A building engineering survey underground disease body detection method and system based on high-density resistivity method

Abstract

This disclosure provides a method and system for detecting underground defects in building engineering based on high-density resistivity, belonging to the field of building engineering technology. The method includes: acquiring current field data of a multi-pole electrode array deployed on the surface of the construction area under different power supply modes; identifying the local distortion amplitude and phase lag characteristics at the edge of the fracture development zone based on the current field data; adaptively adjusting the sampling interval and power supply cycle based on the local distortion amplitude and phase lag characteristics; and performing high-density resistivity data acquisition and defect location according to the adjusted sampling parameters. Through the scheme of this disclosure, the adaptive sampling interval and power supply cycle can be adjusted according to the local distortion amplitude and phase lag characteristics generated by the current field at the edge of the fracture development zone to solve the problem of missed sampling of resistivity abrupt change signals caused by micro-hidden fractures.

Description

Technical Field

[0001] This application relates to the field of building engineering technology, specifically to a method and system for detecting underground defects in building engineering based on the high-density resistivity method. Background Technology

[0002] High-density resistivity (HDR) is a non-destructive and efficient geophysical exploration technique. With its advantages of flexible depth, high spatial resolution, and convenient construction, it has been widely applied in the field of building engineering surveys for detecting underground karst, fissures, and fault zones. It is a crucial technique for identifying potential underground geological hazards in advance and ensuring the safety of construction and the long-term stability of structures. However, in current applications, conventional HDR methods mostly employ standardized data acquisition modes with fixed sampling intervals and fixed power supply cycles. These methods cannot adapt to the varying electric field response characteristics of heterogeneous strata within building sites, especially for small-scale, highly concealed faults, which can cause significant fluctuations at the edges of fissure development zones. The weak local distortion and phase lag of the current field make it easy for fixed sampling parameters to miss such resistivity abrupt changes, making it difficult to accurately capture the electric field response characteristics of tiny hidden faults. At the same time, most existing technologies only identify and adjust the parameters based on the single apparent resistivity amplitude change, without fully combining the dual characteristics of the local distortion amplitude and phase lag of the current field to achieve adaptive and coordinated control of the sampling interval and power supply cycle. This makes it impossible to effectively improve the sensitivity and positioning accuracy of tiny hidden faults while taking into account the efficiency of on-site detection. Problems such as fault boundary positioning deviation, missed detection, and misjudgment are prone to occur, making it difficult to fully meet the application requirements of modern building engineering survey for high-precision and high-reliability detection of underground tiny hidden faults. Summary of the Invention

[0003] In view of this, the present disclosure provides a method for detecting underground defects in building engineering based on the high-density resistivity method, which at least partially solves the problems existing in the prior art.

[0004] In a first aspect, the present invention provides a method for detecting underground defects in building engineering based on the high-density resistivity method, comprising: Acquire current field data of multi-pole electrode arrays deployed on the surface of the construction area under different power supply modes; Based on the current field data, the local distortion amplitude and phase hysteresis characteristics at the edge of the fracture development zone are identified; Based on the local distortion amplitude and phase lag characteristics, the sampling interval and power supply cycle are adaptively adjusted. High-density resistivity data acquisition and lesion location were performed based on the adjusted sampling parameters.

[0005] In one specific implementation, identifying the local distortion amplitude and phase hysteresis characteristics at the edge of the fracture development zone based on the current field data includes: Obtain the current density difference ΔJ between each electrode pair; Calculate the rate of change of current density gradient γ = ΔJ / (Δx × Δy), where Δx is the transverse electrode spacing and Δy is the longitudinal electrode spacing; Based on the above γ, the local distortion amplitude η = |γ| / γ_mean is calculated, where γ_mean is the rate of change of the average current density gradient in the normal formation. If η > the preset threshold η_threshold, it is determined to be the edge of the fracture development zone.

[0006] In one specific implementation, adaptively adjusting the sampling interval and power supply cycle based on the local distortion amplitude and phase lag characteristics includes: The sampling interval Δt = Δt_base × (η_avg / η) is adjusted according to the distortion amplitude η at the edge of the fracture development zone, where Δt_base is the default sampling interval and η_avg is the average distortion amplitude of the background area. Obtain the phase difference Δφ between adjacent points; The power supply cycle T_adjust=T_base+(Δφ / φ_rate) is compensated by combining Δφ, where φ_rate is the preset phase change rate; The compensated power supply cycle is weighted and smoothed using the weighted formula T_final=α×T_adjust+(1-α)×T_base, where α is the weighting factor.

[0007] In one specific implementation, the adaptive adjustment of the sampling interval and power supply cycle includes: Define the local distortion index of the electrode pair as D_index=(η×Δφ) / ρ_real, where ρ_real is the real part of the apparent resistivity of the region; The complexity of the fracture can be determined by D_index. If D_index > D_threshold, it indicates the presence of a small, hidden fracture. If D_index is detected to exceed the threshold, enter high-frequency scanning mode, and the sampling frequency f_sampling=f_base×(D_index / D_threshold); The power supply cycle is set to T_supply=T_base×(D_index / D_threshold) to enhance the sampling coverage of small signals.

[0008] In one specific implementation, the high-density resistivity data acquisition and lesion localization based on the adjusted sampling parameters includes: Perform current injection for multiple electrode arrangements at the adjusted sampling frequency and power supply period; Collect the resistivity response data set and calculate its standard deviation σ_rho = sqrt(1 / n × Σ(rho_i - rho_avg)^2), where rho_i is the single measurement value; Extract the mutation points as the potential disease body positions in the data by setting the standard deviation threshold τ; Use the least squares method to fit the electric field distribution characteristics of the mutation region to determine the boundary of the disease body.

[0009] In a specific embodiment, the high-density resistivity data collection and disease body positioning according to the adjusted sampling parameters further include: Construct a three-dimensional apparent resistivity imaging model and simulate the electric field change based on finite element analysis; Calculate the slope K of the apparent resistivity change curve between electrodes, K = Δρ / Δt, where Δρ is the resistivity change per unit time and Δt is the time increment; Judge whether the absolute value of the slope |K| is greater than the preset critical value K_threshold, and if it is satisfied, mark it as a suspicious point; Evaluate the authenticity and position consistency of the suspicious points in combination with the spatial correlation matrix to avoid misjudgment.

[0010] In a specific embodiment, the obtaining of the current field data under different power supply modes includes: Select the bipolar, tripolar and quadrupolar power supply modes for multiple measurements; For each power supply mode, obtain the electric field response data and calculate its signal-to-noise ratio SNR = P_signal / P_noise, where P_signal is the effective signal power and P_noise is the noise power; Use a conditional judgment statement to judge whether to switch the power supply mode: switch to the high-resolution power supply mode when SNR < SNR_threshold; If the SNR standard is not met for three consecutive scan cycles, enable multi-angle electrode arrangement to improve the signal-to-noise ratio.

[0011] In a specific embodiment, the adaptive adjustment of the sampling interval and power supply period further includes: Convert the local distortion amplitude η and phase difference Δφ into a perturbation index F = η + Δt_phase / T_base, where T_base is the reference power supply period; Determine the regulation intensity by judging the magnitude of the perturbation index F. F > F_threshold indicates that sampling should be significantly enhanced; The sampling interval Δt = Δt_initial × e^(k_F × F) is adjusted using an exponential decay function, where k_F is the adjustment coefficient. Adjust the power supply cycle using the following formula: T_supply = T_base × e^(k_T × F), where k_T is another adjustment coefficient.

[0012] In one specific embodiment, the high-density resistivity data acquisition and lesion localization based on the adjusted sampling parameters further includes: Based on the new sampling interval, block data processing is performed, and the length of each data block is n_block=round(Δt / Δt_resolution); Fast Fourier Transform is used to perform spectral analysis on the block data to extract the low-frequency components and reduce interference; The presence of abrupt changes is determined based on the peak value of the amplitude spectrum, and the kurtosis index Sk=(ρ_max-ρ_mean) / σ_rho characterizes the suddenness. If Sk > Sk_threshold and synchronous oscillations occur across multiple frequency bands, it is determined to be caused by a micro-hidden fault.

[0013] Secondly, the present invention provides a system for detecting underground defects in building engineering based on the high-density resistivity method, comprising: The data acquisition module acquires current field data of the multi-pole electrode array deployed on the surface of the construction area under different power supply modes; The identification module identifies the local distortion amplitude and phase hysteresis characteristics at the edge of the fracture development zone based on the current field data; The adjustment module adaptively adjusts the sampling interval and power supply cycle based on the local distortion amplitude and phase lag characteristics. The positioning module performs high-density resistivity data acquisition and disease location based on the adjusted sampling parameters.

[0014] This disclosure provides a method for detecting underground defects in building engineering based on high-density resistivity, comprising: acquiring current field data of a multi-pole electrode array deployed on the surface of the construction area under different power supply modes; identifying the local distortion amplitude and phase lag characteristics at the edge of the fracture development zone based on the current field data; adaptively adjusting the sampling interval and power supply cycle based on the local distortion amplitude and phase lag characteristics; and performing high-density resistivity data acquisition and defect location according to the adjusted sampling parameters. The scheme of this disclosure can address the problem of missed resistivity abrupt change signals caused by micro-hidden fractures by adjusting the adaptive sampling interval and power supply cycle according to the local distortion amplitude and phase lag characteristics generated by the current field at the edge of the fracture development zone. Attached Figure Description

[0015] In the accompanying drawings, unless otherwise specified, the same reference numerals throughout the various drawings denote the same or similar parts or elements. These drawings are not necessarily drawn to scale. It should be understood that these drawings depict only some embodiments disclosed in this application and should not be construed as limiting the scope of this application.

[0016] Figure 1 This is a flowchart of a method for detecting underground defects in building engineering based on the high-density resistivity method, according to the present invention. Figure 2 This is a flowchart of the present invention for obtaining current field data of a multi-pole electrode array deployed on the surface of the construction area under different power supply modes. Figure 3 This is a block diagram of a building engineering survey underground disease detection system based on the high-density resistivity method according to the present invention. Detailed Implementation

[0017] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0018] Next, refer to Figure 1 This invention describes a method for detecting underground defects in building engineering based on the high-density resistivity method, the method comprising the following steps: S101. Acquire current field data of a multi-pole electrode array deployed on the surface of the construction area under different power supply modes. In the target building construction area, first complete the site surface clearing, removing surface metal debris, accumulated water, and other sources of electromagnetic interference. Then, deploy a high-density multi-pole electrode array along the predetermined survey line direction. The electrodes are low-grounding-resistance copper cylindrical grounding electrodes, with an insertion depth of no less than 20cm. Conductive mud is backfilled at the bottom of the electrodes to ensure good low-impedance contact between the electrodes and the strata. The electrode spacing is set according to the exploration depth and site strata conditions. For shallow exploration (0-20m), a 1m-5m electrode spacing is used; for medium-deep exploration (20-50m), a 5m-10m electrode spacing is used. The electrode array adopts a grid structure with equal horizontal spacing and 2-4 parallel rows vertically, forming a three-dimensional electrode arrangement system covering the entire exploration area. The spacing between adjacent survey lines is consistent with the horizontal electrode spacing to ensure spatial sampling uniformity in the detection area. After electrode deployment, data acquisition was conducted using three mainstream high-density electrical resistivity tomography (EDT) power supply modes: bipolar-dipole, tripolar, and Wenner quadrupole. Pre-set baseline acquisition parameters were: adjustable baseline power supply voltage (12V-48V), baseline power supply current not less than 10mA, default sampling interval Δt_base of 50ms, and baseline power supply period T_base of 200ms. For each power supply mode, 3-5 rounds of current injection and electric field response acquisition were performed. After each round, the signal-to-noise ratio (SNR) of the acquired data under that power supply mode was calculated in real time. When the SNR fell below a preset 30dB threshold, automatic... Switching to the high-resolution Wenner quadrupole power supply mode, if the signal-to-noise ratio still does not meet the threshold requirement after three consecutive scan cycles, the electrode array layout angle is adjusted. A multi-angle cross-arrangement of 0°, 45°, and 90° is adopted to suppress environmental electromagnetic interference and improve the data signal-to-noise ratio. Finally, the original current field data such as voltage response, current intensity, apparent resistivity, and mutual impedance of each electrode pair under different power supply modes are obtained, and the original data is preprocessed. The preprocessing operations include removing abnormal jump values, filtering 50Hz power frequency interference, baseline drift correction, and removing abnormal grounding resistance values ​​to complete the acquisition of high-quality original current field data.

[0019] S102, Based on the current field data, identify the local distortion amplitude and phase lag characteristics at the edge of the fracture development zone. Based on the preprocessed current field data, first select a background area with uniform strata and no diseased formations within the exploration area, and calculate the average current density J_mean of normal strata through multi-point weighted averaging; then perform spatial difference calculation on the current field data of adjacent electrode pairs one by one to obtain the current density difference ΔJ between adjacent electrode pairs. Combined with the transverse electrode spacing Δx and the longitudinal electrode spacing Δy of the electrode array, calculate the current density gradient change rate γ=ΔJ / (Δx×Δy) to quantify the degree of spatial change of the current field and capture the local abrupt change characteristics of the current field. Based on the calculated rate of change of current density gradient γ, the local distortion amplitude η = |γ| / γ_mean is further calculated, where γ_mean is the rate of change of the formation's average current density gradient. The local distortion amplitude characterizes the degree of distortion of the current field relative to a normal homogeneous formation. The calculated local distortion amplitude η is compared with a preset distortion threshold η_threshold (in this embodiment, η_threshold is set to 2.5). When η > η_threshold, the location of the measuring point is determined to be the edge of a fracture development zone. Simultaneously, cross-correlation analysis and phase calculation are performed on the time-domain signals of the current field from adjacent sampling points to extract the phase difference Δφ of the electric field response between adjacent measuring points. This completes the accurate identification of the dual characteristics of local distortion amplitude and phase lag at the edge of the fracture development zone, providing core data basis for the adaptive adjustment of subsequent sampling parameters.

[0020] S103, based on the local distortion amplitude and phase lag characteristics, adaptively adjust the sampling interval and power supply cycle. Based on the identified local distortion amplitude η and phase difference Δφ, first calculate the average distortion amplitude η_avg of the background area of ​​the survey region. Using the default sampling interval Δt_base set in step S101 as a benchmark, complete the initial adaptive adjustment of the sampling interval according to the relative magnitude of the distortion amplitude. The adjustment formula is Δt=Δt_base×(η_avg / η), realizing differentiated control with larger distortion amplitude, shorter sampling interval, and higher spatial sampling density, ensuring full signal coverage acquisition in areas with strong distortion and avoiding missed acquisition of abrupt signal changes. Simultaneously, based on the identified phase difference Δφ, combined with the preset phase change rate φ_rate (in this embodiment, φ_rate is set to π / 100 rad / ms, i.e., the phase changes by π / 100 radians per millisecond), phase lag compensation is performed on the reference power supply cycle T_base to obtain the compensated power supply cycle T_adjust=T_base+(Δφ / φ_rate), eliminating the signal acquisition deviation and phase distortion caused by phase lag; then, a weighting factor α is introduced, with a value ranging from 0.3 to 0.7, which can be flexibly adjusted according to the ambient noise level of the site. A lower value is used in high noise environments to reduce adjustment fluctuations, and a higher value is used in low noise environments to improve adjustment sensitivity. The final power supply cycle is calculated using the weighted formula T_final=α×T_adjust+(1-α)×T_base, realizing the coordinated adaptive control of the sampling interval and the power supply cycle. In addition, when the local distortion amplitude η of the measurement point is detected to exceed the threshold by more than 3 times, the high-frequency fine scanning mode is automatically triggered to further compress the sampling interval, optimize the power supply cycle, and improve the ability to capture weak distortion signals, thereby fundamentally solving the problem of missed sampling of resistivity change signals caused by small hidden fractures.

[0021] S104. High-density resistivity data acquisition and lesion location are performed based on the adjusted sampling parameters. Based on the core sampling parameters such as sampling interval, power supply cycle, and sampling frequency adaptively adjusted in step S103, 5-8 rounds of cyclic current injection and data acquisition are performed on the electrode array in the survey area. Each round of acquisition covers all effective electrode pairs within the array, obtaining a high-resolution, high-signal-to-noise ratio resistivity response dataset. Data quality is verified on the acquired dataset. The standard deviation of the resistivity value measured at each measuring point is calculated as σ_rho=sqrt(1 / nΣ(rho_i-rho_avg)²), where rho_i is the resistivity value measured once, rho_avg is the average value of multiple measurements at that measuring point, and n is the number of measurements. A standard deviation threshold τ is set, and invalid data with a standard deviation exceeding the threshold τ are removed. Valid data is retained, and resistivity abrupt change points are extracted and marked as potential lesion locations. Subsequently, the electric field distribution characteristics of the abrupt change region were fitted using the least squares method to accurately delineate the spatial boundary and distribution range of the disease. At the same time, a three-dimensional apparent resistivity imaging model of the survey area was constructed. Based on finite element analysis, the distribution and variation of the underground electric field were simulated. The location of the marked potential disease was verified by combining the spatial correlation matrix. Misjudgments caused by environmental noise and poor electrode contact were eliminated. Finally, the accurate location, boundary delineation and three-dimensional imaging output of the underground disease were completed, providing a reliable geological basis for architectural engineering survey and design and construction safety protection.

[0022] In another embodiment, the high-density resistivity data acquisition and lesion localization based on the adjusted sampling parameters specifically includes performing multiple rounds of high-density resistivity data acquisition on the construction area according to the adaptively adjusted sampling interval and power supply cycle. During each round of acquisition, the apparent resistivity time series and phase response of each electrode pair are recorded synchronously. For the time-domain signal of each measurement point, the continuous wavelet transform (CWT) is used to decompose the signal into different scales. A scale factor matching the adjusted sampling interval is selected, and the wavelet coefficient modulus maxima at that scale are extracted as abrupt change candidate points. Then, spatial clustering is performed on the modulus maxima points under different power supply modes. If the same location has a modulus maxima in more than three power supply modes, it is determined to be a lesion signal point. Finally, the Canny edge detection algorithm is used to perform edge tracking on all signal points, outputting a closed lesion boundary. The depth coordinates of the boundary points are corrected by combining the phase compensation parameters of the power supply cycle to eliminate the depth offset caused by phase lag, thus completing the lesion localization.

[0023] In another embodiment, after acquiring a high-density apparent resistivity dataset based on the adjusted sampling parameters, a three-dimensional resistivity grid model is constructed. The grid resolution is proportional to the adjusted sampling interval (the smaller the sampling interval, the higher the grid density). The resistivity gradient vector of each grid node is calculated using the finite difference method to form a gradient field in the entire space. The position of the electrode pair when the local distortion index is greater than the threshold is used as the seed point. A three-dimensional region growth algorithm is executed in the gradient field. The growth criterion is that the angle between the resistivity gradient directions of adjacent grid nodes is less than 30° and the gradient amplitude change rate is less than 50%. After the region growth converges, all grid nodes in the growth area are marked as the interior of the disease body, and their convex hulls are extracted as the three-dimensional boundary of the disease body. Finally, the boundary coordinates are anisotropically scaled according to the adjustment coefficient of the power supply cycle to compensate for the difference in electric field diffusion depth caused by the change of power supply cycle, and the spatial location and volume of the underground disease body are output.

[0024] In one embodiment of the present invention, identifying the local distortion amplitude and phase hysteresis characteristics at the edge of a fracture development zone based on current field data specifically includes the following steps: The first step, after completing the preprocessing of the raw current field data (including 50Hz power frequency notch filtering, baseline drift correction, removal of grounding resistance anomalies, and removal of bad circuit data), is to use the multi-pole electrode array deployed in the construction area as a basis and adopt the 3×3 grid sliding window traversal method to fully cover all adjacent electrode pairs in the measurement line. The sliding window step size is consistent with the transverse electrode spacing Δx to ensure that no measurement points are missed and no area overlap is deviated during the traversal process. For each power supply electrode-measuring electrode pair within the window, based on the fundamental theory of stable current fields, and combined with the synchronously acquired stable injection current intensity I, inter-electrode potential difference U, electrode spatial layout coordinates, and measured formation apparent resistivity ρ, the formation current density vector value J corresponding to that measuring point is calculated point by point. The current density calculation formula is J=(I×ρ) / (2π×r²)×e_r, where r is the straight-line distance from the measuring point to the power supply electrode, and e_r is the unit vector of the current conduction direction. Vector calculation synchronously obtains the amplitude and direction information of the current density, compensating for the deficiency of traditional scalar calculations in identifying current field direction distortion. For the calculation results of multiple electrode pairs at the same measuring point, median filtering is used to remove outliers exceeding ±3 standard deviations. The arithmetic mean of the remaining valid data is then taken to obtain the final current density scalar value for that measuring point. Finally, the current density field grid construction of the entire exploration area is completed, with the grid resolution perfectly matching the electrode array layout resolution. The second step is to complete the standardized calculation of current density differences and gradient change rates. For the constructed gridded current density field, spatial difference calculations are performed along the transverse and longitudinal orthogonal directions of the survey line: In the transverse direction, the absolute value of the current density difference between two adjacent measuring points is calculated to obtain the transverse current density difference ΔJ_x=|J(x+Δx,y)-J(x,y)|; in the longitudinal direction, the absolute value of the current density difference between two adjacent measuring points is calculated to obtain the longitudinal current density difference ΔJ_y=|J(x,y+Δy)-J(x,y)|; the maximum value in the two orthogonal directions is taken as the final current density difference ΔJ=max(ΔJ_x,ΔJ_y) of the measuring point, ensuring that the sudden change in current field caused by the development zone of fractures with arbitrary orientation can be captured, and avoiding the omission of oblique fractures caused by single-direction difference. Based on the obtained ΔJ, and combined with the corresponding transverse electrode spacing Δx and longitudinal electrode spacing Δy at the measuring point, the rate of change of current density gradient at the measuring point is calculated as γ = ΔJ / (Δx × Δy), where the units of Δx and Δy are unified to meters (m), and the unit of ΔJ is A / m². The final unit of γ is A / m. 4This gradient change rate can accurately quantify the variation amplitude of current density per unit area, eliminating the influence of differences in electrode spacing and exploration depth on the identification of current field changes, and standardizing the identification results under different deployment parameters and different exploration scenarios. The third step involves pre-calibrating the background area within the exploration area. The selection of the background area must simultaneously meet three hard conditions: first, it must be located within the exploration area, far away from known geological structures, underground pipelines, artificial structures, and other sources of electromagnetic interference; second, the strata must be homogeneous, with a continuous distribution range of not less than 10m × 10m, covering no less than 20% of the total number of measuring points; and third, the apparent resistivity variation coefficient within the area must not exceed 10%, ensuring that the strata homogeneity meets the benchmark calibration requirements. The arithmetic mean of the current density gradient change rate of all measuring points in the background area is calculated to obtain the average current density gradient change rate γ_mean. Simultaneously, the average distortion amplitude η_avg of the background area is calculated to provide a benchmark value for subsequent sampling parameter adjustments. Based on the above calibration parameters, the local distortion amplitude η = |γ| / γ_mean of each measuring point in the entire exploration area is calculated point by point. This relative value calculation method can eliminate the differences in background current density caused by different lithologies and different injection current intensities, and realize the uniform quantification of distortion degree under different sites and different acquisition parameters. A distortion threshold η_threshold is preset. This threshold is set according to the statistical results of the distortion amplitude in the background area. Three times the maximum value of η in the background area is taken as the benchmark threshold. In this embodiment, for the Quaternary loose strata building exploration scenario, η_threshold is set to 2.5. For bedrock exposed strata, η_threshold is increased to 3.0 to reduce misjudgment caused by the original heterogeneity of the strata. The η of each measuring point is compared with η_threshold one by one. If η > η_threshold, the measuring point is determined to be a feature point on the edge of the fracture development zone. At the same time, DBSCAN spatial clustering analysis is performed on all feature points to remove isolated discrete feature points and retain the continuously distributed feature point set, thus completing the spatial delineation of the edge of the fracture development zone. Simultaneously, cross-correlation analysis is performed on the time-domain signals of the current field of adjacent feature points. Using the fundamental phase of the power supply current as a reference, the phase difference between the measured signal and the reference signal is calculated to obtain the phase difference Δφ between adjacent points. Finally, the entire process of identifying the local distortion amplitude and phase lag characteristics at the edge of the fracture development zone is completed.

[0025] Furthermore, in this embodiment, the adaptive adjustment of the sampling interval and power supply cycle based on the local distortion amplitude and phase lag characteristics includes the following specific steps: The first step involves adaptively adjusting the sampling interval based on the local distortion amplitude η of the feature points at the edge of the fracture development zone, the pre-calibrated average distortion amplitude η_avg of the background area, and the pre-set default sampling interval Δt_base. The adjustment formula is Δt = Δt_base × (η_avg / η). Here, the default sampling interval Δt_base is the instrument's time-domain sampling interval, pre-graded according to the exploration depth and site noise level: in this embodiment, Δt_base is set to 50ms for shallow building engineering exploration (0-30m), 100ms for medium-deep exploration (30-60m), and 200ms for deep exploration (over 60m). Differential adjustments are made for different feature points at the edge of the fracture development zone based on their respective η values. A larger η value indicates more severe current field distortion, resulting in a shorter adjusted sampling interval Δt and a higher time-domain sampling density, ensuring complete capture of rapidly changing resistivity abrupt changes. Simultaneously, safe upper and lower limits for the sampling interval are set. The minimum sampling interval is not less than 10ms to avoid overloading the instrument's acquisition system and excessive amplification of high-frequency noise due to excessively short sampling intervals; the maximum sampling interval does not exceed four times the reference value to avoid missing effective signals due to excessively long sampling intervals, ensuring that the adjusted sampling interval is always within the instrument's safe operating range and effective detection range. The second step involves dynamic phase compensation of the power supply cycle based on the calculated phase difference Δφ. First, the phase change rate φ_rate is preset. This parameter is set in conjunction with the instrument's power supply frequency and sampling frequency. In this embodiment, the instrument's reference power supply frequency is 5Hz, corresponding to a reference power supply cycle of 200ms. The preset phase change rate φ_rate is set to π / 100rad / ms, meaning a phase change of π / 100 radians per millisecond. Based on the above parameters, the phase-compensated power supply cycle T_adjust = T_base + (Δφ / φ_rate) is calculated, where T_base is the set reference power supply cycle, which is set to 200ms in this embodiment. When a positive phase difference Δφ is detected, it indicates a phase delay in the measured signal. Phase compensation is achieved by extending the power supply cycle, eliminating signal truncation errors caused by phase lag and ensuring the acquisition of a complete steady-state electric field response signal. When Δφ is 0 or negative, T_adjust equals T_base, and no additional compensation is performed to avoid reduced detection efficiency due to excessively extended power supply cycles. The third step involves introducing a weighting factor α to perform weighted smoothing on the compensated power supply cycle. The weighting calculation formula is T_final = α × T_adjust + (1-α) × T_base, where α is the weighting factor, ranging from 0.2 to 0.8. Its core function is to control the adjustment intensity of the power supply cycle, avoiding instability and poor data continuity caused by frequent parameter jumps.The weighting factor α is dynamically set according to the site's environmental noise level and geological conditions: in suburban sites with low electromagnetic interference and uniform geological formations, α is set to 0.6-0.8 to maximize the phase compensation effect and improve the sensitivity of parameter adjustment; in densely populated urban areas, under high-voltage lines, and near subways, where electromagnetic interference is high, α is set to 0.2-0.4 to reduce the fluctuation of the power supply cycle and improve the anti-interference capability and data stability of the acquisition system; in conventional building engineering survey sites, α is set to a default value of 0.5 to balance adjustment sensitivity and system stability. For adjacent measuring points on the same survey line, a 5-point moving average method is used to perform secondary smoothing on the calculated T_final to ensure that the change in the power supply cycle of adjacent measuring points does not exceed 50% of the baseline value, avoiding discontinuous data acquisition caused by parameter abrupt changes. Finally, a stable final power supply cycle T_final that adapts to the site characteristics is obtained, completing the coordinated adaptive adjustment of the sampling interval and power supply cycle.

[0026] Next, the steps of adaptively adjusting the sampling interval and power supply cycle in this embodiment will be further described.

[0027] The first step is to simultaneously calculate the real part of the apparent resistivity ρ_real corresponding to the measurement point based on the obtained local distortion amplitude η and phase difference Δφ of the measurement point. The extraction of the real part of the apparent resistivity adopts a standardized process: the time-domain response signal of the electric field collected at the measurement point is subjected to a windowed Fast Fourier Transform (FFT). The Hanning window is selected as the window function to suppress spectral leakage. The number of FFT points is set to 1024 points, and the frequency resolution is not less than 1Hz. The frequency component that is consistent with the fundamental frequency of the power supply is extracted, and the real and imaginary parts of the component are calculated. The real part of the apparent resistivity ρ_real of the measurement point is calculated based on the real part. The influence of harmonic interference and reactive power components on the calculation results is eliminated to ensure that ρ_real can truly reflect the inherent resistivity characteristics of the formation. Based on the three core parameters mentioned above, the local distortion index D_index=(η×Δφ) / ρ_real for each measuring point is defined and calculated point by point. This index integrates the amplitude distortion of the current field, the phase lag characteristics, and the inherent resistivity of the strata, and can comprehensively quantify the development degree and complexity of underground fractures. Compared with a single amplitude or phase index, it has higher sensitivity to identify weak electric field distortions caused by small hidden fractures, and can effectively identify centimeter- to meter-level small hidden fractures that traditional methods cannot capture. The second step is to pre-set the distortion index threshold D_threshold. This threshold is set based on the statistical results of D_index in the background area of ​​the exploration region, taking 5 times the average D_index of the background area as the benchmark threshold. In this embodiment, for the detection needs of small hidden fractures in conventional building engineering exploration, D_threshold is set to 0.8 by default; for deep foundation pit engineering, shield tunneling shaft exploration, and high-rise building foundation exploration scenarios with extremely high requirements for fracture detection accuracy, D_threshold is lowered to 0.5 to further improve the ability to identify weak distortion signals. The D_index and D_threshold of each measuring point within the entire exploration area are compared one by one. If D_index > D_threshold, it is determined that there is a small hidden fracture in the area where the measuring point is located. For this measuring point and all measuring points within the surrounding 3×3 grid, a high-frequency scanning mode is automatically triggered to achieve targeted and dense detection of suspected areas of small hidden fractures. At the same time, the area that triggers the high-frequency scanning mode is spatially marked and spatially superimposed with the edge of the fracture development zone to form a key detection area, avoiding the problems of reduced detection efficiency and excessive data redundancy caused by indiscriminate full-area densification. In the third step, after entering the high-frequency scanning mode, the sampling frequency is adaptively adjusted based on the set reference sampling frequency f_base. The adjustment formula is f_sampling=f_base×(D_index / D_threshold), where the reference sampling frequency f_base is set to 20Hz by default, corresponding to the default sampling interval of 50ms.The adjusted sampling frequency is directly proportional to D_index. A larger D_index indicates a more complex fracture and weaker signal distortion, resulting in a higher sampling frequency and greater time-domain sampling density. This ensures complete capture of transient resistivity changes caused by minute, latent fractures, addressing signal omission at its source. Simultaneously, safety upper and lower limits are set for the sampling frequency: the highest sampling frequency does not exceed 200Hz to avoid high-frequency noise intrusion; the lowest sampling frequency is not lower than the reference sampling frequency to ensure basic detection accuracy. The power supply cycle is also adaptively adjusted using the formula T_supply = T_base × (D_index / D_threshold). By extending the power supply cycle, the excitation energy of weak signals is increased, the signal accumulation time is extended, and the signal-to-noise ratio of the acquired data is improved. For strong distortion measurement points where D_index exceeds the threshold by more than 3 times, a multi-cycle superposition acquisition mode is simultaneously activated. The number of superpositions is positively correlated with the ratio of D_index / D_threshold, and the maximum number of superpositions does not exceed 32 times. This further improves the ability to identify weak signals and ensures that resistivity change signals caused by tiny hidden fractures are completely and accurately acquired.

[0028] In addition, in this embodiment, the high-density resistivity data acquisition and disease location based on the adjusted sampling parameters include the following steps.

[0029] First, based on the adaptively adjusted core sampling parameters such as sampling interval, power supply cycle, sampling frequency, and number of stacking operations, data acquisition was conducted in a three-tiered classification system for the exploration area: background area, fracture zone, and micro-fracture zone. For the background area where the high-frequency scanning mode was not triggered, conventional acquisition was performed using baseline sampling parameters to achieve basic data coverage of the entire area. For the edge areas of fracture development zones, intensified acquisition was conducted using adjusted sampling intervals and power supply cycles to improve the detection accuracy of fracture zone boundaries. For suspected micro-concealed fracture areas that triggered the high-frequency scanning mode, refined and repeated acquisition was performed using optimized parameters under the high-frequency scanning mode, focusing on capturing weak abrupt changes. During the acquisition process, for each key detection area, three power supply modes were used sequentially: Wenner quadrupole, dipole-dipole, and differential tripolar. Each power supply mode underwent at least eight rounds of cyclic measurement. Each round of measurement used a rolling electrode arrangement, with the rolling step size of the electrode arrangement matching the adjusted sampling interval to ensure that the measurement data of adjacent arrangements overlapped by at least three electrodes, improving the spatial continuity of the data. After each round of data acquisition, the signal-to-noise ratio (SNR) of the data in that round is checked in real time. When the SNR is lower than 30 dB, the number of stacking operations is automatically increased until the SNR meets the requirements. This results in a high-resolution resistivity response dataset covering the entire survey area and categorized by region. The dataset contains comprehensive information for each measuring point, including its coordinates, electrode arrangement, power supply parameters, resistivity values ​​measured multiple times, acquisition time, and SNR. Secondly, the acquired resistivity response dataset undergoes a point-by-point full-process data quality check. First, for each measuring point, the arithmetic mean rho_avg of the repeatedly measured resistivity values ​​is calculated using the formula rho_avg=(1 / n)×Σ(rho_i), where i=1,2,...,n, and n is the number of repeated measurements at that measuring point, n≥8. Based on the obtained rho_avg, the standard deviation σ_rho=sqrt((1 / n)×Σ(rho_i-rho_avg)²) of the resistivity measurements at that measuring point is calculated. This standard deviation quantifies the stability and dispersion of the measurement data at that measuring point. A relative standard deviation threshold τ is pre-set, i.e., τ = (σ_rho / ρ_avg) × 100%, where σ_rho is the standard deviation of resistivity measurements and ρ_avg is the average value of resistivity measurements. In this embodiment, τ is set to 8% for routine survey areas and 5% for key detection areas. Measurement data exceeding the standard deviation threshold are deemed invalid and removed. For the removed measurement points, Kriging interpolation is performed using valid data from adjacent measurement points to complete the dataset and ensure spatial integrity. Simultaneously, a third-order trend surface analysis is performed on the dataset to remove global abnormal jump values ​​caused by poor electrode contact or interference from underground metal pipelines, ultimately obtaining a high-quality, effective resistivity dataset.Furthermore, based on the filtered effective resistivity dataset, a sliding window method with a 5×5 measuring point grid was used to extract resistivity abrupt change points, with a sliding step size of one measuring point. The coefficient of variation and gradient change of resistivity values ​​within the window were calculated. When the resistivity gradient change within the window exceeded three times the maximum gradient value of the background area, the measuring point at the center of the window was marked as a resistivity abrupt change point, i.e., the location of potential disease. DBSCAN spatial clustering analysis was performed on all marked abrupt change points to remove isolated discrete abrupt change points, retaining the continuously distributed set of abrupt change points, thus delineating the core area of ​​potential disease. For the identified abrupt change region, the least squares method is used to fit the spatial distribution curve of resistivity and the electric field distribution characteristics of the region. The fitting model adopts a second-order bivariate polynomial fitting model, and the fitting formula is ρ(x,y)=a0+a1x+a2y+a3x²+a4xy+a5y², where a0-a5 are the fitting coefficients, x and y are the planar coordinates of the measuring points, and ρ is the apparent resistivity value of the measuring points. The fitting coefficients are solved by the least squares method to ensure that the sum of squared residuals between the fitted values ​​and the measured values ​​is minimized, and the goodness of fit R² is not less than 0.95. Based on the spatial distribution surface of resistivity obtained by fitting, the first and second derivatives of the surface are solved. The extreme points of the first derivative and the zero-crossing points of the second derivative are used as the boundary inflection points of the disease. By connecting all the boundary inflection points, the planar boundary and distribution range of the disease are accurately delineated, and the precise location and boundary delineation of the underground disease are finally completed, providing accurate geological disease data support for building engineering survey, design and construction.

[0030] In this case, the high-density resistivity data acquisition and lesion localization based on the adjusted sampling parameters specifically includes the following steps: The first step involves constructing a three-dimensional geological entity model covering the entire exploration area, based on the spatial layout coordinates of the electrode array, preliminary geological stratification data of the exploration area, and the selected effective resistivity dataset. The model extends 20m beyond each end of the exploration line laterally and covers 1.5 times the maximum designed exploration depth vertically, ensuring that boundary effects do not interfere with the electric field simulation results in the core exploration area. A tetrahedral unstructured mesh is used to partition the model, with the minimum mesh size set to half the lateral electrode spacing and the maximum mesh size not exceeding 1 / 10 of the minimum exploration depth. Mesh refinement is applied in fracture development zones and areas with suspected micro-hidden faults to ensure simulation accuracy. Based on Maxwell's governing equations for a stable current field, the finite element method is used to simulate the underground electric field distribution, with strict boundary conditions: the top surface of the model is the air-stratum interface, and the normal component of the current density is set to 0; the bottom and sides of the model are set as infinity boundaries, with a boundary potential of 0. The measured apparent resistivity data is used as the electrical parameter input model of the formation unit to simulate the distribution and variation of the underground electric field under different power supply electrodes, resulting in a simulated electric field response dataset. The simulation results are compared and corrected with the measured current field data, and the formation electrical parameters are iteratively optimized until the relative error between the simulated and measured values ​​is less than 10%, ultimately obtaining a high-precision three-dimensional apparent resistivity imaging model. The second step involves calculating the slope of the apparent resistivity change curve and accurately marking suspicious points. Based on the constructed three-dimensional apparent resistivity imaging model, the apparent resistivity time series of each measuring point at different acquisition times and under different power supply modes is extracted. Using the adaptively adjusted sampling interval as the time increment Δt, the apparent resistivity change Δρ between two adjacent sampling times is calculated. Δρ is the difference between the measured apparent resistivity at the later time and the earlier time. The slope of the apparent resistivity change curve, K = Δρ / Δt, is further calculated. This slope can accurately quantify the rate of change of apparent resistivity per unit time and sensitively capture transient resistivity abrupt changes caused by minute hidden fractures. A slope threshold K_threshold is preset. This threshold is set based on the statistical results of K values ​​in the background area of ​​the exploration region, taking three times the maximum absolute value of K values ​​in the background area as the benchmark value. In this embodiment, K_threshold is set to 5 Ω·m / s in a typical Quaternary loose strata exploration scenario; K_threshold is increased to 10 Ω·m / s in high-resistivity bedrock outcrops; and K_threshold is decreased to 3 Ω·m / s in low-resistivity water-saturated loose strata. The absolute slope value |K| of each measuring point in the entire exploration area is compared with K_threshold one by one. If |K|>K_threshold, the measuring point is marked as a suspected defect point, and the three-dimensional spatial coordinates, slope value, apparent resistivity time series data, corresponding acquisition parameters, and power supply mode of the suspected point are recorded simultaneously. The third step is to construct a spatial correlation matrix and verify the authenticity of the suspected points.For all marked suspicious points, a three-dimensional spatial correlation matrix is ​​constructed. The matrix dimension is the number of suspicious points × the number of suspicious points, and the matrix elements are the spatial correlation coefficients between two suspicious points. The correlation coefficients are calculated using a dual-weighting method. The first weight is the Pearson correlation coefficient of the apparent resistivity time series of the two suspicious points, quantifying the temporal consistency of the abnormal signal. The second weight is a Gaussian decay weight based on the spatial distance between the two points, quantifying the spatial correlation of the abnormal signal. The closer the spatial distance, the higher the weight. A correlation coefficient threshold of 0.7 is preset. If the correlation coefficients of a suspicious point with three or more adjacent suspicious points are all greater than the threshold, it indicates that the abnormal signal has continuous spatial consistency and is determined to be a real disease anomaly point. If the correlation coefficient is lower than the threshold and there are no adjacent related anomalies, it is determined to be random noise or a false anomaly caused by poor electrode contact and is discarded. Three-dimensional spatial DBSCAN clustering analysis was performed on all real anomalies. Combined with the electrical distribution characteristics of the three-dimensional apparent resistivity imaging model, the three-dimensional spatial distribution range, burial depth, scale and orientation of the disease were accurately delineated. This completed the three-dimensional accurate positioning of underground disease, which solved the problems of depth deviation and boundary ambiguity in traditional two-dimensional planar positioning. At the same time, it significantly reduced the probability of false anomaly misjudgment.

[0031] In another embodiment, such as Figure 2 As shown, obtaining current field data of a multi-pole electrode array deployed on the surface of the construction area under different power supply modes includes the following steps: S201: Select bipolar, tripolar, and quadrupole power supply modes for multiple measurements. After completing the multi-pole electrode array layout, electrode grounding treatment, and grounding resistance test in the construction area, a multi-power supply mode system adapted to different detection requirements is pre-constructed, covering three mainstream high-density electrical resistivity tomography (EDT) power supply modes: bipolar-dipole power supply mode, tripolar (Wener-Schlumberger) power supply mode, and Wenner quadrupole power supply mode. The reference operating parameters for each mode are set separately. In the bipolar-dipole power supply mode, the spacing between power supply electrodes A and B is set to 5-10 times the spacing between measuring electrodes M and N. The electrodes are arranged in a rolling pattern along the survey line, which is suitable for large-area rapid scanning and features large detection depth and high construction efficiency. In the tripolar power supply mode, the infinity power supply electrode B is arranged in the direction perpendicular to the survey line, at a distance of not less than 5 times the maximum designed exploration depth from the endpoint of the survey line, ensuring that the infinity boundary condition is met. This mode is suitable for medium-deep medium-resolution detection. In the Wenner quadrupole power supply mode, the four electrodes A, M, N, and B are arranged at equal intervals along the survey line, satisfying AM=MN=NB. This mode features high signal-to-noise ratio and high lateral resolution, and is suitable for high-resolution fine detection. The reference electrical parameters for the three modes are uniformly configured as follows: power supply voltage is linearly adjustable from 12V to 48V, maximum output current is not less than 500mA, reference sampling frequency is 20Hz, reference power supply cycle is 200ms, the default number of superpositions per round of measurement is 4, the qualified threshold for grounding resistance is set to 10Ω, and when the grounding resistance of a single electrode exceeds the threshold, conductive mud is backfilled to reduce the grounding resistance until the requirements are met.

[0032] S202: For each power supply mode, acquire the electric field response data and calculate its signal-to-noise ratio (SNR) = P_signal / P_noise, where P_signal is the effective signal power and P_noise is the noise power. Following the priority order of "bipolar-dipole mode → tripolar mode → Wenner quadrupole mode," perform full-permutation cyclic measurements on the electrode array sequentially. Each power supply mode completes three rounds of complete repeated measurements, with each round covering all effective electrode pairs within the array, ensuring no measurement points are missed. After each measurement of an electrode pair is completed, calculate the SNR of that set of measurement data in real time using the formula SNR = P_signal / P_noise, where P_signal is the effective signal power, obtained by extracting the signal component with the same frequency as the power supply fundamental frequency through bandpass filtering and calculating the mean square power of this component; P_noise is the noise power, obtained by calculating the average power of the signal across the entire frequency band outside the fundamental frequency. Simultaneously, convert the SNR to decibels using the formula SNR_dB = 10 × lg(SNR) for threshold determination. For all measurement data of each power supply mode, the average signal-to-noise ratio, minimum signal-to-noise ratio, and data pass rate are statistically analyzed and used as the comprehensive signal-to-noise ratio evaluation index for that power supply mode.

[0033] S203: Switch to the high-resolution power supply mode when SNR < SNR_threshold. The signal-to-noise ratio threshold SNR_threshold is preset. In this embodiment, it is set to 30 dB in the conventional exploration scenario. In high electromagnetic interference scenarios such as densely built urban areas, under high-voltage lines, and around subways, the threshold is lowered to 20 dB. For the currently used power supply mode, if its comprehensive average SNR < SNR_threshold, immediately automatically stop the current mode measurement, switch to the Wenner quadrupole power supply mode with high resolution and high anti-interference ability, and restart the full-array measurement to improve the data signal-to-noise ratio. If the comprehensive average SNR of the measured data is still lower than SNR_threshold after 3 consecutive complete scan cycles of switching to the Wenner quadrupole mode, enable the multi-angle electrode layout optimization scheme: on the basis of the original 0° main survey line, add two parallel survey lines in the directions of 45° and 90°. The spacing between the new survey lines and the electrode spacing are exactly the same as those of the original main survey line. The electrode layout positions form a cross-grid structure with the original array to achieve multi-angle full-coverage measurement of the exploration area. At the same time, perform secondary grounding optimization on the electrodes at the intersection points, backfill with low-resistance conductive powder, and reduce the grounding resistance to below 5 Ω. Through multi-angle cross measurement, effectively suppress the false anomalies caused by environmental electromagnetic interference and surface inhomogeneities, improve the signal-to-noise ratio and spatial resolution of the current field data, and finally obtain multi-mode, multi-angle, and high-quality original current field data, providing a reliable data basis for subsequent distortion feature identification and parameter adjustment.

[0034] At this time, the adaptive adjustment of the sampling interval and the power supply period includes the following steps: First, convert the local distortion amplitude η and the phase difference Δφ into a perturbation index F = η + Δt_phase / T_base, where T_base is the reference power supply period, and Δt_phase = Δφ / (2πf), and f is the power supply frequency. Define and calculate the perturbation index F for each measurement point in the entire exploration area point by point based on the identified local distortion amplitude η, phase difference Δφ of the measurement point, and the preset reference power supply period T_base. The calculation formula is F = η + Δt_phase / T_base. Among them, η is the dimensionless local distortion amplitude, quantifying the amplitude distortion degree of the current field; Δt_phase is the phase lag time, in milliseconds (ms), quantifying the phase distortion degree of the current field; T_base is the reference power supply period, in milliseconds (ms), and the default value in this embodiment is 200 ms; the finally obtained perturbation index F is a dimensionless index, which can comprehensively quantify the comprehensive perturbation degree of the underground medium heterogeneity, fracture development, and buried faults on the current field, and at the same time integrates the three dimensions of amplitude distortion, phase lag, and reference power supply parameters. Compared with a single amplitude or phase index, it can more comprehensively and accurately reflect the strength of formation anomalies and provide a quantitative basis for the adjustment intensity of sampling parameters.

[0035] Next, the regulation intensity is determined by judging the magnitude of the disturbance index F. F > F_threshold indicates that sampling should be significantly enhanced. The disturbance index threshold F_threshold is preset. This threshold is set based on the statistical results of the F values in the background area of the exploration area. Four times the average value of the F values of all measuring points in the background area is taken as the reference threshold to ensure the statistical reliability of the threshold. In this embodiment, in the scenario of conventional building engineering exploration, F_threshold is set to 0.05. For exploration scenarios with high sensitivity requirements for minor anomalies, such as deep foundation pits, high-rise building foundations, and shield launching shafts, F_threshold is lowered to 0.03 to further improve the ability to identify weak disturbances. The F value of each measuring point in the entire exploration area is compared with F_threshold one by one to complete the three-level classification determination of the regulation intensity of sampling parameters: when F ≤ F_threshold, it is determined as the background area without obvious formation disturbance, and the reference sampling parameters are used without additional adjustment; when F_threshold < F ≤ 3×F_threshold, it is determined as the medium disturbance area, corresponding to medium-strength parameter adjustment; when F > 3×F_threshold, it is determined as the strong disturbance area, triggering the maximum-strength parameter regulation, and at the same time marked as the key encrypted detection area.

[0036] Then, the sampling interval Δt = Δt_initial × e^(k_F × F) is adjusted using an exponential decay function, where k_F is the adjustment coefficient. For measuring points identified as having ground disturbances, the sampling interval and power supply cycle are adaptively adjusted using an exponential decay function to achieve precise targeted control where "the stronger the disturbance, the higher the sampling density, and the more suitable the power supply cycle," while avoiding parameter abrupt changes caused by linear adjustment. The adjustment formula for the sampling interval is Δt = Δt_initial × e^(k_F × F), where Δt_initial is the initial sampling interval, i.e., the set default sampling interval Δt_base, which is 50ms by default in this embodiment; k_F is the sampling interval adjustment coefficient, ranging from -0.8 to -0.2. The negative sign indicates that the larger the F value, the smaller the adjusted sampling interval Δt. In this embodiment, k_F is set to -0.5 by default in normal scenarios, and lowered to -0.8 in areas of strong disturbance to improve adjustment sensitivity. For the power supply cycle, the adjustment formula is T_supply = T_base × e^(k_T × F), where k_T is the power supply cycle adjustment coefficient, ranging from 0.2 to 0.8. A positive sign indicates that the larger the F value, the longer the adjusted power supply cycle T_supply. In this embodiment, k_T is set to 0.5 by default in normal scenarios, and is increased to 0.8 in areas with strong disturbances. Simultaneously, safety boundaries for parameter adjustment are set: the minimum sampling interval is not less than 10ms, and the maximum value does not exceed 4 times the initial value; the minimum power supply cycle is not less than 0.5 times the base value, and the maximum value does not exceed 5 times the base value, to avoid overloading the instrument acquisition system, high-frequency noise contamination, or missed effective signals due to excessive parameter adjustment. For the adjustment results of adjacent measuring points, a 3-point moving average method is used for smoothing to ensure continuous parameter changes without abrupt changes between adjacent measuring points, improving the stability and data continuity of the data acquisition system.

[0037] Furthermore, in this embodiment, the high-density resistivity data acquisition and lesion localization based on the adjusted sampling parameters include the following steps: The first step is to standardize and segment the acquired high-resolution resistivity time series data based on the adaptively adjusted sampling interval Δt. First, the length n_block of each data block is determined using the formula n_block = round(Δt / Δt_resolution), where Δt_resolution is the minimum sampling time resolution of the acquisition instrument, set to 1 ms in this embodiment; round is a rounding function to ensure that the length of each data block is an integer and perfectly matches the adjusted sampling interval. For example, when the adjusted sampling interval Δt is 50 ms, n_block = round(50 / 1) = 50, meaning each data block contains 50 consecutive sampling points. The segmentation process uses a 50% overlap sliding window segmentation method, meaning that there is 50% overlap in sampling points between adjacent data blocks. This avoids the truncation of transient abrupt signals caused by data segmentation, ensuring that each resistivity abrupt feature is completely contained within at least one data block, thus eliminating the problem of missed abrupt signal detection at the data processing level.

[0038] The second step involves preprocessing the resistivity time-domain data after each block by detrending. Least squares fitting is used to fit and remove linear baseline drift, eliminating systematic biases caused by electrode polarization and temperature drift. Then, a windowed Fast Fourier Transform (FFT) is applied to the preprocessed time-domain data for spectral analysis. A Hamming window is used to effectively suppress spectral leakage. The number of FFT points is consistent with the data block length n_block, and the frequency resolution is 1 / Δt to ensure the accuracy of the spectral analysis matches the sampling interval. The transformed spectral data is then divided into frequency bands and effective components are extracted. Low-frequency effective components from 0.1Hz to 10Hz are retained, as this band represents the main response frequency band of underground strata electrical anomalies. Simultaneously, high-frequency noise components above 10Hz, including 50Hz power frequency interference and its higher harmonics, and random electromagnetic noise, are removed, resulting in the denoised amplitude and phase spectra. The signal-to-noise ratio (SNR) of the spectrum data for each data block is calculated synchronously. For data blocks with an SNR lower than 10dB, wavelet soft thresholding is used for secondary noise reduction to ensure the quality and reliability of the spectrum data.

[0039] The third step involves extracting the peak amplitude ρ_max corresponding to each data block based on the denoised amplitude spectrum data. Simultaneously, the average value ρ_mean of the apparent resistivity time-series data within that data block, and the standard deviation σ_rho of the apparent resistivity, are calculated. Based on these parameters, the kurtosis index Sk = (ρ_max - ρ_mean) / σ_rho, which characterizes the degree of signal abrupt change, is calculated. This kurtosis index accurately quantifies the suddenness and anomaly of the apparent resistivity signal. The larger the Sk value, the more significant the abrupt change in the signal, and the higher the probability of a hidden underground fault. A kurtosis threshold Sk_threshold is preset; in this embodiment, it is set to 3 based on the normal distribution 3σ criterion. That is, when Sk > Sk_threshold, a significant resistivity abrupt change is determined to exist within the data block. Simultaneously, multi-band synchronization analysis was performed on the spectral data of this data block. The effective low-frequency band was divided into three sub-bands: 0.1-1Hz, 1-5Hz, and 5-10Hz. If synchronous amplitude peak oscillations occurred in all three sub-bands, i.e., the peaks appeared at the same time and their trends were highly consistent, then the abrupt change was determined to be an effective geological anomaly caused by a small, concealed underground fault. If a peak appeared only in a single sub-band without multi-band synchronous oscillation characteristics, it was determined to be random noise interference and was removed. Finally, the accurate identification of small, concealed faults was achieved. Combined with the spatial coordinate information of the measuring points, the plane location, strike determination, and distribution range of the fault were completed. This solved the core problems of traditional methods, such as low recognition rate, easy omission, and easy misjudgment of weak abrupt change signals caused by small, concealed faults.

[0040] Furthermore, in this embodiment, acquiring the current field data of the multi-pole electrode array deployed on the surface of the construction area under different power supply modes includes the following steps: The first step, after completing the surface clearing and coordinate layout of the construction area, is to accurately locate the multi-pole electrode array deployment points based on the preset exploration depth and lateral resolution requirements. The deployment points are laid out using a grid-like geodetic plane coordinate system. The spacing between deployment points in the lateral survey line direction is completely consistent with the designed electrode spacing Δx, and the spacing between deployment points in the longitudinal parallel survey line is equal to Δx. The plane coordinate layout error of all deployment points is controlled within ±2cm to ensure the absolute accuracy of the electrode spatial position. In this embodiment, the electrode is a pure copper cylindrical grounding electrode with a diameter of d=1.5cm and a total length of 30cm. The depth of the electrode is not less than 20cm. The bottom of the electrode is backfilled with conductive mud to ensure that the electrode forms a low-impedance good contact with the stratum. For each pair of electrodes that have been deployed (including power supply electrode pairs, measurement electrode pairs, and power supply-measurement cross electrode pairs), the straight-line distance L_ij between any electrode i and electrode j is calculated based on the plane coordinates of the two points. The calculation accuracy is accurate to 0.01m. At the same time, electrode pairs with L_ij≤2d are removed to avoid direct coupling interference between electrodes and ensure that all electrode pairs participating in the calculation are effective electrode pairs. Based on the theory of stable current field in a uniform half-space, the mutual impedance Z_ij between effective electrodes is calculated pair by pair. The calculation formula is Z_ij=ρ / (2π)×ln((L_ij+d) / (L_ij-d)), where ρ is the background apparent resistivity of the strata in the exploration area, which is obtained through small-scale pre-test calibration during the preliminary site reconnaissance. In this embodiment, the background apparent resistivity of the Quaternary loose strata is taken as 100Ω·m, and that of the intact bedrock strata is taken as 500Ω·m; d is the electrode radius, and the unit is the same as L_ij, which is meters (m). Through the pre-calculation of mutual impedance, the theoretical impedance benchmark value of each pair of electrodes is calibrated in advance, providing a standard basis for subsequent grounding resistance verification and data validity judgment, and at the same time completing the validity screening of the electrode array.

[0041] The second step involves completing the electrode layout and mutual impedance pre-calculation. For each pair of effective power supply electrodes, at least six rounds of constant current injection are performed. The injection current intensity remains constant in each round, controlled between 10mA and 200mA, and is adaptively adjusted according to the formation resistivity to ensure that the real-time fluctuation of the injection current does not exceed ±0.5%, meeting the constant current accuracy requirements of high-density electrical resistivity tomography (EDS) data acquisition. During each round of current injection, voltage data of all measurement electrode pairs within the electrode array are continuously acquired simultaneously. The acquisition frequency is consistent with the set reference sampling frequency, and the acquisition duration of a single round fully covers the preset reference power supply cycle. Simultaneously, auxiliary parameters such as the real-time value of the injection current, power supply duration, single-electrode grounding resistance, and ambient background noise level are recorded for each round. For multiple rounds of data acquisition for the same electrode pair, the Grubbs criterion is used to remove abnormal jump values ​​exceeding ±3 standard deviations. The remaining valid data are then arithmetically averaged to obtain the final voltage response data and current injection data for that electrode pair. This process eliminates the influence of random electromagnetic noise on the acquisition results, ensuring the stability and reliability of the data.

[0042] The third step involves constructing an electrode array admittance matrix Y covering the entire exploration area, based on the obtained full electrode pair voltage response data V and injection current data I. The admittance matrix has an N×N dimension, where N is the total number of electrodes in the electrode array. Rows correspond to the measurement electrode numbers, and columns correspond to the power supply electrode numbers. Matrix element Y_ij represents the admittance value when the j-th electrode is powered and the i-th electrode is measured. The calculation formula is Y_ij = V_ij / I_j, where V_ij is the steady-state voltage response value measured by the i-th electrode when the j-th electrode injects a constant current I_j, and I_j is the actual injection current intensity of the j-th electrode. For missing elements in the matrix without valid measurement data, Kriging interpolation of adjacent valid elements is used to fill in the gaps, ensuring the integrity and continuity of the admittance matrix. The constructed admittance matrix Y is subjected to standardized singular value decomposition, with the decomposition formula Y = U × Σ × V^H, where U is an N × N dimensional left singular matrix, Σ is an N × N dimensional diagonal singular value matrix, the elements on the diagonal are the singular values ​​of the admittance matrix, arranged in descending order, and V^H is the conjugate transpose of the N × N dimensional right singular matrix. The principal component corresponding to the component with the largest singular value is extracted, i.e., the first principal component. This principal component can represent more than 90% of the effective information in the admittance matrix and fully reflect the overall spatial distribution characteristics of the formation current field. At the same time, the noise components corresponding to higher-order singular values ​​are removed, realizing dimensionality reduction and noise reduction of the original data and purification of effective information.

[0043] The fourth step involves preliminary identification of abrupt changes in the current field based on the first principal component obtained from SVD decomposition. First, a 3×3 sliding window is used to traverse the deployment points of the entire electrode array, calculating the spatial gradient rate of change and coefficient of variation of the first principal component within the window. A gradient amplitude threshold is pre-set, taking three times the maximum gradient amplitude of the first principal component in the background area of ​​the survey region as the baseline threshold. When the gradient amplitude within the window exceeds the threshold, it is determined that the region corresponding to that window has abrupt changes in the current field. The region is then spatially marked, and the marking results are simultaneously synchronized to the fracture development zone edge identification process, providing preliminary target points for subsequent distortion feature identification and significantly improving the efficiency and accuracy of feature identification. Simultaneously, the validity of the original current field data is judged by the reconstruction error of the first principal component. When the reconstruction error of a single measurement point exceeds 10%, it is determined that the collected data in that region is abnormal, and current injection and data acquisition are immediately restarted to ensure the high quality and high reliability of the final acquired current field data, providing accurate basic data support for subsequent adaptive adjustment of sampling parameters and precise location of the disease.

[0044] At this point, based on the local distortion amplitude and phase lag characteristics, the adaptive adjustment of the sampling interval and power supply cycle includes: The first step is to construct a hierarchical logic threshold judgment mechanism based on the local distortion amplitude η of the measuring points in the entire exploration area calculated point by point. The core trigger threshold is set to η≥3. This threshold is based on the set distortion threshold η_threshold=2.5. If it exceeds the basic threshold by more than 20%, it means that there is a strong distortion in the current field at the measuring point, which corresponds to a high degree of formation fracture development and a very high probability of the presence of small hidden fractures, thus meeting the triggering conditions of the fine adjustment process. The logic threshold judgment adopts a dual verification mode of "point-by-point traversal + spatial clustering" to avoid false triggering caused by random noise: First, the η value of all measuring points in the entire exploration area is compared one by one. When the η of a single measuring point is ≥3, it is marked as a point to be triggered. Then, DBSCAN spatial clustering analysis is performed on all points to be triggered. When the number of points to be triggered in a cluster is ≥3, the fine adjustment process is automatically triggered immediately for all measuring points in the area covered by the cluster and within a 5m radius around it. For isolated single points to be triggered, they are judged as false anomalies caused by random noise or poor electrode contact, and the fine adjustment process is not triggered to avoid invalid parameter adjustments affecting the overall detection efficiency.

[0045] The second step involves adjusting the sampling interval using a phased strategy of "gradual compression and step-by-step verification" to avoid instability and poor data continuity caused by sudden changes in the sampling interval. This is specifically implemented in two phases: The first phase is the initial compression phase, which uses the set default sampling interval Δt_base as a benchmark and initially adjusts the sampling interval to Δt_1 = Δt_base × 0.8. In this embodiment, Δt_base defaults to 50ms, corresponding to an initially adjusted Δt_1 = 40ms. One round of pre-acquisition is performed using the adjusted Δt_1 to verify the signal-to-noise ratio and distortion feature recognition of the pre-acquisition data in real time. When the signal-to-noise ratio of the pre-acquisition data is ≥30dB and the spatial gradient change characteristics of the current field can be clearly captured, the second phase of in-depth refinement adjustment begins. If the signal-to-noise ratio does not meet the requirements, Δt_1 is kept unchanged, and the number of acquisition and superposition times is increased synchronously until the data quality meets the verification standard. The second stage is the in-depth refinement stage. Building upon the first stage, the sampling interval is gradually compressed in steps of 10% of Δt_base. After each compression, a round of pre-acquisition and data quality verification is performed simultaneously. Only after successful verification is the next compression performed, until the sampling interval is adjusted to the final refined sampling interval Δt_final = Δt_base × 0.5. In this embodiment, Δt_final = 25ms, achieving a sampling density that is twice that of the baseline value. This significantly improves the temporal resolution for weak, abrupt signals, preventing the omission of resistivity abrupt signals caused by minute latent breaks. Simultaneously, a minimum safe threshold of 10ms is set for the sampling interval to ensure that the adjusted sampling interval remains within the instrument's safe acquisition range, avoiding excessive amplification of high-frequency noise due to excessively short sampling intervals.

[0046] The third step involves adjusting the sampling interval in stages while designing a dedicated phase delay compensation algorithm for the phase difference Δφ to eliminate signal acquisition distortion and phase deviation caused by phase lag. First, the effective starting point for phase compensation is determined using the formula Δφ_threshold=max(Δφ,0). This means compensation is only applied to positive phase lag (the measured signal phase lags behind the power supply current reference phase), while no additional compensation is applied to negative phase lead to avoid reverse phase distortion caused by overcompensation. In this embodiment, the unit of Δφ is uniformly set to radians (rad) to ensure consistency with the unit system used in subsequent frequency calculations. Next, the instrument's operating frequency step Δf is calibrated. Δf is set in conjunction with the instrument's power supply frequency range and sampling frequency. In this embodiment, the instrument's reference operating frequency is 5Hz, the frequency adjustment range is 0.5Hz-20Hz, and the operating frequency step Δf is set to 0.1Hz to ensure the precision of frequency adjustment meets the phase compensation requirements. Based on the above parameters, the time difference δ corresponding to phase compensation is calculated using the formula δ=Δφ_threshold / (2π×Δf), where 2π is the conversion coefficient between angular frequency and frequency, and δ is in seconds (s). After conversion to milliseconds (ms), it is used for compensation adjustment of the final power supply cycle. The fourth step is the fine-tuning and smoothing optimization of the final power supply cycle. Based on the phase compensation time difference δ calculated above, the reference power supply cycle is finely compensated and adjusted. The final power supply cycle T_final is calculated using the formula T_final=T_base+δ, where T_base is the set reference power supply cycle, which is defaulted to 200ms in this embodiment. This compensation adjustment achieves precise matching between the power supply cycle and the phase lag time, ensuring the acquisition of a complete steady-state electric field response signal and eliminating signal truncation errors caused by phase lag. Simultaneously, a double safety boundary is set for the adjusted final power supply cycle: the minimum power supply cycle is not less than 100ms, and the maximum power supply cycle does not exceed 1000ms, avoiding incomplete signal acquisition due to an excessively short power supply cycle or low overall detection efficiency due to an excessively long power supply cycle. For the final power supply cycle adjustment results of adjacent measuring points, a 5-point moving average method is used for smoothing to ensure that the change in power supply cycle between adjacent measuring points does not exceed 30% of the baseline value, thus avoiding instability of the acquisition system caused by frequent parameter jumps. After completing the full-process fine adjustment of sampling interval and power supply cycle, the adjusted parameters are synchronized to subsequent data acquisition processes to carry out targeted fine detection, significantly improving the ability to capture resistivity abrupt changes caused by minute latent fractures, and completely solving the signal omission problem of traditional fixed parameter acquisition mode.

[0047] In addition, such as Figure 3 As shown, the present invention also provides a building engineering survey underground defect detection system 300 based on the high-density resistivity method, comprising: Data acquisition module 301 acquires current field data of multi-pole electrode arrays deployed on the surface of the construction area under different power supply modes; The identification module 302 identifies the local distortion amplitude and phase lag characteristics at the edge of the fracture development zone based on the current field data. The adjustment module 303 adaptively adjusts the sampling interval and power supply cycle based on the local distortion amplitude and phase lag characteristics. The positioning module 304 performs high-density resistivity data acquisition and disease location based on the adjusted sampling parameters.

[0048] The functions of each module of the underground lesion detection system 300 for building engineering exploration based on the high-density resistivity method of the present invention have been described above with reference to the method embodiment, and will not be repeated here.

[0049] The above description is the preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this invention, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for detecting underground defects in building engineering based on high-density resistivity method, characterized in that, include: Acquire current field data of multi-pole electrode arrays deployed on the surface of the construction area under different power supply modes; Based on the current field data, the local distortion amplitude and phase hysteresis characteristics at the edge of the fracture development zone are identified; Based on the local distortion amplitude and phase lag characteristics, the sampling interval and power supply cycle are adaptively adjusted. High-density resistivity data acquisition and lesion location were performed based on the adjusted sampling parameters.

2. The method for detecting underground defects in building engineering based on high-density resistivity method according to claim 1, characterized in that, The identification of local distortion amplitude and phase hysteresis characteristics at the edge of the fracture development zone based on the current field data includes: Obtain the current density difference ΔJ between each electrode pair; Calculate the rate of change of current density gradient γ = ΔJ / (Δx × Δy), where Δx is the transverse electrode spacing and Δy is the longitudinal electrode spacing; Based on the above γ, the local distortion amplitude η = |γ| / γ_mean is calculated, where γ_mean is the rate of change of the average current density gradient in the normal formation. If η > the preset threshold η_threshold, it is determined to be the edge of the fracture development zone.

3. The method for detecting underground defects in building engineering based on high-density resistivity method according to claim 2, characterized in that, The adaptive adjustment of the sampling interval and power supply cycle based on the local distortion amplitude and phase lag characteristics includes: The sampling interval Δt = Δt_base × (η_avg / η) is adjusted according to the distortion amplitude η at the edge of the fracture development zone, where Δt_base is the default sampling interval and η_avg is the average distortion amplitude of the background area. Obtain the phase difference Δφ between adjacent points; The power supply cycle T_adjust=T_base+(Δφ / φ_rate) is compensated by combining Δφ, where φ_rate is the preset phase change rate; The compensated power supply cycle is weighted and smoothed using the weighted formula T_final=α×T_adjust+(1-α)×T_base, where α is the weighting factor.

4. The method for detecting underground defects in building engineering based on high-density resistivity method according to claim 3, characterized in that, The adaptive adjustment of the sampling interval and power supply cycle includes: Define the local distortion index of the electrode pair as D_index=(η×Δφ) / ρ_real, where ρ_real is the real part of the apparent resistivity of the region; The complexity of the fracture can be determined by D_index. If D_index > D_threshold, it indicates the presence of a small, hidden fracture. If D_index is detected to exceed the threshold, enter high-frequency scanning mode, and the sampling frequency f_sampling=f_base×(D_index / D_threshold); The power supply cycle is set to T_supply=T_base×(D_index / D_threshold) to enhance sampling coverage.

5. The method for detecting underground defects in building engineering based on high-density resistivity method according to claim 3, characterized in that, The process of acquiring high-density resistivity data and locating diseased organisms based on the adjusted sampling parameters includes: Current injection is performed on multiple electrode arrangements under the adjusted sampling frequency and power supply cycle; The resistivity response dataset was collected and its standard deviation σ_rho=sqrt(1 / n×Σ(rho_i-rho_avg)^2), where rho_i is the single measurement value and rho_avg is the arithmetic mean of the resistivity values ​​of multiple repeated measurements at the measuring point. By setting a standard deviation threshold τ, mutation points are extracted from the data as potential disease locations; The electric field distribution characteristics of the mutation region are fitted by the least square method to determine the boundary of the disease body.

6. The method for detecting underground defects in building engineering based on high-density resistivity method according to claim 5, characterized in that, The high-density resistivity data acquisition and disease body positioning according to the adjusted sampling parameters include: Construct a three-dimensional apparent resistivity imaging model and simulate the electric field change based on finite element analysis; Calculate the slope K = Δρ / Δt of the apparent resistivity change curve between electrodes, where Δρ is the resistivity change per unit time and Δt is the time increment; Judge whether the absolute value of the slope |K| is greater than the preset critical value K_threshold, and if it is satisfied, mark it as a suspicious point; Evaluate the authenticity and position consistency of the suspicious points in combination with the spatial correlation matrix to avoid misjudgment.

7. The method for detecting underground defects in building engineering based on high-density resistivity method according to claim 6, characterized in that, The acquisition of the current field data of the multi-pole distance electrode array arranged on the surface of the construction area under different power supply modes includes: Select bipolar, tripolar and quadrupolar power supply modes for multiple measurements; For each power supply mode, obtain the electric field response data and calculate its signal-to-noise ratio SNR = P_signal / P_noise, where P_signal is the effective signal power and P_noise is the noise power; When SNR < SNR_threshold, switch to the high-resolution power supply mode; If the SNR standard is not satisfied for three consecutive scanning cycles, enable multi-angle electrode arrangement to improve the signal-to-noise ratio.

8. The method for detecting underground defects in building engineering based on high-density resistivity method according to claim 7, characterized in that, The further adaptive adjustment of the sampling interval and the power supply period includes: Convert the local distortion amplitude η and the phase difference Δφ into a perturbation index F = η + Δt_phase / T_base, where T_base is the reference power supply period, and Δt_phase = Δφ / (2πf), and f is the power supply frequency; Determine the regulation intensity by judging the magnitude of the perturbation index F. F > F_threshold indicates that sampling should be enhanced; Adjust the sampling interval Δt = Δt_initial×e^(k_F×F) using an exponential decay function, where k_F is the adjustment coefficient; Adjust the power supply period T_supply = T_base×e^(k_T×F) using the following formula, where k_T is another adjustment coefficient.

9. A method for detecting underground defects in building engineering based on high-density resistivity method according to claim 8, characterized in that, The high-density resistivity data acquisition and disease body positioning according to the adjusted sampling parameters include: Perform block data processing based on the new sampling interval, and the length of each data block n_block = round(Δt / Δt_resolution); Perform spectral analysis on the block data using the fast Fourier transform and extract the low-frequency part to reduce interference; Judge whether there are mutation characteristics based on the peak value of the amplitude spectrum. The kurtosis index Sk = (ρ_max - ρ_mean) / σ_rho characterizes the suddenness; If Sk > Sk_threshold and synchronous oscillations in multiple frequency bands occur, it is judged to be caused by a small hidden fracture.

10. A system for detecting underground defects in building engineering based on the high-density resistivity method, characterized in that, Include: A data acquisition module that acquires the current field data of the multi-pole distance electrode array arranged on the surface of the construction area under different power supply modes; An identification module that identifies the local distortion amplitude and phase lag characteristics at the edge of the fracture development zone based on the current field data; An adjustment module that adaptively adjusts the sampling interval and the power supply period based on the local distortion amplitude and phase lag characteristics; The positioning module performs high-density resistivity data acquisition and disease location based on the adjusted sampling parameters.

Images