number of bentonite mud samples in historical tests, using the value corresponding to the 85th percentile of the cumulative distribution function as the boundary for judging high-intensity response. The specific method of the second comparison operation is as follows: traverse each element in the weighted structural response matrix, calculate the difference between the value of each element in the weighted structural response matrix and the second strength benchmark threshold, and arrange all the differences of elements into a difference matrix according to the element's position.

[0043] Intermediate regression parameters are generated based on the results of the second comparison operation. The intermediate regression parameters are generated as follows: the sum of the squares of all elements in the difference matrix is ​​divided by the total number of elements in the difference matrix to obtain the mean squared error value, which is then used as the intermediate regression parameter. The total number of elements in the difference matrix is ​​150 multiplied by 3, i.e., 450.

[0044] Substitute the intermediate regression parameters into the recursive update formula for the nonlinear stress resistance factor, and repeat the steps of generating and updating the intermediate regression parameters until the rate of change of the intermediate regression parameters is less than a preset rate of change threshold. Then stop iterating and output the current nonlinear stress resistance factor. The recursive update formula incorporates a momentum factor and has the following form:

[0045] in, Indicates the current iteration step number. The value can be an integer greater than or equal to 2; Indicates the first The nonlinear stress factor output by the iteration step is dimensionless. Indicates the first The nonlinear stress factor output by the iteration step is dimensionless. Indicates the first The nonlinear stress factor output by the iteration step is dimensionless. This represents the learning step size coefficient. Set to 0.01. The value of 0.01 is based on the fact that in multiple iterative convergence tests, this value can make the nonlinear compressive factor stabilize within 30 steps and prevent divergent oscillations. Indicates the first The intermediate regression parameters generated by the iteration step are dimensionless. Represents the momentum factor. Set to 0.8. The value of 0.8 is chosen because, in typical second-order optimization problems, this value effectively suppresses iterative oscillations while preserving sufficient convergence speed. Initial nonlinear stress factor. and All are set to 0. In each iteration step, the weighted structural response matrix is ​​recalculated using the nonlinear stress factor from the previous iteration step. Specifically, each element of the weighted structural response matrix is ​​replaced with the result of multiplying the nonlinear stress factor from the previous iteration step by the corresponding element of the structural response matrix, and then multiplying it by the corresponding element of the first adjustment weight matrix. Then, the second comparison operation and the generation of intermediate regression parameters are re-executed, and the newly generated intermediate regression parameters are... Substitute into the recursive update formula to calculate The rate of change of the intermediate regression parameters is calculated according to... Calculation, where Indicates the first The intermediate regression parameters generated in the iterative step have a preset change rate threshold of 0.001. When the change rate of the intermediate regression parameters is less than 0.001, the iteration stops, and the current... Output as a nonlinear stress resistance factor.

[0046] After obtaining the nonlinear compressive strength factor, it is mapped to a pre-defined compressive strength grading system to generate compressive strength test results. A pre-constructed compressive strength grading mapping table containing multiple continuous intervals is used. This table includes five continuous intervals: the first interval is for nonlinear compressive strength factor values ​​less than 0.20, corresponding to a strength level of "low"; the second interval is for nonlinear compressive strength factor values ​​greater than or equal to 0.20 and less than 0.40, corresponding to a strength level of "low-medium"; the third interval is for nonlinear compressive strength factor values ​​greater than or equal to 0.40 and less than 0.60, corresponding to a strength level of "medium"; the fourth interval is for nonlinear compressive strength factor values ​​greater than or equal to 0.60 and less than 0.80, corresponding to a strength level of "high-medium"; and the fifth interval is for nonlinear compressive strength factor values ​​greater than or equal to 0.80, corresponding to a strength level of "high". The target continuous interval in the compressive strength grading mapping table where the nonlinear compressive strength factor falls is queried, and the strength level identifier corresponding to the target continuous interval is extracted as the preliminary strength level of the current subsample. For the three nonlinear compressive strength factors calculated for the three subsamples, three preliminary strength levels were obtained by independently looking up the table.

[0047] The preliminary strength grades of all subsamples are merged through majority voting, and the strength grade with the most votes is taken as the compressive strength test result of the bentonite mud sample. The specific method of majority voting merging is as follows: count the strength grade identifier that appears most frequently among the three preliminary strength grades. If all three preliminary strength grades are different, the strength grade identifier with the median value is taken. If two are the same and one is different, the strength grade identifier with 2 votes is taken as the final compressive strength test result, and the final compressive strength test result is recorded and output.

[0048] In practice, after generating the compressive strength test results, these results are compared with historical test results to calculate the test deviation. Historical test results are stored in the system database; these are the most recent compressive strength test results obtained from bentonite slurry from the same source. Both the compressive strength test results and historical test results are represented by strength grade identifiers, which include five categories: "Low," "Low-Medium," "Medium," "High-Medium," and "High." Before comparison, the strength grade identifiers are converted to grade values, with the following conversion rules: "Low" corresponds to grade value 1, "Low-Medium" to grade value 2, "Medium" to grade value 3, "High-Medium" to grade value 4, and "High" to grade value 5. The test deviation is calculated as the absolute value of the difference between the grade value corresponding to the current test result and the grade value corresponding to the historical test result; the unit of the test deviation is the grade difference number.

[0049] When the test deviation exceeds the preset deviation alarm threshold, a recalibration process is triggered. The preset deviation alarm threshold is set to a difference of two levels. The basis for setting the deviation alarm threshold to 2 is: considering the natural fluctuation range of the strength grade of bentonite slurry due to different degrees of hydration and settling time in actual use, after statistically analyzing the distribution of grade changes between two consecutive measurements of a large number of field test samples, the probability of grade changes within two levels exceeds 90%. Therefore, the deviation alarm threshold is set to a difference of two levels to filter out normal fluctuations and only respond to abnormal jumps. When the test deviation is greater than 2, it is determined that there may be system drift or sensor failure in the current test, and the recalibration process is immediately triggered. In the recalibration process, multiple preset strength benchmark thresholds are reacquired. The reacquisition method is: calling three sets of bentonite standard samples with known strength grades from the standard sample library stored in the system. The strength grades of the three sets of bentonite standard samples are "low", "medium" and "high", corresponding to nonlinear compressive strength factor reference values ​​of 0.10, 0.50 and 0.90, respectively. Three sets of bentonite standard specimens were subjected to dynamic perturbation excitation, response waveform data acquisition, and structural response matrix construction according to the testing procedure, resulting in structural response matrices for the three sets of standard specimens. Using the structural response matrices of the three sets of standard specimens and known reference values ​​for the nonlinear compressive strength factor, new first and second strength benchmark thresholds were determined through inverse calculation, minimizing the sum of squared errors between the recalculated nonlinear compressive strength factor and the reference value. The new first and second strength benchmark thresholds replaced the original strength benchmark thresholds, and the nonlinear compressive strength factor was recalculated iteratively using the updated strength benchmark thresholds, thereby regenerating the compressive strength test results.

[0050] Applying dynamic perturbation excitation to bentonite slurry samples involves selecting three waveform types—sine wave, triangle wave, and square wave—from a pre-set perturbation waveform library. The pre-set perturbation waveform library is stored in the non-volatile memory of the signal generator. The library contains digital sequences of sine waves, triangle waves, and square waves, each with a frequency of 50 Hz. Each waveform sequence has 10,000 sampling points and a sampling rate of 10,000 Hz. The selected perturbation waveforms are grouped according to waveform type to generate multiple sets of perturbation excitation sequences. The sine wave perturbation excitation sequence consists of 50 sine waves with frequencies ranging from 0.1 Hz to 1000 Hz on a logarithmic scale; the triangle wave perturbation excitation sequence consists of 50 triangle waves with frequencies ranging from 0.1 Hz to 1000 Hz on a logarithmic scale; and the square wave perturbation excitation sequence consists of 50 square waves with frequencies ranging from 0.1 Hz to 1000 Hz on a logarithmic scale. In each set of perturbation excitation sequences, each frequency point is continuously excited for 10 complete waveform cycles.

[0051] Following a fixed sequence of sine wave, triangular wave, and square wave, multiple sets of disturbance excitation sequences were sequentially applied to the bentonite slurry sample in the test chamber. The disturbance waveform application sequence parameters were set in the signal generator's control program; the sequence parameter values ​​were fixed at 1 for a sine wave, 2 for a triangular wave, and 3 for a square wave, and could not be changed. The signal generator output the corresponding disturbance excitation sequence one by one according to the sequence parameter values ​​1 to 3, driving the electromagnetic vibrator to apply dynamic disturbance to the bentonite slurry sample. Each time the disturbance excitation sequence was switched, a no-excitation resting time window was inserted. The duration of the no-excitation resting time window was set to 60 seconds. The basis for setting the duration to 60 seconds was: through actual measurement of the pore pressure dissipation curves inside the bentonite slurry sample after three waveform disturbances, the pore pressure recovered to more than 95% of the pre-disturbance level after 60 seconds. Therefore, setting it to 60 seconds was to fully eliminate the residual influence of the previous disturbance on the response of the next disturbance.

[0052] During the application of each set of perturbation excitation sequences, the corresponding response waveform sub-data was independently acquired. The acquisition method was consistent with the aforementioned frequency domain response acquisition method. Four piezoelectric sensors attached to the outer wall of the sample container synchronously acquired the vibration response signals. After analog-to-digital conversion at a sampling frequency of 10000Hz, the signals were amplified by a charge amplifier and a data acquisition card. Sine wave response waveform sub-data was acquired during the application of the sinusoidal perturbation excitation sequence, triangular wave response waveform sub-data was acquired during the application of the triangular wave perturbation excitation sequence, and square wave response waveform sub-data was acquired during the application of the square wave perturbation excitation sequence. The three sets of response waveform sub-data were stored independently and labeled with waveform type tags.

[0053] Each set of response waveform sub-data is subjected to a Fourier transform to obtain the corresponding frequency domain response sub-data. A Fast Fourier Transform is performed on the signal from each sensor channel in the sinusoidal response waveform sub-data, and the spectral amplitude within the frequency range of 0.1Hz to 1000Hz is taken to form the sinusoidal frequency domain response sub-data. The same operation is performed on the triangular wave response waveform sub-data to form the triangular wave frequency domain response sub-data. The same operation is performed on the square wave response waveform sub-data to form the square wave frequency domain response sub-data. Each set of frequency domain response sub-data contains a sequence of spectral amplitude values ​​from four sensor channels.

[0054] Extract the dominant frequency energy value and dominant frequency offset from each set of frequency domain response sub-data. In the sinusoidal wave frequency domain response sub-data, retrieve the largest amplitude value in the spectral amplitude sequence as the sinusoidal wave dominant frequency energy value, record the frequency value corresponding to the largest amplitude point, and use the absolute value of the difference between the frequency value corresponding to the largest amplitude point and the fundamental frequency (50Hz) of the sinusoidal wave perturbation excitation sequence as the sinusoidal wave dominant frequency offset. In the triangular wave frequency domain response sub-data, retrieve the largest amplitude value in the spectral amplitude sequence as the triangular wave dominant frequency energy value, record the frequency value corresponding to the largest amplitude point, and use the absolute value of the difference between the frequency value corresponding to the largest amplitude point and the fundamental frequency (50Hz) of the triangular wave perturbation excitation sequence as the triangular wave dominant frequency offset. In the square wave frequency domain response sub-data, retrieve the largest amplitude value in the spectral amplitude sequence as the square wave dominant frequency energy value, record the frequency value corresponding to the largest amplitude point, and use the absolute value of the difference between the frequency value corresponding to the largest amplitude point and the fundamental frequency (50Hz) of the square wave perturbation excitation sequence as the square wave dominant frequency offset.

[0055] The dominant frequency energy value and dominant frequency offset are concatenated according to waveform type order to form a multidimensional response feature vector as the response waveform data. The waveform type order is fixed as sine wave first, triangle wave in the middle, and square wave last. The resulting multidimensional response feature vector is a 6-dimensional vector, with the six dimensions corresponding to: sine wave dominant frequency energy value, sine wave dominant frequency offset, triangle wave dominant frequency energy value, triangle wave dominant frequency offset, square wave dominant frequency energy value, and square wave dominant frequency offset. The multidimensional response feature vector serves as the response waveform data for subsequent structural response matrix construction.

[0056] Each element of the multidimensional response eigenvector is used as an initial column vector of the structural response matrix. The multidimensional response eigenvector has six elements. The first element is used as the first initial column vector of the structural response matrix, consisting of a single scalar value in one row and one column. The second element is used as the second initial column vector, and so on, up to the sixth element. The structural response matrix has six initial column vectors, each consisting of one row and one column.

[0057] Calculate the cross-correlation value between each initial column vector and the remaining initial column vectors. The cross-correlation value is calculated using the Pearson correlation coefficient. For the , The initial column vector and the first Initial column vectors, cross-correlation values The calculation formula is:

[0058] in, and All are initial column vector indices. and The values ​​range from 1 to 6 and ; Indicates the first The element values ​​of the initial column vector, i.e., the values ​​of the elements in the multidimensional response feature vector. The numerical values ​​of each dimension element; Indicates the first The element values ​​of the initial column vector, i.e., the values ​​of the elements in the multidimensional response feature vector. The numerical values ​​of each dimension element; Indicates the first The mean of the initial column vectors, since each initial column vector has only one element, the mean is equal to the element value itself; Indicates the first The mean of an initial column vector, where the mean is equal to the element value itself. Cross-correlation value. The value of is in the range of -1 to 1. The cross-correlation value is calculated pairwise for all initial column vectors, resulting in a 6x6 cross-correlation matrix.

[0059] Based on the magnitude of the cross-correlation values, the initial column vectors are reordered, with the column vectors having the largest cross-correlation values ​​arranged as adjacent columns, resulting in a reordered matrix. The specific reordering method is as follows: The element with the largest absolute cross-correlation value (excluding the diagonal) in the cross-correlation matrix is ​​located, and the indices of the two initial column vectors with the largest cross-correlation values ​​are determined. These two initial column vectors are then placed in the first and second columns. Next, in the remaining initial column vectors, the absolute cross-correlation value between each remaining initial column vector and the last column of the already sorted column vectors is calculated. The remaining initial column vector with the largest absolute cross-correlation value is then placed in the next column. This process is repeated until all initial column vectors are sorted, resulting in a reordered matrix with one row and six columns.

[0060] Principal component analysis (PCA) is performed on the reordered matrix to extract principal component components whose cumulative contribution rate reaches a preset contribution rate threshold. These principal component components form the dimensionality-reduced structural response matrix. The reordered matrix is ​​a 1x6 matrix; after transposing it to obtain a 6x1 matrix, PCA is performed. Eigenvalues ​​and eigenvectors of the covariance matrix are calculated. The eigenvalues ​​are sorted from largest to smallest, and the contribution rate of each eigenvalue is calculated. The contribution rate of a single eigenvalue is the eigenvalue divided by the sum of all eigenvalues, and the cumulative contribution rate is the sum of the contribution rates starting from the largest eigenvalue. The preset contribution rate threshold is set to 0.90. The rationale for setting this threshold is that in the multi-waveform response PCA of bentonite mud, the first few principal components typically contain the majority of data variation information; setting it to 0.90 effectively reduces the dimensionality of the features while preserving sufficient information. The first few principal component components whose cumulative contribution rate first reaches or exceeds 0.90 are selected, and their corresponding eigenvectors are extracted. These eigenvectors are used as columns of the dimensionality-reduced matrix. The reordered matrix is ​​projected onto the dimensionality-reduced matrix to obtain the dimensionality-reduced structural response matrix. If the cumulative contribution rate of the first two principal components has reached 0.90 or higher, the reduced structural response matrix contains two columns, each corresponding to a projection value in the direction of a principal component.

[0061] In specific implementation, please refer to Figure 4 A bentonite slurry compressive strength testing system includes a memory, a processor, and a computer program stored in the memory and running on the processor. The memory is a non-volatile solid-state memory with a capacity of 256GB, storing an operating system, device drivers, a signal processing algorithm library, and a computer program implementing the bentonite slurry compressive strength testing method. The processor is a 64-bit multi-core central processing unit with a clock speed of 2.5GHz. The processor is connected to the memory via a system bus, and reads and executes computer program instructions from the memory.

[0062] When executing the computer program, the processor implements the steps of a method for testing the compressive strength of bentonite slurry. First, the processor acquires a bentonite slurry sample and places it in a test chamber with dynamic loading capability. The processor sends a sampling command to an automatic sampling device via a serial communication interface. The automatic sampling device extracts multiple sub-samples from different depths of the bentonite slurry, performs air bubble removal and isothermal curing, and then places the sub-samples into the sample container within the test chamber. The processor acquires placement attitude information and sample volume information using a six-axis attitude sensor and a laser rangefinder connected to the bottom of the sample container, and stores these information in its memory.

[0063] When the processor executes the step of applying dynamic perturbation excitation to the bentonite slurry sample, it sends a perturbation excitation parameter configuration command to the signal generator. The signal generator generates a digital sequence of perturbation waveforms based on this command. This digital sequence, after passing through a digital-to-analog converter and a power amplifier, drives the electromagnetic vibrator. The test chamber includes a cylindrical rigid sample container, an electromagnetic vibrator, and a base. The electromagnetic vibrator is fixed to the base, and the sample container is screwed onto the drive end of the electromagnetic vibrator via a threaded connection. The processor controls the signal generator to output a sinusoidal wave perturbation excitation sequence, a triangular wave perturbation excitation sequence, or a square wave perturbation excitation sequence, causing the electromagnetic vibrator to apply dynamic perturbation excitation to the bentonite slurry sample in the sample container.

[0064] When the processor executes the step of acquiring response waveform data, it sends an acquisition start command to the multi-channel piezoelectric sensor acquisition module. The multi-channel piezoelectric sensor acquisition module includes four piezoelectric sensors, a four-channel charge amplifier, a four-channel synchronous analog-to-digital converter (ADC), and a first-in-first-out (FIFO) data buffer. The four piezoelectric sensors are attached to the outer wall of the sample container. The signal lines of the four piezoelectric sensors are connected to the input terminals of the four-channel charge amplifier, and the output terminals of the charge amplifiers are connected to the input terminals of the four-channel FIFO. The FIFO converts the analog voltage signal into a 16-bit digital signal at a sampling frequency of 10000Hz. The digital signal is then transmitted to the processor via the FIFO data buffer through a high-speed serial peripheral interface bus. The processor stores the received vibration response signal in segments according to the time periods of different disturbance excitation sequences, forming response waveform sub-data.

[0065] When the processor executes the step of constructing the structural response matrix, it calls the signal processing algorithm library in memory to perform operations on the response waveform data, such as extracting peak amplitude and phase delay, calculating dynamic stiffness coefficients, normalizing and performing singular value decomposition or Fourier transform, extracting the dominant frequency energy and dominant frequency offset, calculating cross-correlation, reordering, and performing principal component analysis, to generate the structural response matrix. The data structure of the structural response matrix is ​​stored in the processor's cache as a two-dimensional floating-point array.

[0066] When the processor performs the iterative calculation of the nonlinear compressive strength factor, it reads multiple preset strength benchmark thresholds from memory. These thresholds include a first strength benchmark threshold and a second strength benchmark threshold. The processor iteratively calculates the nonlinear compressive strength factor according to a recursive update formula. The momentum factor and learning step size coefficient introduced in the recursive update formula are stored as preset parameters in the parameter configuration area of ​​memory. In each iteration, the processor calculates the intermediate regression parameters and determines whether the rate of change of the intermediate regression parameters is less than a preset rate of change threshold. When the rate of change is less than the rate of change threshold, the iteration stops and the nonlinear compressive strength factor is output.

[0067] When the processor executes the step of generating compressive strength test results, it reads the compressive strength grading mapping table from memory. This table is stored using a lookup table data structure, and the nonlinear compressive strength factor is used as the index input to query the corresponding strength grade identifier. The processor then performs a majority voting fusion algorithm on the strength grade identifiers of all sub-samples. The algorithm determines the strength grade identifier with the most votes through a counting and sorting method, and outputs the strength grade identifier with the most votes as the compressive strength test result for the bentonite mud sample. The output methods include displaying the compressive strength test results on an LCD screen connected to the processor, and simultaneously uploading the results to a remote data server via an Ethernet communication module.

[0068] The processor also implements a test deviation comparison and recalibration process when executing the computer program. The processor reads the most recent compressive strength test result of bentonite slurry from the same source from the historical test database in memory, and calculates the absolute value of the difference between the current compressive strength test result's grade value and the historical compressive strength test result's grade value as the test deviation value. When the test deviation value exceeds a preset deviation alarm threshold, the processor triggers the recalibration process, sends a standard sample test command to the signal generator, drives the automatic sampling device to switch to the bentonite standard sample in the standard sample library, obtains the structural response matrix of the standard sample according to the normal test process, solves in reverse and updates the first and second strength reference thresholds, stores the updated strength reference thresholds in memory, and recalculates the nonlinear compressive strength factor using the updated strength reference thresholds.

[0069] In some embodiments, the processor is connected to the disturbance excitation module and the multi-channel piezoelectric sensor acquisition module via an industrial Ethernet bus. The disturbance excitation module includes a signal generator, a power amplifier, and an electromagnetic exciter. The processor sends a frequency point list, waveform type identifier, and excitation duration parameters to the disturbance excitation module via the industrial Ethernet bus. The disturbance excitation module autonomously generates a disturbance excitation sequence based on the received parameters, realizing distributed control of the disturbance excitation and reducing the real-time control load on the processor.

[0070] In some embodiments, the multi-channel piezoelectric sensor acquisition module has a built-in digital signal processor. The digital signal processor performs real-time bandpass filtering on the acquired vibration response signal. The passband frequency range of the bandpass filter is automatically adjusted according to the frequency range of the currently applied disturbance excitation sequence to filter out environmental mechanical vibration and electromagnetic interference, thereby improving the signal-to-noise ratio of the response waveform data.

[0071] In some embodiments, the testing system further includes a temperature control module, which comprises a heating film attached to the outer wall of the sample container, a temperature sensor, and a temperature controller. The temperature controller is connected to the processor via a serial bus. When performing the isothermal curing process, the processor sends a temperature setpoint to the temperature controller. The temperature controller adjusts the heating power of the heating film according to the temperature setpoint to maintain the internal temperature of the sample container at the preset temperature value, thus maintaining the temperature consistency of the bentonite slurry sample during the testing process.

[0072] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. A method for testing the compressive strength of bentonite slurry, characterized in that, The method includes: Obtain bentonite slurry samples and place them in a test chamber with dynamic loading function; A dynamic perturbation excitation is applied to the bentonite slurry sample, and the response waveform data of the bentonite slurry sample under the dynamic perturbation excitation are collected, specifically including: A sinusoidal perturbation sequence with linearly increasing frequency is sequentially applied to each subsample in the test cavity; During the application of the sinusoidal perturbation sequence, the vibration response signal of each sub-sample is synchronously acquired by multiple piezoelectric sensors attached to the outer wall of the sample container. The vibration response signal of each subsample is segmented in the time domain to obtain response waveform segments corresponding to multiple time windows for each subsample; All response waveform segments of each subsample are arranged in frequency order to form the response waveform data; The frequency range of the linearly increasing sinusoidal perturbation sequence is from 0.1 Hz to 1000 Hz, and the frequency intervals between adjacent frequency points are distributed on a logarithmic scale. The structural response matrix of the bentonite mud sample is constructed based on the response waveform data, specifically including: Extract the peak amplitude and phase delay of each subsample in the response waveform segment of each time window; Based on the peak amplitude and the phase delay, calculate the dynamic stiffness coefficient of each subsample at each frequency point; Arrange the dynamic stiffness coefficients of all subsamples at all frequency points according to the subsample index and frequency index to form the initial stiffness matrix; Normalization and singular value decomposition are performed on the initial stiffness matrix to obtain the reduced-order structural response matrix; Based on the structural response matrix and multiple preset strength benchmark thresholds, the nonlinear compressive strength factor of the bentonite slurry sample is calculated iteratively. The nonlinear compressive strength factor is mapped to a preset compressive strength grading system to generate the compressive strength test results of the bentonite slurry sample.

2. The method for testing the compressive strength of bentonite slurry according to claim 1, characterized in that, The process of obtaining bentonite slurry samples and placing them in a test chamber with dynamic loading capabilities specifically includes: Multiple sub-samples were extracted from different depth locations of the bentonite mud to be tested; Each subsample was subjected to bubble removal and isothermal curing treatment. Each sub-sample, after being filtered to remove air bubbles and cured at a constant temperature, is placed sequentially into the sample container of the test chamber; Record the placement orientation and volume information of each subsample in the sample container.

3. The method for testing the compressive strength of bentonite slurry according to claim 2, characterized in that, The bubble removal process employs a vacuum degassing device to continuously process each subsample for 5 to 10 minutes under negative pressure.

4. The method for testing the compressive strength of bentonite slurry according to claim 3, characterized in that, When performing normalization on the initial stiffness matrix, the maximum value of the dynamic stiffness coefficient of all subsamples at all frequency points is used as the normalization benchmark.

5. The method for testing the compressive strength of bentonite slurry according to claim 4, characterized in that, The iterative calculation of the nonlinear compressive strength factor of the bentonite slurry sample based on the structural response matrix and multiple preset strength benchmark thresholds specifically includes: Assign an initial weight coefficient to each element in the structure response matrix; The structural response matrix is ​​compared with a first strength benchmark threshold, and the initial weight coefficient is adjusted according to the result of the first comparison to obtain the first adjusted weight. The structural response matrix after the first adjustment weight is compared with the second intensity benchmark threshold, and intermediate regression parameters are generated based on the result of the second comparison. Substitute the intermediate regression parameters into the recursive update formula of the nonlinear stress resistance factor, and repeat the intermediate regression parameter generation step and the recursive update step until the rate of change of the intermediate regression parameters is less than the preset rate of change threshold, then stop the iteration and output the current nonlinear stress resistance factor.

6. The method for testing the compressive strength of bentonite slurry according to claim 5, characterized in that, The recursive update formula introduces a momentum factor to suppress the iterative oscillation of the nonlinear resistance factor.

7. A bentonite slurry compressive strength testing system, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method for testing the compressive strength of bentonite slurry as described in any one of claims 1 to 6.

Images