A pile foundation settlement detection method and system applied to a gradual sedimentation type

Abstract

The application discloses a pile foundation settlement detection method and system applied to a gradually changing sediment, applied to the technical field of data processing, and comprises the following steps: obtaining initial density data, conductivity data and pile foundation penetration resistance data of the sediment; profile graphs are analyzed and identified to obtain gradually changing density data; the profile graphs are obtained by cutting a three-dimensional image of the sediment in a selected direction; the gradually changing density data is subjected to differential processing to obtain density distribution information; the penetration resistance data is subjected to mechanical characteristic enhancement processing to obtain density continuous distribution information; based on the density continuous distribution information, a bearing capacity reduction coefficient is determined; the conductivity data is subjected to characteristic extraction processing to obtain a sediment activity index; and the sediment activity index and the bearing capacity reduction coefficient are input into a pile foundation settlement calculation model to obtain pile foundation settlement data. The method provided by the application can realize accurate detection of pile foundation settlement.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, and in particular to a pile foundation settlement detection method and system applied to a gradually changing type of sediment. BACKGROUND

[0002] As an important part of the building substructure, the pile foundation can stably transmit the load to the surrounding soil, and has advantages such as high bearing capacity, high seismic capacity, and low settlement, and is widely used in highway, railway and building construction in China as a main supporting structure.

[0003] In the prior art, the pile foundation settlement calculation method is to divide the foundation into several thin layers according to the theory, calculate the compression amount of each layer under the additional stress, and then sum up to obtain the total settlement. However, this method relies on parameters obtained from indoor fixed tests, and there are differences between these parameters and the actual parameters of the sediment pit, which leads to a large deviation between the settlement calculation result and the actual engineering settlement, thereby seriously threatening the safety and stability and service life of the engineering structure. SUMMARY

[0004] The present application provides a pile foundation settlement detection method and system applied to a gradually changing type of sediment, to solve the technical problem of a large deviation between the settlement calculation result and the actual engineering settlement in the prior art, to realize accurate detection of the pile foundation settlement, and to ensure the safety and stability and service life of the engineering structure.

[0005] To solve the above technical problems, the present application provides a pile foundation settlement detection method applied to a gradually changing type of sediment, the method comprising:

[0006] Obtaining initial density data, electrical conductivity data of the sediment in the target sediment pit, and penetration resistance data of the pile foundation;

[0007] Analyzing and identifying a plurality of cross-sectional graphs obtained, determining air gap information in the sediment based on all analysis and identification results, and correcting the initial density data based on the air gap information to obtain gradually changing density data;

[0008] The cross-sectional graph is obtained by cutting a three-dimensional image of the sediment in a selected direction, the direction is determined by the settlement interaction direction between the target sediment pit and the sediment, and the cutting action scale is greater than the average particle size of the sediment;

[0009] Differential processing the gradually changing density data, processing the obtained density sensitive information based on the acoustic impedance technology to obtain density distribution information;

[0010] performing mechanical feature enhancement processing on the penetration resistance data, performing density calibration reconstruction processing on the density distribution information based on obtained penetration resistance mechanical feature parameters, and obtaining density continuous distribution information;

[0011] determining a bearing capacity reduction coefficient based on the density continuous distribution information;

[0012] performing feature extraction processing on the electrical conductivity data to obtain a sediment activity index;

[0013] inputting the sediment activity index and the bearing capacity reduction coefficient into a pile foundation settlement calculation model constructed by a neural network model to obtain pile foundation settlement data.

[0014] As one of the preferred solutions, the inputting the sediment activity index and the bearing capacity reduction coefficient into a pile foundation settlement calculation model constructed by a neural network model to obtain pile foundation settlement data includes:

[0015] processing the sediment activity index and the bearing capacity reduction coefficient input into the pile foundation settlement calculation model by using a dual-path attention mechanism to determine sediment adaptive weights and bearing capacity adaptive weights;

[0016] inputting the sediment activity index, the bearing capacity reduction coefficient, the sediment adaptive weights, and the bearing capacity adaptive weights into the pile foundation settlement calculation model to obtain pile foundation settlement data, with the pile body elastic compression amount of the pile foundation as a constraint, wherein the pile body elastic compression amount is obtained from the pile length and the pile cross-sectional area of the pile foundation.

[0017] As one of the preferred solutions, the determining a bearing capacity reduction coefficient based on the density continuous distribution information includes:

[0018] processing the density continuous distribution information by using a Gaussian filter to obtain a smooth density curve;

[0019] performing inflection point analysis on the smooth density curve to obtain density inflection point information;

[0020] performing feature dimension reduction processing on the penetration resistance mechanical feature parameters according to a principal component analysis method to obtain mechanical feature information;

[0021] performing inflection point analysis on the mechanical feature information by using a maximum value method to obtain mechanical inflection point information;

[0022] obtaining the bearing capacity reduction coefficient based on the density inflection point information and the mechanical inflection point information.

[0023] As one of the preferred solutions, the differential processing of the gradual density data is based on the acoustic impedance technology to process the obtained density sensitive information to obtain the density distribution information, including:

[0024] The wavelet threshold technology is used to perform noise reduction processing on the gradual density data to obtain purified density information;

[0025] The transformed instantaneous frequency analysis method is used to perform differential processing on the purified density information to obtain the density sensitive information;

[0026] The impedance inversion processing is performed on the density sensitive information to obtain the density distribution information.

[0027] As one of the preferred solutions, the impedance inversion processing of the density sensitive information to obtain the density distribution information includes:

[0028] The reflection coefficient conversion processing is performed on the density sensitive information to obtain reflection coefficient sequence data;

[0029] The reflection coefficient sequence data is processed by using the integral inversion technology with constraints to obtain acoustic impedance distribution data;

[0030] The wave velocity-density decoupling processing is performed on the acoustic impedance distribution data to obtain the density distribution information.

[0031] As one of the preferred solutions, the mechanical characteristic enhancement processing of the penetration resistance data is based on the obtained penetration resistance mechanical characteristic parameters to perform density calibration reconstruction processing on the density distribution information to obtain density continuous distribution information, including:

[0032] The wavelet packet decomposition technology is used to perform mechanical characteristic enhancement processing on the penetration resistance data to obtain penetration resistance mechanical characteristic parameters;

[0033] The state mapping processing is performed on the penetration resistance mechanical characteristic parameters to obtain an equivalent density field;

[0034] The density calibration reconstruction processing is performed on the equivalent density field and the density distribution information by using Bayesian fusion to obtain the density continuous distribution information.

[0035] As one of the preferred solutions, before obtaining the profile, the pile settlement detection method applied to the gradual sediment also includes:

[0036] Obtain the size data of a plurality of samples in the sediment;

[0037] The size data of a plurality of samples is input into a cutting model constructed by a neural network algorithm to determine the average particle size of the sediment;

[0038] obtain a work log constructed by historical pile foundation settlement data;

[0039] analyze the work log by using a large language model to obtain a settlement interaction direction between the target slag pit and the settled slag.

[0040] As one of the preferred solutions, the profile graph is obtained by cutting the three-dimensional image of the settled slag in the selected direction, further comprising:

[0041] Perform fan profile analysis on the settled slag with the center of the slag pit as the origin;

[0042] Based on the fan profile of the settled slag, use a sensor array to collect and process data to obtain initial echo signal data;

[0043] Use synthetic aperture focusing technology to perform imaging algorithm processing on the initial echo signal data to obtain the profile graph.

[0044] As one of the preferred solutions, based on the density inflection point information and the mechanical inflection point information, the bearing capacity reduction coefficient is obtained, comprising:

[0045] Quantitative fitting processing is performed on the density inflection point information to obtain a density critical gradient value;

[0046] Use mechanical parameter correlation modeling technology to perform eigenvalue extraction processing on the mechanical inflection point information to obtain a mechanical critical threshold value;

[0047] Based on the density critical gradient value and the mechanical critical threshold value, determine the bearing capacity reduction coefficient.

[0048] The present application further provides a pile foundation settlement detection system for gradually changing settled slag, comprising:

[0049] An acquisition module is configured to acquire initial density data, electrical conductivity data of settled slag in a target slag pit, and penetration resistance data of a pile foundation;

[0050] An identification module is configured to analyze and identify a plurality of profile graphs, determine air gap information in the settled slag based on all analysis and identification results, and correct the initial density data based on the air gap information to obtain gradually changing density data;

[0051] A cutting module is configured to obtain the profile graph by cutting a three-dimensional image of the settled slag in a selected direction, wherein the direction is determined by a settlement interaction direction between the target slag pit and the settled slag, and the cutting action scale is greater than the average particle size of the settled slag.

[0052] a differential module configured to perform differential processing on the gradually changing density data, process density sensitive information obtained based on acoustic impedance technology, and obtain density distribution information;

[0053] a reconstruction module configured to perform mechanical characteristic enhancement processing on the penetration resistance data, perform density calibration reconstruction processing on the density distribution information based on obtained penetration resistance mechanical characteristic parameters, and obtain density continuous distribution information;

[0054] a determination module configured to determine a bearing capacity reduction coefficient based on the density continuous distribution information;

[0055] an extraction module configured to perform feature extraction processing on the electrical conductivity data, and obtain a sediment activity index;

[0056] a generation module configured to input the sediment activity index and the bearing capacity reduction coefficient into a pile foundation settlement calculation model constructed by a neural network model, and obtain pile foundation settlement data.

[0057] Compared with the prior art, the present application has at least one of the following advantages:

[0058] The present application obtains initial density data, electrical conductivity data and penetration resistance data of a target sediment pit, analyzes and identifies a plurality of cross-section graphs obtained, determines air gap information in the sediment based on all analysis and identification results, corrects the initial density data based on the air gap information, and obtains gradually changing density data; the cross-section graphs are obtained by cutting a three-dimensional image of the sediment in a selected direction, the direction is determined by a settlement interaction direction between the target sediment pit and the sediment, and the cutting action scale is greater than the average particle size of the sediment; the gradually changing density data is subjected to differential processing, density sensitive information obtained based on acoustic impedance technology is processed, and density distribution information is obtained; the penetration resistance data is subjected to mechanical characteristic enhancement processing, density calibration reconstruction processing is performed on the density distribution information based on obtained penetration resistance mechanical characteristic parameters, and density continuous distribution information is obtained; a bearing capacity reduction coefficient is determined based on the density continuous distribution information; the electrical conductivity data is subjected to feature extraction processing, and a sediment activity index is obtained; the sediment activity index and the bearing capacity reduction coefficient are input into a pile foundation settlement calculation model constructed by a neural network model, and pile foundation settlement data is obtained.

[0059] Compared with the prior art, the application obtains the initial density, the conductivity and the pile foundation penetration resistance data of the sediment, analyzes the air gap information by the three-dimensional cross-section profile determined by the interaction direction of the slag pit and the sediment settlement, corrects the initial density to obtain the gradual density data; then, the density distribution information is obtained by differentiating the gradual density data and combining the acoustic impedance technology, and the mechanical characteristics of the penetration resistance data are enhanced to calibrate and reconstruct the density distribution to obtain the continuous density distribution information, and then the bearing capacity reduction coefficient is determined, and the sediment activity index is obtained by the conductivity feature extraction; finally, the two key parameters are input into the pile foundation settlement calculation model constructed by the neural network to accurately output the pile foundation settlement data, realize the accurate detection of the pile foundation settlement, and finally guarantee the safety stability and service life of the engineering structure. BRIEF DESCRIPTION OF DRAWINGS

[0060] Figure 1 is a process schematic diagram of the pile foundation settlement detection method for the gradually changing sediment in one of the embodiments of the application;

[0061] Figure 2 is a structural schematic diagram of the sediment cross-section profile in which the average particle diameter of the sediment is slightly larger than the profile reference;

[0062] Figure 3 is a structural schematic diagram of the sediment fan-shaped cross-section analysis diagram in one of the embodiments of the application;

[0063] Figure 4 is a schematic diagram of the sediment cross-section air gap in one of the embodiments of the application;

[0064] Figure 5 is a structural schematic diagram of the pile foundation settlement detection system for the gradually changing sediment in one of the embodiments of the application.

[0065] Reference signs:

[0066] Among them, 11, the acquisition module; 12, the identification module; 13, the cross-cutting module; 14, the difference module; 15, the reconstruction module; 16, the determination module; 17, the extraction module; 18, the generation module; 21, the profile diagram; 22, the fan-shaped cross-section; 23, the air gap. DETAILED DESCRIPTION

[0067] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments of the application. The purpose of providing these embodiments is to make the disclosure of the application more thorough and comprehensive. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of the application.

[0068] In the description of the present application, it should be noted that, unless otherwise defined, all technical and scientific terms used in the present application have the same meanings as those commonly understood by those skilled in the art. The terms used in the specification of the present application are only for the purpose of describing specific embodiments and are not intended to limit the present application. Those skilled in the art can understand the specific meanings of the above terms in the present application according to the specific circumstances.

[0069] An embodiment of the present application provides a pile foundation settlement detection method applied to a gradual change type of sediment, and specifically, please refer to Figure 1 , Figure 1 A flowchart of the pile foundation settlement detection method applied to the gradual change type of sediment in one embodiment of the present application is shown, and the method comprises the following steps:

[0070] S1: obtaining initial density data, conductivity data and pile foundation penetration resistance data of sediment in a target sediment pit;

[0071] S2: analyzing and identifying a plurality of cross-section graphs obtained, determining air gap information in the sediment based on all analysis and identification results, and correcting the initial density data based on the air gap information to obtain gradual change density data;

[0072] S3: the cross-section graph is obtained by cutting a three-dimensional image of the sediment in a selected direction, the direction is determined by the settlement interaction direction between the target sediment pit and the sediment, and the cutting action scale is greater than the average particle size of the sediment;

[0073] S4: differentiating the gradual change density data, processing the obtained density sensitive information based on acoustic impedance technology to obtain density distribution information;

[0074] S5: performing mechanical characteristic enhancement processing on the penetration resistance data, performing density calibration reconstruction processing on the density distribution information based on the obtained penetration resistance mechanical characteristic parameters to obtain density continuous distribution information;

[0075] S6: determining a bearing capacity reduction coefficient based on the density continuous distribution information;

[0076] S7: performing feature extraction processing on the conductivity data to obtain a sediment activity index;

[0077] S8: inputting the sediment activity index and the bearing capacity reduction coefficient into a pile foundation settlement calculation model constructed by a neural network model to obtain pile foundation settlement data.

[0078] Specifically, the initial density data of the sediment can be measured by the water immersion method, i.e., the sample sediment is immersed in water, and the initial density of the sediment is determined based on the displacement volume. However, when the sample is immersed in water, the air gap of the sediment is closed or not completely discharged due to the surface tension of water, resulting in a smaller measured displacement volume. That is, the displacement volume measured by the water immersion method includes the volume of the air gap, but air itself does not provide bearing capacity, so this method cannot truly reflect the density of the sediment.

[0079] To this end, the initial density needs to be corrected. The correction process includes: analyzing and identifying a plurality of obtained cross-sectional images, determining the air gap information in the sediment based on all analysis and identification results, and correcting the initial density data based on the air gap information to obtain the gradual density data.

[0080] Specifically, the cross-sectional image inside the sediment is obtained, the air gap data is obtained by analyzing and identifying the cross-sectional image, and the initial density is corrected based on the air gap data to obtain the gradual density data.

[0081] It should be noted that the cross-sectional image is obtained by cutting the three-dimensional image of the sediment in a selected direction, and the direction is determined by the settlement interaction direction between the target slag pit and the sediment. The cutting action scale is greater than the average particle size of the sediment.

[0082] Preferably, the cutting action scale is the minimum thickness of the cutting.

[0083] From the direction selection, the settlement interaction direction is the path where the sediment in the slag pit is most significantly deformed and the physical state changes most concentratedly, such as the sediment in a circular slag pit being pressed and interacted with the pit wall along the radial direction, and the sediment in a square slag pit being deformed along the diagonal direction. By cutting the cross-sectional image in this direction, the sensitive area of the sediment settlement can be directly cut, and the key information such as the layered structure and air gap distribution of the sediment in this direction can be fully presented, ensuring that the sediment characteristics reflected by the cross-sectional image are directly matched with the mechanical mechanism of pile settlement, and providing targeted samples for subsequent correction of density data using air gap and deduction of bearing capacity reduction coefficient.

[0084] From the scale, the sediment is a mixture of particles of different particle sizes. If the cutting scale is too small, the cross-sectional image will focus on the microstructure of a single or a few particles, rather than the macrostructure of the whole sediment. The data from this micro perspective cannot reflect the density distribution and air gap of the whole sediment as a medium, while the pile settlement calculation relies on the mechanical response of the whole sediment, rather than the micro characteristics of individual particles. Therefore, controlling the cutting scale can avoid the micro interference of small components and fully present the macro features of the sediment in the settlement interaction direction, such as the macro layering and large-scale air gap, ensuring that the air gap information and density data obtained by analysis conform to the physical state of the whole sediment.

[0085] Therefore, before analyzing and identifying the cross-section diagram, the interaction direction and particle size data are determined, specifically, the size data of a plurality of samples in the sediment is obtained; the size data of the plurality of samples is input into a cutting model constructed by a neural network algorithm to determine the average particle size of the sediment; a work log constructed from historical pile foundation settlement data is obtained; and a large language model is used to analyze the work log to obtain the settlement interaction direction between the target slag pit and the sediment.

[0086] First, the size data of a plurality of samples of the sediment is obtained, which covers the particle characteristics of different regions and depths of the sediment, avoiding the one-sidedness caused by a single or small number of samples, and providing multiple original data for subsequent model calculation.

[0087] The sample size data is input into a cutting model constructed by a neural network, which has strong nonlinear fitting and data clustering capabilities, can automatically learn the correlation between the particle size parameters in the sample and the overall average particle size, exclude the extreme interference of large particles / micro-particles on the average particle size, and perform weighting according to the particle number ratio, volume ratio, etc. Compared with traditional manual calculation, it can more accurately eliminate measurement errors and sample deviations, and output an average particle size that conforms to the actual particle distribution characteristics of the sediment, providing accurate numerical basis for subsequent cutting action with a size greater than the average particle size, and avoiding improper cutting size caused by inaccurate average particle size calculation.

[0088] Similarly, the settlement interaction direction is obtained from the work log constructed from historical pile foundation settlement data, which contains a large number of similar projects, such as the same slag pit shape, similar sediment type, similar geological conditions of pile foundation engineering, actual records, covering slag pit parameters, sediment characteristics, construction load, monitored settlement interaction direction, final settlement result, etc. Multidimensional information, use a large language model to analyze these multidimensional information, the large language model can break through the limitations of traditional keyword retrieval, and realize deep understanding of unstructured log data such as construction record text, monitoring report description, etc. On the one hand, automatically extract similar scene features in the log; on the other hand, correlate the settlement interaction direction conclusions of these similar cases, and fine-tune them in combination with the special differences of the target project, and finally output the settlement interaction direction adapted to the target slag pit and sediment.

[0089] In an embodiment, as shown in Figure 2 , the method comprises the following steps. Figure 2This is a structural schematic diagram of a cross-sectional view of sediment, with the average particle size slightly larger than the sediment's average particle size as the cross-sectional reference in one embodiment of the present invention. Figure 21 is the cross-sectional view. The specific analysis process includes: performing particle size analysis on the sediment sample data to obtain the original particle size data of several samples in the sediment; determining the average particle size of the sediment based on the original particle size data; determining the cutting action scale based on the average particle size of the sediment as the cross-sectional reference; and processing the sediment based on the cutting action scale to obtain several cross-sectional views of the sediment.

[0090] Considering that the particle size characteristics of sediment directly determine the formation pattern of its internal air gaps, fine particles are easy to compact and have small and uniform air gaps, while coarse particles are prone to forming large pores, laser particle size analyzers and other technical means are used to detect the size, particle size distribution range, and proportion of particles of different sizes in sediment samples to obtain the original particle size data.

[0091] Since the particle size of sediment directly affects the smallest identifiable structural unit, i.e., the profile thickness, the average particle diameter essentially sets a benchmark for the accuracy of subsequent profiles. This ensures that the profile captures structural information that matches the particle size, avoiding structural distortion due to overly coarse profiles or redundancy due to overly fine profiles. The average particle diameter is a representative particle diameter calculated by weighting the proportion of different particle sizes in the original particle size data. Therefore, using the average particle diameter of the sediment as the profile benchmark determines the cutting action scale, ensuring that each profile covers the structural differences at different depths of the sediment, and that the spacing between adjacent profiles is sufficient to reflect the spatial changes in the gaps between particles and air.

[0092] Based on the cutting action scale, the sediment sample is processed into layers using physical cutting or non-destructive imaging methods. Each layer corresponds to a cross section, and images of each cross section are acquired using equipment such as high-resolution cameras and CT imaging systems. Finally, several cross-sectional images reflecting the internal structure of the sediment at different locations are obtained.

[0093] In another embodiment, a fan-shaped profile analysis of the sediment is performed with the center of the slag pit as the origin, such as... Figure 3 As shown, Figure 3 This is a fan-shaped cross-sectional analysis diagram of sediment in one embodiment of the present invention. In the diagram, 22 represents the fan-shaped cross-section, which specifically includes: based on the fan-shaped cross-section of the sediment, data acquisition and processing are performed using a sensor array to obtain initial echo signal data; synthetic aperture focusing technology is used to process the initial echo signal data using an imaging algorithm to obtain cross-sectional image sequence data; and based on the cross-sectional image sequence data, a cross-sectional view of the sediment is obtained.

[0094] The fan-shaped profile is adopted instead of a single linear profile to realize three-dimensional coverage and multi-directional sampling of the settled slag in the slag pit. The radial structure of the fan shape can extend from the origin to the periphery of the slag pit, covering settled slag areas of different radii and depths, and ensuring that the collected samples can reflect the overall distribution characteristics of the settled slag in the slag pit.

[0095] Specifically, a sensor array matching the fan-shaped profile is adopted. Multiple sensors in the array can be synchronously deployed along different radial directions and depths of the fan shape, and simultaneously collect signals at different positions, thereby improving the data collection efficiency.

[0096] The collection principle of the echo signal is based on physical detection logic such as ultrasonic waves and electromagnetic waves. That is, the sensor emits a detection signal into the settled slag. When the signal encounters different media such as solid particles, air gaps, and water in the settled slag, it will be reflected, forming an echo signal. The strength and delay time of the echo signal are directly related to the medium characteristics inside the settled slag. Therefore, the structure and morphology inside the settled slag can be restored through the echo signal data.

[0097] The synthetic aperture focusing technology precisely calibrates, superimposes, and focuses the dispersed echo signals collected by multiple sensors in the sensor array according to the spatial position and time delay, thereby greatly improving the spatial resolution of the signals. After processing by the synthetic aperture focusing technology, the originally dispersed echo signals are focused to the corresponding spatial positions, and the signals of different medium interfaces are clearly separated, and finally converted into profile image sequence data. The profile image sequence data includes continuous images generated along different depths and different radial directions of the fan-shaped profile, including layer-by-layer images from the surface of the slag pit to the bottom and radius-by-radius images from the origin to the edge. Each image can intuitively display the internal structure of the settled slag at the corresponding position, such as the position, size, and distribution of air gaps and the aggregation state of solid particles.

[0098] The profile image sequence data contains a large number of continuous image frames, but not all images have analytical value. Therefore, based on the spatial positioning logic of the fan-shaped profile (with the origin as the center and the radial direction covering the range), effective image frames reflecting the main area of the settled slag are selected. Secondly, the selected effective images are calibrated for consistency to ensure that images at different positions can be compared and spliced. Finally, the settled slag profile is obtained.

[0099] In yet another embodiment, the settled slag is subjected to rotational sectioning. Specifically, the cone beam computed tomography technology is used to perform projection data collection and processing on the cross section obtained by rotational sectioning processing, to obtain two-dimensional projection image sequence data. Based on an image reconstruction algorithm, the two-dimensional projection image sequence data is processed to obtain the profile of the settled slag.

[0100] The essence of rotating cross-section is to make the cross-sectioning motion revolve around the central axis of the sediment sample, so as to achieve 360-degree circular sampling of the sediment. This covers the radial and axial structure of the sediment from different angles, ensuring that the data collected later can reflect the complete structural features of the three-dimensional space inside the sediment and avoid the one-sidedness of information caused by a single perspective.

[0101] Therefore, cone-beam computed tomography (CBCT) technology can cover a larger detection range in one go. Combined with rotation, it can simultaneously collect projection data from multiple angles, which not only improves the acquisition efficiency but also avoids sample disturbance caused by multiple scans. For example, sediment samples are prone to structural changes due to multiple operations, which can lead to the closure of air gaps.

[0102] When the detector captures the intensity signal of the penetrating rays, it converts it into grayscale values ​​to form a two-dimensional projection image. As it continues to rotate, the detector will collect a two-dimensional projection image at each rotation angle, eventually forming a sequence of data. For example, if one image is collected for every 1 degree of rotation, 360 projection images at different angles will be collected.

[0103] Each projection image records the medium distribution characteristics of the sediment at the corresponding angle. All the sequence images together constitute a data mosaic that reconstructs the three-dimensional structure of the sediment. Therefore, by using image reconstruction algorithms to process the two-dimensional projection image sequence data, a cross-sectional view of the sediment can be obtained.

[0104] Since the cross-sectional view obtained by the above method can only visually observe the morphology of the air gap, such as Figure 4 As shown, Figure 4 This is a schematic diagram of the air gap in the sediment cross-section in one embodiment of the present invention, where 23 represents the air gap. Since the cross-sectional view cannot be directly used for density correction, the core of image feature extraction is to convert the air gaps in the image into precise quantization parameters using digital image processing technology. The specific operation process is as follows:

[0105] First, by using algorithms such as threshold segmentation and edge detection, the sediment particle region and the air gap region are distinguished in the image. Then, by using pixel statistics or morphological analysis, key parameters such as the volume ratio of the air gap, the equivalent diameter of a single air gap, and the spatial distribution density of the air gap are calculated, and finally the air gap data is obtained.

[0106] Specifically, before threshold segmentation and edge detection, the quality of the sediment image needs to be addressed. Sediment particles are mostly irregular mixtures of soil and rock, which are prone to image blurring due to uneven lighting, particle shadows, and background noise, directly affecting the subsequent segmentation accuracy. Therefore, image preprocessing is required. The core objective is to enhance the grayscale difference between particles and air and eliminate interference information, which can be achieved through grayscale processing, noise suppression, and contrast enhancement.

[0107] For how to accurately separate the pixel area representing the air gap from the pretreated image, the threshold segmentation and edge detection strategy should be selected according to the characteristics of the sediment image. Threshold segmentation is a coarse segmentation method based on gray difference. The core logic is to set a gray threshold T, and the pixels with gray > T in the image are judged as air gap, and the pixels with gray ≤ T are judged as sediment particles.

[0108] After segmentation, the clear outline of particle-air is extracted by edge detection. When high-precision outline is needed, Canny edge detection is selected; when fast and rough outline extraction is needed, Sobel edge detection is selected.

[0109] The air gap volume ratio = total number of air gap pixels / total number of image pixels, provided that the pixel size has been calibrated during image acquisition, that is, by using a calibration board with known size to determine the actual physical size corresponding to one pixel, ensuring the accuracy of the conversion relationship between pixel statistics and actual volume; The volume ratio directly reflects the compactness of the sediment. The higher the ratio, the looser the sediment; The lower the ratio, the denser the sediment.

[0110] Since the air gap is mostly irregular in shape, it cannot be directly measured in diameter, and it needs to be realized through equivalent area. The equivalent diameter determines the permeability of the sediment. The larger the diameter, the easier it is to form a connected channel for air gap, and rainwater or groundwater is easy to penetrate, causing the sediment to soften and the bearing capacity to be reduced; The smaller the diameter, the gap is mostly isolated, and the permeability is weak, and the sediment stability is better.

[0111] The spatial distribution density is obtained by the number of air gaps / actual volume of the image. The number of independent air gaps in the image needs to be counted through connected region analysis to avoid misjudgment of connected gaps as multiple, and then the density is calculated in combination with the actual volume of the image; The uniformity of the spatial distribution density of the air gap is directly related to the settlement risk of the sediment. Non-uniform distribution can cause local loosening of the sediment, and uneven settlement is easy to occur later; When the distribution is uniform and the density is low, the sediment is uniformly stressed, and the settlement risk is small.

[0112] By integrating the volume ratio, equivalent diameter and spatial distribution density obtained into the air gap data set, and comparing it with the design standard of the sediment project, it is further verified whether the sediment bearing capacity meets the design value.

[0113] By converting the structural difference into numerical parameters, quantitative support is provided for density correction.

[0114] The formula for the initial density measured by the water immersion method is:

[0115]

[0116] where, is the initial density, is the total mass of the sample, is the total volume of the sample, and the total volume of the sample = the volume of the settled particles + the volume of the air gap, since the air itself does not provide carrying capacity, the initial density is actually a false density containing invalid air, and the core of the correction is to remove the volume of the air gap from the total volume of the sample, and to calculate the effective density with the real particle volume, that is,

[0117]

[0118] wherein, is the corrected density, is the air gap data, obtained by the volume of the air gap / the total volume of the sample .

[0119] Since the porosity at different positions in the profile is different, the corrected density will also change with the position, and finally form the gradient density data.

[0120] Next, the gradient density data is subjected to differential processing, the density sensitive information obtained is processed based on the acoustic impedance technology to obtain the density distribution information, and the specific process includes: the wavelet threshold technology is used to carry out noise reduction processing on the gradient density data to obtain purified density information; the purified density information is subjected to differential processing based on the transformed instantaneous frequency analysis method to obtain the density sensitive information; the density sensitive information is subjected to impedance inversion processing to obtain the density distribution information.

[0121] Specifically, the wavelet threshold technology is a signal denoising method based on wavelet transform, and the core logic is: first, the gradient density data is decomposed into wavelet coefficients of different frequency scales through wavelet transform, then the coefficients are screened according to the preset threshold rule, the effective coefficients reflecting the real change of the density are retained, and the noise coefficients generated by the measurement error and the environmental interference are suppressed or removed, and finally the processed coefficients are reconstructed into new density data, i.e. purified density information, through wavelet inverse transform.

[0122] Then, the purified density information is extracted through signal transformation (such as Hilbert-Huang transform, wavelet transform, etc.), the instantaneous change frequency of the density at different positions, such as different depths and different radial positions of the sediment profile, is analyzed, the differential processing of the transformed signal is carried out, i.e. the density difference value of adjacent sampling points is calculated, the density difference value is combined with the instantaneous frequency characteristics, and finally the data reflecting the sensitive area of the density change, i.e. the density sensitive information, is formed.

[0123] Through the establishment of the correlation model of the density sensitive information and the acoustic impedance, the change amount in the density sensitive information is converted into corresponding acoustic impedance change amount data, through the inversion algorithm, the density value of the sediment in the whole space range is inversely deduced according to the acoustic impedance change amount, and the inverse deduction result is presented in the form of an image, i.e. the density distribution information.

[0124] Specifically, in an embodiment, the density-sensitive information is subjected to impedance inversion processing to obtain the density distribution information, including: the density-sensitive information is subjected to reflection coefficient conversion processing to obtain reflection coefficient sequence data; the reflection coefficient sequence data is processed by using a band-constrained integral inversion technique to obtain acoustic impedance distribution data; the acoustic impedance distribution data is subjected to wave velocity-density decoupling processing to obtain the density distribution information.

[0125] From the physical nature, the density-sensitive information is an indirect representation of the internal density variation of the sediment, and the density-sensitive information is converted into reflection coefficient sequence data conforming to the physical law of acoustic wave propagation by reflection coefficient, which builds a bridge for subsequent acoustic impedance inversion of density distribution, and ensures that subsequent analysis is always based on actual physical phenomena, avoiding data from deviating from the application scenario. It should be noted that the reflection coefficient is a core propagation parameter of acoustic waves at the interface between two different media, and its size is directly determined by the acoustic impedance difference of the media on both sides of the interface.

[0126] However, in the actual sediment sample, the density distribution is continuously changing, and there may be local abnormal areas, which are prone to problems of disconnection between the calculation results and the actual physical scene. Therefore, by using the band-constrained integral inversion technique, physical constraint conditions such as the acoustic impedance being a positive value, the overall impedance range of the sediment conforming to the known physical properties of the same material, and the impedance change of adjacent regions being continuous are introduced to modify and limit unreasonable results in the integral calculation process, and finally the acoustic impedance distribution data obtained can match the measured information of the reflection coefficient sequence and conform to the actual physical properties of the sediment, ensuring the reliability of the data.

[0127] Acoustic impedance is a coupled parameter of density and wave velocity, and density cannot be directly determined by impedance data. Therefore, to obtain the density distribution, the density must be separated from the coupling relationship between impedance and wave velocity through decoupling processing. In actual processing, the decoupling is usually realized by combining the physical properties of the sediment or auxiliary measurement data, and finally the acoustic impedance distribution data is converted into density distribution information directly reflecting the spatial variation of the internal density of the sediment through wave velocity-density decoupling.

[0128] The density distribution information is subjected to density calibration reconstruction processing based on the obtained penetration resistance mechanical characteristic parameters to obtain continuous density distribution information, including: the penetration resistance data is subjected to mechanical characteristic enhancement processing by using a wavelet packet decomposition technique to obtain penetration resistance mechanical characteristic parameters; the penetration resistance mechanical characteristic parameters are subjected to state mapping processing to obtain an equivalent density field; the equivalent density field and the density distribution information are subjected to density calibration reconstruction processing by using Bayesian fusion to obtain the continuous density distribution information.

[0129] Penetration resistance data is the recorded data of the reaction force experienced by the equipment when the pile foundation is inserted into the sediment.

[0130] Considering that the obtained penetration resistance data not only contains information on the mechanical properties of the sediment, but also includes noise from equipment vibration, and that key mechanical characteristics, such as resistance peak, slope of change and fluctuation frequency, are often masked by noise, it is difficult to directly correlate them with sediment density.

[0131] Therefore, wavelet packet decomposition technology is used for noise reduction, purification and feature extraction. First, high-frequency interference noise in the original penetration resistance data is separated and filtered out, while retaining the effective signal that reflects the true mechanical state of the sediment. Then, mechanical characteristic parameters that are strongly correlated with the sediment density are extracted from the purified signal.

[0132] Then, in order to correlate the obtained mechanical characteristic parameters with the sediment density, a preset mapping model, such as a linear regression model or a machine learning mapping model, was used to convert the extracted mechanical characteristic parameters into corresponding density values. For example, if the experimentally calibrated resistance increases by 10 kPa, the density increases by 0.2 g / cm³. 3 The local density can be inferred from the measured resistance. Therefore, the mechanical-density correspondence of a single or local measuring point can be combined with the movement trajectory of the penetrating equipment to expand into an equivalent density field covering the entire sediment.

[0133] To ensure data accuracy, Bayesian fusion is used to perform density calibration and reconstruction on the equivalent density field and the density distribution information, thereby obtaining more reliable results.

[0134] Specifically, Bayesian fusion first evaluates the reliability weights of the two types of data. If the measurement error of the original density distribution information in a certain region is small, then the original data in that region has a high weight; if the equivalent density field has dense measurement points and clear mechanical characteristics in a certain region, then the equivalent density field in that region has a high weight. Through weight allocation, the bias of individual data is corrected.

[0135] Furthermore, to address the spatial discontinuity caused by the sparse measurement points of the equivalent density field, Bayesian fusion takes the spatial continuity of the original density distribution information as a basis, combines it with the local accurate data of the equivalent density field, and fills in the density values ​​of the unmeasured points through probability interpolation, ultimately obtaining spatially continuous and data-reliable density continuous distribution information.

[0136] Preferably, probability interpolation is a spatial prediction method based on Bayesian posterior distribution.

[0137] Based on the obtained density continuous distribution information, the bearing capacity reduction coefficient is determined, including: the density continuous distribution information is processed by using Gaussian filtering to obtain a smooth density curve; the smooth density curve is analyzed for inflection points to obtain density inflection point information; the mechanical characteristic parameters of the penetration resistance are processed for feature dimension reduction according to a principal component analysis method to obtain mechanical characteristic information; the mechanical characteristic information is analyzed for inflection points by using a maximum value method to obtain mechanical inflection point information; and the bearing capacity reduction coefficient is obtained based on the density inflection point information and the mechanical inflection point information.

[0138] Although the density continuous distribution information has eliminated most errors through the previous processing, it may still contain high-frequency noise caused by local particle random distribution and slight shaking of the measuring equipment. These noises will interfere with the subsequent inflection point analysis and misidentify the noise fluctuations as real density change inflection points. Gaussian filtering retains the overall trend of the density data and suppresses local meaningless fluctuations, thereby providing data support for accurately identifying the key nodes of the density change in the subsequent process.

[0139] An inflection point is a point where the second derivative of a function changes from positive to negative or from negative to positive, reflecting the mutation of the curve change rate. In the smooth density curve, the inflection point corresponds to the position where the sediment density change trend changes significantly. In the specific operation, the second derivative of the smooth density curve is calculated to find the point where the second derivative is zero, which is the density inflection point. Then, combined with engineering experience, such as the density value and depth position at the inflection point, the inflection points with actual physical meaning are screened out to form the density inflection point information.

[0140] The density inflection point is essentially a sign of the stratification of the internal structure of the sediment. The density change law on both sides of the inflection point is different, which means that there are essential differences in the sediment particle accumulation state and density between the corresponding regions.

[0141] The number of the penetration resistance mechanical characteristic parameters extracted in the previous stage is large, and some of the parameters have strong correlation. Directly using them for analysis will lead to complex calculation and increased interference factors. The principal component analysis eliminates redundant information by dimension reduction and retains the core features that best reflect the mechanical nature of the sediment. Then, the maximum value method is used to find the point with the largest change rate in the mechanical characteristic information to determine the mechanical inflection point.

[0142] The mechanical inflection point is a sign of the stratification of the mechanical properties of the sediment. The mechanical characteristics on both sides of the inflection point are significantly different, which means that there are essential differences in the bearing capacity of the sediment. Similar to the density inflection point, the mechanical inflection point information provides a stratification basis for the sediment from the mechanical perspective and can be mutually verified with the density inflection point. Ideally, the density inflection point and the mechanical inflection point should be in similar positions.

[0143] The density inflection point and the mechanical inflection point are combined to demarcate the bearing layer of the sediment, for each layer, the average density and the average mechanical characteristics are calculated, and the bearing capacity reduction coefficient of each layer is calculated according to the engineering specification, and the weighted average of the layer reduction coefficient of the whole area is obtained by weighting the thickness proportion of each layer, and the final bearing capacity reduction coefficient is obtained.

[0144] The density inflection point reflects the layer of physical density, and the mechanical inflection point reflects the layer of actual bearing capacity. The combination of the two can avoid the limitation of a single index. Through the comprehensive layer information of the two, the calculated bearing capacity reduction coefficient can not only reflect the physical nature of the sediment, but also conform to the actual mechanical performance.

[0145] The sediment with high conductivity indicates that the water content is high and the ions are active. In the later stage, additional settlement may occur due to water evaporation and colloid shrinkage, and the long-term settlement is larger. Therefore, the sediment activity index can be obtained by performing feature extraction processing on the conductivity data, solving the problem that the existing technology only calculates short-term settlement and ignores the influence of long-term material activity.

[0146] In an embodiment, based on the density inflection point information and the mechanical inflection point information, the bearing capacity reduction coefficient is obtained, including: performing quantitative fitting processing on the density inflection point information to obtain a density critical gradient value; performing eigenvalue extraction processing on the mechanical inflection point information by using a mechanical parameter correlation modeling technology to obtain a mechanical critical threshold value; and determining the bearing capacity reduction coefficient based on the density critical gradient value and the mechanical critical threshold value.

[0147] Specifically, the density inflection point information is essentially the density mutation feature record of the sediment at different depths and different settlement stages, such as the position data where the density at a certain depth of the sediment changes from uniform growth to sudden drop, the numerical point where the density change rate suddenly changes, etc. The quantitative fitting processing will be based on the original monitoring data of the density inflection point, and a suitable mathematical model such as linear regression, nonlinear fitting function, etc. is selected for data reconstruction: on the one hand, the monitoring errors in the original data are eliminated to ensure that the fitting result conforms to the real law of the density change of the sediment; on the other hand, through the slope change of the fitting curve, the second derivative extreme value and other characteristics, the critical change rate where the density changes from normal distribution to abnormal decay / mutation, i.e. the density critical gradient value, is located.

[0148] The purpose of this step is to convert the fuzzy density inflection point phenomenon into a quantifiable and calculable density critical gradient value. This value not only accurately reflects the critical state of the failure change of the sediment density, but also provides a unified density dimension quantitative basis for the subsequent calculation of the bearing capacity reduction coefficient in combination with the mechanical parameters, avoiding the coefficient deviation caused by the original inflection point information which cannot be directly compared and calculated.

[0149] Next, the core logic of the mechanical parameter correlation modeling technology is to first build a correlation model of the mechanical inflection point information and the key parameters of the sediment, and to clearly define the collaborative variation law of each parameter when the mechanical inflection point occurs; then, based on the model, the characteristic values of the mechanical inflection point information are extracted, and not simply the peak or mutation point of the mechanical curve is selected, but the influence of the density and other related parameters is combined to filter out the critical index that can represent the mechanical performance of the sediment, i.e. the mechanical critical threshold.

[0150] Considering that the essence of the bearing capacity reduction coefficient is the decay ratio of the actual bearing capacity of the sediment relative to the ideal state, and the core influencing factor is the density characteristics and mechanical properties of the sediment, the density critical gradient value and the mechanical critical threshold value are used as the core input, and the final coefficient is determined through the collaborative calculation of the two.

[0151] An associated calculation model of the density critical gradient value, the mechanical critical threshold value and the bearing capacity reduction coefficient is established, and the model is used to calculate the bearing capacity reduction coefficient.

[0152] In addition, in another embodiment of the present application, the AI algorithm model can also be realized, which can learn the statistical rules, characteristic relationships and patterns in the data through massive data training. In the application process, the trained algorithm model can obtain the desired data results, and the processing process of the algorithm model is as follows:

[0153] The first training method is to link the AI algorithm model with the database / knowledge base, which can be pre-stored in the cloud / local server or formulated by the industry safety specification standard in the field. The second training method is: first, build a sample data set, which not only includes the density critical gradient value and the mechanical critical threshold value, but also assigns corresponding data result labels to them, which are used to represent the bearing capacity reduction coefficient corresponding to the selected density critical gradient value and the selected mechanical critical threshold value; then, based on the learning algorithm, the neural network of the AI algorithm model is trained and learned based on the above sample data set. In the training process, the AI algorithm model can be retrained or fine-tuned according to the method of the prior art to improve the generalization ability of the model; finally, the second AI algorithm model is obtained after training. In actual application, the density critical gradient value and the mechanical critical threshold value are input into the second AI algorithm model, which is analyzed and processed by the model, and then the bearing capacity reduction coefficient is output.

[0154] Through the processing process of the algorithm model, the AI algorithm model can have the input density critical gradient value, the mechanical critical threshold, the ability of logical deduction and output of the bearing capacity reduction coefficient. It should be noted that the above two training methods are only examples, and those skilled in the art can also select other appropriate ways according to the scene, or can combine the two training methods, for example, using a specific field database to perform secondary training on the trained AI algorithm model to strengthen the ability to obtain the bearing capacity reduction coefficient, which will not be described in detail in the embodiments of the present application. Similarly, the training and learning of the AI algorithm model can also be supervised learning (convolutional neural network, recurrent neural network, etc.), unsupervised learning (generative adversarial network, autoencoder), semi-supervised learning (self-training, consistency regularization), reinforcement learning (gradient policy method), federated learning or transfer learning, etc. General learning paradigm, which is not specifically limited in the embodiments of the present application.

[0155] In step S8, the sediment activity index and the bearing capacity reduction coefficient are input into a pile foundation settlement calculation model constructed by a neural network model to obtain pile foundation settlement data, specifically including: processing the sediment activity index and the bearing capacity reduction coefficient input into the pile foundation settlement calculation model by using a double-path attention mechanism to determine sediment adaptive weights and bearing capacity adaptive weights; inputting the sediment activity index, the bearing capacity reduction coefficient, the sediment adaptive weights and the bearing capacity adaptive weights into the pile foundation settlement calculation model with the pile body elastic compression amount of the pile foundation as a constraint to obtain pile foundation settlement data, wherein the pile body elastic compression amount is obtained from the pile length and the pile cross-sectional area of the pile foundation.

[0156] The double-path attention mechanism is designed with two independent attention calculation paths for the two core input parameters of the sediment activity index and the bearing capacity reduction coefficient, rather than using fixed weights.

[0157] In the sediment activity index path, by learning a large amount of engineering data, based on the engineering experimental data of the actual settlement corresponding to different active sediments, the contribution weight of the parameter to the settlement is calculated; if the sediment activity is very high in a certain project, such as high water content and high ion concentration, it is easy to shrink due to water loss in the later period, and the model will automatically increase the sediment adaptive weight, so that the influence of the activity index on the settlement calculation is more significant.

[0158] Similarly, for the bearing capacity reduction coefficient, the model dynamically adjusts the bearing capacity adaptive weight by learning the correlation between the reduction coefficient and the settlement.

[0159] The double-path attention mechanism dynamically allocates weights through data driving, so that the model can adapt to different engineering scenes, avoid the one-size-fits-all deviation caused by fixed weights, and ensure that the parameter influence is consistent with the actual working condition.

[0160] The elastic compression amount of the pile body is the compression amount of the pile body material due to elastic deformation under the action of the load, and the calculation is based on the basic formula of material mechanics and is directly related to the pile length and the pile cross-sectional area.

[0161] When inputting the sedimentation activity index, the bearing capacity reduction coefficient, the sedimentation adaptive weight, and the bearing capacity adaptive weight into the neural network model, the elastic compression amount of the pile body needs to be taken as a physical constraint condition and integrated into the calculation process. Finally, the model outputs the pile foundation settlement data that meets the data rules and physical constraints through nonlinear fitting.

[0162] After obtaining the pile foundation settlement data, the method further includes: obtaining a building risk database constructed by historical pile foundation settlement data; inputting the pile foundation settlement data into the building risk database for analysis to determine a settlement risk level; and executing a safety response strategy corresponding to the settlement risk level.

[0163] The construction of the building risk database needs a large number of completed and long-term monitored pile foundation projects as samples. Each sample needs to include two types of core data: one is historical pile foundation settlement data, including settlement amount, settlement rate, differential settlement value, etc. at different use stages, and the other is risk results and environmental information corresponding to the project.

[0164] From the database, the same type and environment of historical engineering samples are selected, and the pile foundation settlement data of the project, such as the current settlement amount, settlement rate, and differential settlement value, are compared with the filtered historical sample data to determine the settlement risk level. The measures corresponding to each risk level are shown in the following table.

[0165]

[0166] After obtaining the pile foundation settlement data, the method further includes supplementing key links from the dimensions of data depth verification, risk root tracing, long-term dynamic control, and result review optimization to form a more complete pile foundation settlement management closed loop.

[0167] Specifically, the effectiveness verification of the pile foundation settlement data can be operated from data outlier screening, multi-source data cross verification, and environmental factor correction.

[0168] Data outlier screening is to identify abnormal data by combining the principles of monitoring equipment, such as total station and settlement meter, and engineering scene, for example, short-term soil rebound caused by heavy rain, jump data caused by monitoring instrument failure, or missing data caused by monitoring point being blocked, etc.

[0169] Multi-source data cross-validation is to compare the current settlement data with other monitoring indicators of the same pile foundation, and the settlement trend of adjacent pile foundations. For example, if the settlement of a pile foundation suddenly increases, but the adjacent pile foundation does not change significantly and the pile stress is normal, it is necessary to check whether it is a monitoring point error; if the surrounding strata are sinking synchronously, it is a regional geological problem.

[0170] The environmental factor correction is to eliminate the interference of non-structural factors such as temperature, humidity, and groundwater level changes on the settlement data. For example, short-term micro-settlement of concrete pile foundation caused by temperature shrinkage needs to be corrected by combining local meteorological data and material thermal expansion and contraction coefficient to avoid misjudgment as structural settlement.

[0171] Data anomaly value screening can avoid risk misjudgment caused by invalid data or interference data. If abnormal data is used directly without verification, it may cause over-treatment; if the environmental factor correction is ignored, it may also miss the real structural settlement hidden danger, leading to inaccurate risk control.

[0172] The settlement risk source can be processed through geological condition backtracking, engineering construction tracing, and structure correlation analysis.

[0173] Geological condition backtracking can combine geological reports in the survey stage, such as stratum distribution, soil bearing capacity, and groundwater level change history, to analyze whether the settlement is related to geological anomalies; engineering construction operation tracing is to check the pile foundation construction process and later operation load to determine whether there is settlement caused by construction defects; structure correlation analysis is based on the evaluation of the mutual influence of pile foundation settlement and superstructure.

[0174] The pile foundation settlement data, risk judgment result, root cause analysis report, response strategy execution record, and treatment effect feedback are sorted and archived according to the specification to form a complete pile foundation settlement control file.

[0175] In addition, if the pile foundation settlement involves public safety, the local housing and construction department, urban management department, or surrounding residents should be informed of the risk situation and disposal progress in a timely manner to avoid causing panic and to seek external cooperation.

[0176] Another embodiment of the present application provides a pile foundation settlement detection system applied to gradual sedimentation, specifically, please refer to Figure 5 , Figure 5 The pile foundation settlement detection system applied to gradual sedimentation in one embodiment of the present application is shown, which includes:

[0177] The acquisition module 11 is used to acquire the initial density data, electrical conductivity data of the sediment in the target sediment pit, and the penetration resistance data of the pile foundation;

[0178] A recognition module 12 is configured to recognize the obtained profile graphs, determine air gap information in the sediment based on all the recognition results, and correct the initial density data based on the air gap information to obtain gradual density data.

[0179] A cutting module 13 is configured to cut the three-dimensional image of the sediment in a selected direction to obtain the profile graph, the direction being determined by the sedimentation interaction direction between the target slag pit and the sediment, and the cutting scale being greater than the average particle size of the sediment.

[0180] A difference module 14 is configured to perform difference processing on the gradual density data, process the obtained density sensitive information based on the acoustic impedance technology, and obtain density distribution information.

[0181] A reconstruction module 15 is configured to perform mechanical characteristic enhancement processing on the penetration resistance data, perform density calibration reconstruction processing on the density distribution information based on the obtained penetration resistance mechanical characteristic parameters, and obtain density continuous distribution information.

[0182] A determination module 16 is configured to determine a bearing capacity reduction coefficient based on the density continuous distribution information.

[0183] An extraction module 17 is configured to perform feature extraction processing on the electrical conductivity data to obtain a sediment activity index.

[0184] A generation module 18 is configured to input the sediment activity index and the bearing capacity reduction coefficient into a pile foundation settlement calculation model constructed by a neural network model to obtain pile foundation settlement data.

[0185] Compared with the prior art, the embodiment of the present application has at least one of the following advantages:

[0186] The application obtains initial density data, conductivity data and pile foundation penetration resistance data of the target slag pit; a plurality of profile graphs are analyzed and identified, and the air gap information in the settled slag is determined based on all the analysis and identification results, and the initial density data is corrected based on the air gap information to obtain gradual density data; wherein the profile graph is obtained by cutting the three-dimensional image of the settled slag in a selected direction, the direction is determined by the settlement interaction direction between the target slag pit and the settled slag, and the cutting action scale is greater than the average particle size of the settled slag; the gradual density data is subjected to difference processing, the obtained density sensitive information is processed based on acoustic impedance technology to obtain density distribution information; the penetration resistance data is subjected to mechanical characteristic enhancement processing, and the density distribution information is subjected to density calibration reconstruction processing based on the obtained penetration resistance mechanical characteristic parameters to obtain density continuous distribution information; based on the density continuous distribution information, a bearing capacity reduction coefficient is determined; the conductivity data is subjected to feature extraction processing to obtain a settled slag activity index; the settled slag activity index and the bearing capacity reduction coefficient are input into a pile foundation settlement calculation model constructed by a neural network model to obtain pile foundation settlement data.

[0187] Compared with the prior art, the application obtains initial density, conductivity and pile foundation penetration resistance data of the settled slag, analyzes the air gap information from the three-dimensional cutting profile graph determined by the slag pit and the settled slag settlement interaction direction, corrects the initial density to obtain gradual density data; then the gradual density data is differentiated and combined with the acoustic impedance technology to obtain the density distribution information, and the penetration resistance data is subjected to mechanical characteristic enhancement to calibrate and reconstruct the density distribution to obtain the density continuous distribution information, and then the bearing capacity reduction coefficient is determined, and the settled slag activity index is obtained by extracting the conductivity features; finally, the two key parameters are input into the pile foundation settlement calculation model constructed by the neural network to accurately output the pile foundation settlement data, realize accurate detection of the pile foundation settlement, and finally ensure the safety and stability of the engineering structure and the service life.

[0188] The above-described embodiments only express several embodiments of the application, which are described in detail and specifically, but should not be understood as limiting the scope of the patent of the application. It should be noted that for ordinary skilled persons in the art, without departing from the concept of the application, a number of modifications and improvements can be made, which are within the scope of protection of the application. Therefore, the protection scope of the patent of the application should be subject to the appended claims.

Claims

1. A method for detecting settlement of a pile foundation applied to a progressive type of sedimentation, characterized by, The method comprises the following steps: obtaining initial density data, electrical conductivity data and pile foundation penetration resistance data of the settled sediment in the target sediment pit; analyzing and identifying a plurality of cross-section graphs obtained, and determining air gap information in the settled sediment based on all analysis and identification results, and correcting the initial density data with the air gap information to obtain gradual density data; wherein the cross-section graph is obtained by cutting a three-dimensional image of the settled sediment in a selected direction, the direction being determined by the settlement interaction direction between the target sediment pit and the settled sediment, and the cutting action scale being greater than the average particle size of the settled sediment; differentially processing the gradual density data, processing the obtained density sensitive information based on acoustic impedance technology to obtain density distribution information; performing mechanical characteristic enhancement processing on the penetration resistance data, and performing density calibration reconstruction processing on the density distribution information based on the obtained penetration resistance mechanical characteristic parameters to obtain density continuous distribution information; determining a bearing capacity reduction coefficient based on the density continuous distribution information; performing feature extraction processing on the electrical conductivity data to obtain a settled sediment activity index; inputting the settled sediment activity index and the bearing capacity reduction coefficient into a pile foundation settlement calculation model constructed by a neural network model to obtain pile foundation settlement data.

2. The pile settlement detection method for the progressive type of settlement according to claim 1, wherein The method of inputting the settled sediment activity index and the bearing capacity reduction coefficient into the pile foundation settlement calculation model constructed by the neural network model to obtain the pile foundation settlement data comprises: processing the settled sediment activity index and the bearing capacity reduction coefficient input into the pile foundation settlement calculation model by using a dual-path attention mechanism to determine a settled sediment adaptive weight and a bearing capacity adaptive weight; inputting the settled sediment activity index, the bearing capacity reduction coefficient, the settled sediment adaptive weight and the bearing capacity adaptive weight into the pile foundation settlement calculation model with the pile body elastic compression amount of the pile foundation as a constraint to obtain the pile foundation settlement data, wherein the pile body elastic compression amount is obtained from the pile length and the pile cross-sectional area of the pile foundation.

3. The method for pile settlement detection applied to the progressive type of siltation according to claim 1, characterized in that, The method of determining a bearing capacity reduction coefficient based on the density continuous distribution information comprises: processing the density continuous distribution information by using a Gaussian filter to obtain a smooth density curve; performing inflection point analysis on the smooth density curve to obtain density inflection point information; performing feature dimension reduction processing on the penetration resistance mechanical characteristic parameters according to a principal component analysis method to obtain mechanical characteristic information; performing inflection point analysis on the mechanical characteristic information by using a maximum value method to obtain mechanical inflection point information; obtaining the bearing capacity reduction coefficient based on the density inflection point information and the mechanical inflection point information.

4. The method for pile settlement detection applied to the progressive type of siltation according to claim 1, characterized in that, The method of differentially processing the gradual density data, processing the obtained density sensitive information based on acoustic impedance technology to obtain density distribution information comprises: performing noise reduction processing on the gradual density data by using a wavelet threshold technology to obtain purified density information; differentially processing the purified density information based on a transform transient frequency analysis method to obtain the density sensitive information; performing impedance inversion processing on the density sensitive information to obtain the density distribution information.

5. The method for pile settlement detection applied to the progressive type of siltation according to claim 4, characterized in that, The impedance inversion processing is performed on the density-sensitive information to obtain the density distribution information, including: The reflection coefficient conversion processing is performed on the density-sensitive information to obtain reflection coefficient sequence data; The reflection coefficient sequence data is processed by using the integral inversion technology with constraints to obtain acoustic impedance distribution data; The acoustic impedance distribution data is processed by using the wave velocity-density decoupling processing to obtain the density distribution information.

6. The pile settlement detection method for the progressive type of settlement according to claim 1, wherein The mechanical characteristic enhancement processing is performed on the penetration resistance data, and the density calibration reconstruction processing is performed on the density distribution information based on the obtained penetration resistance mechanical characteristic parameters to obtain continuous density distribution information, including: The mechanical characteristic enhancement processing is performed on the penetration resistance data by using the wavelet packet decomposition technology to obtain penetration resistance mechanical characteristic parameters; The state mapping processing is performed on the penetration resistance mechanical characteristic parameters to obtain an equivalent density field; The density calibration reconstruction processing is performed on the equivalent density field and the density distribution information by using the Bayesian fusion to obtain the continuous density distribution information.

7. The pile settlement detection method for the progressive type of settlement according to claim 1, wherein Before the profile graph is analyzed and identified, the pile foundation settlement detection method applied to the gradual change type sediment further includes: Obtaining scale data of a plurality of samples in the sediment; The scale data of the plurality of samples is input into a cutting model constructed by a neural network algorithm to determine the average particle size of the sediment; Obtaining a work log constructed by historical pile foundation settlement data; Using a large language model to analyze the work log to obtain a settlement interaction direction between the target sediment pit and the sediment.

8. The method for pile settlement detection applied to the progressive type of siltation according to claim 1, characterized in that, The profile graph is obtained by cutting a three-dimensional image of the sediment in a selected direction, and further includes: Performing fan profile analysis on the sediment with the center of the sediment pit as the origin; Based on the fan profile of the sediment, data acquisition processing is performed by using a sensor array to obtain initial echo signal data; Using synthetic aperture focusing technology, the initial echo signal data is processed by using an imaging algorithm to obtain the profile graph.

9. The method for pile settlement detection applied to the progressive type of siltation according to claim 3, characterized in that, Based on the density inflection point information and the mechanical inflection point information, the bearing capacity reduction coefficient is obtained, including: Quantitative fitting processing is performed on the density inflection point information to obtain a density critical gradient value; Using mechanical parameter correlation modeling technology, characteristic value extraction processing is performed on the mechanical inflection point information to obtain a mechanical critical threshold value; Based on the density critical gradient value and the mechanical critical threshold value, the bearing capacity reduction coefficient is determined.

10. A pile settlement detection system applied to a progressive type of settlement of silt, characterized by, Including: An acquisition module is configured to acquire initial density data, electrical conductivity data of sediment in a target sediment pit, and penetration resistance data of a pile foundation; An identification module is configured to analyze and identify a plurality of profile graphs, determine air gap information in the sediment based on all analysis and identification results, and correct the initial density data based on the air gap information to obtain gradual density data; A cutting module is configured to cut a three-dimensional image of the sediment in a selected direction to obtain a profile graph, wherein the direction is determined by a settlement interaction direction between the target sediment pit and the sediment, and an action scale of the cutting is greater than an average particle size of the sediment. A difference module is configured to perform differential processing on the gradual density data, process density-sensitive information obtained based on acoustic impedance technology, and obtain density distribution information; A reconstruction module is configured to perform mechanical characteristic enhancement processing on the penetration resistance data, perform density calibration reconstruction processing on the density distribution information based on obtained penetration resistance mechanical characteristic parameters, and obtain density continuous distribution information; A determination module is configured to determine a bearing capacity reduction coefficient based on the density continuous distribution information; An extraction module is configured to perform feature extraction processing on the electrical conductivity data, and obtain a sediment activity index; A generation module is configured to input the sediment activity index and the bearing capacity reduction coefficient into a pile foundation settlement calculation model constructed by a neural network model, and obtain pile foundation settlement data.

Images