Sheet-pile wall bearing capacity detection method and system

Abstract

The invention relates to the technical field of sheet-pile walls, in particular to a sheet-pile wall bearing capacity detection method and system.The method comprises the steps that material parameters and structure response data of all monitoring points are obtained, and material characteristic abnormal factors are calculated by integrating the material parameter gradient amplitude, the eight-neighborhood variable coefficient and the deviation with a design benchmark; constructing a local range by taking a monitoring point as a center, determining a weight by combining a structure response difference value and a gradient direction included angle, calculating a weighted variance, and calculating a structure response abnormal factor by fusing a response gradient amplitude, the weighted variance and a response level normalized difference value; when the two monitoring points both exceed a threshold value, screening the monitoring points as target monitoring points; and carrying out local outlier analysis on the target point based on an LOF algorithm, and identifying an abnormal bearing area. According to the method, the two-dimensional features of material performance degradation and structure response abnormity are fused, the sensitivity and recognition precision of local damage are improved, and intelligent detection of the bearing capacity of the slab-pile wall and potential disease area positioning are effectively achieved.

Description

Technical Field

[0001] This invention relates to the field of pile-slab wall technology, and in particular to a method and system for testing the bearing capacity of pile-slab walls. Background Technology

[0002] Pile-slab walls, as a common retaining structure, are widely used in slope reinforcement, roadbed support, and foundation pit engineering. Their safety and load-bearing capacity directly affect the overall stability and operational safety of the project. During long-term service, pile-slab walls may be affected by complex loads, environmental erosion, and material aging, leading to defects such as concrete cracking, steel corrosion, stress concentration, or deformation coordination failure, thereby weakening the structure's load-bearing capacity. Therefore, conducting scientific and effective load-bearing capacity testing and health status assessment of pile-slab walls has significant engineering application value.

[0003] Traditional load-bearing capacity testing methods often rely on single indicators, such as analyzing structural response data like stress, strain, or displacement. This makes it difficult to comprehensively reflect the coupling characteristics of material performance degradation and abnormal mechanical response within the structure. While some existing methods incorporate non-destructive testing techniques or finite element simulation, they still suffer from insufficient sensitivity to localized minor damage, low anomaly identification accuracy, and high false positive rates in practical monitoring. Furthermore, conventional statistical methods typically assume data follows a specific distribution, making it difficult to effectively capture the nonlinear and non-stationary evolution of structural responses. In recent years, anomaly detection algorithms based on multi-source monitoring data have been increasingly applied to the field of structural health monitoring in civil engineering. The Local Outlier Factor (LOF) algorithm has shown potential in structural anomaly identification due to its ability to identify outliers in low-density areas. However, directly applying the LOF algorithm to process raw monitoring data is susceptible to noise interference and lacks consideration of the co-evolution mechanism of material properties and structural response, resulting in poor interpretability and limited engineering applicability of the detection results.

[0004] The above content is only used to help understand the technical solution of the present invention and does not represent an admission that the above content is prior art. Summary of the Invention

[0005] The main objective of this invention is to provide a method and system for detecting the bearing capacity of pile-slab walls, aiming to solve the technical problem that existing methods for detecting the bearing capacity of pile-slab walls are unable to effectively integrate the synergistic characteristics of material parameter degradation and structural response anomalies, resulting in insufficient sensitivity to local damage, low accuracy in anomaly identification, and poor interpretability.

[0006] To achieve the above objectives, the present invention provides a method for testing the bearing capacity of a pile-slab wall, the method comprising:

[0007] Obtain the material parameters and structural response data corresponding to each monitoring point in the monitoring data of the pile-slab wall structure; calculate the material property anomaly factor of the monitoring point based on the gradient magnitude of the material parameters of the monitoring point, the coefficient of variation in the eight neighborhoods of the monitoring point, and the difference between the material parameters of the monitoring point and the design reference parameters.

[0008] A local range of the monitoring point is constructed with the monitoring point as the center. Based on the absolute value of the structural response difference between the monitoring point and other monitoring points in its local range and the gradient direction angle, the weighting coefficient of other monitoring points in the local range to the monitoring point is calculated, and the weighted variance of the structural response in the local range of the monitoring point is obtained.

[0009] The structural response anomaly factor of the monitoring point is calculated based on the gradient magnitude of the structural response data at the monitoring point, the weighted variance of the structural response in the local area of ​​the monitoring point, and the difference between the structural response data at the monitoring point and the minimum value of the structural response in the monitoring data of the pile-slab wall structure.

[0010] The target monitoring point is obtained when both the material property anomaly factor and the structural response anomaly factor of the monitoring point are greater than the corresponding threshold. The target monitoring point is then processed based on the LOF algorithm to obtain the load-bearing capacity test result of the pile-slab wall.

[0011] Optionally, the acquisition of material parameters and structural response data corresponding to each monitoring point in the pile-slab wall structure monitoring data includes:

[0012] After applying graded loads to the pile-slab wall, material parameter datasets and structural response datasets are collected from the structural monitoring data. The material parameter datasets and structural response datasets are preprocessed to obtain the material parameters and structural response data corresponding to each monitoring point. The preprocessing method includes at least normalizing the material parameter datasets and structural response datasets. The material parameters include the elastic modulus of concrete and the yield strength of steel bars, and the structural response data includes stress, strain, and displacement.

[0013] Optionally, calculating the material property anomaly factor at the monitoring point includes:

[0014] Obtain the design material parameter data of the pile-slab wall and perform normalization processing to obtain the material reference parameters; calculate the material property anomaly factor at the x-th monitoring point. :

[0015] ;

[0016] in, Let x be the gradient magnitude of the material parameters at the x-th monitoring point. , , Let $x$ be the standard deviation, mean, and coefficient of variation of the material parameters in the eight neighborhood of the $x$-th monitoring point, respectively. , Let be the measured material parameters and the material reference parameters at the x-th monitoring point, respectively. ln() is the logarithmic function with base e, || is the absolute value sign, and norm() is the linear normalization function.

[0017] Optionally, constructing the local range of the monitoring point with the monitoring point as the center includes:

[0018] The local range of a preset monitoring point is n×n. Taking the monitoring point as the center, n×n-1 spatially adjacent monitoring points are obtained around the monitoring point to construct the local range of the monitoring point.

[0019] Optionally, calculating the weighting coefficients of other monitoring points within the local area for that monitoring point includes:

[0020] ;

[0021] in, Let a be the weighting coefficient of the a-th monitoring point within the local range of the x-th monitoring point, and let a be the weighting coefficient of the a-th monitoring point within the x-th monitoring point. Let be the cosine of the angle between the x-th monitoring point and the a-th monitoring point within the local range of the x-th monitoring point. Let x be the absolute value of the stress difference between the x-th monitoring point and the a-th monitoring point. , These represent the maximum and minimum stress values ​​in the monitoring data of the pile-slab wall structure, respectively.

[0022] Optionally, calculating the structural response anomaly factor at the monitoring point includes:

[0023] Calculate the structural response anomaly factor at the x-th monitoring point. :

[0024] ;

[0025] Let x be the gradient magnitude of the stress value at the x-th monitoring point. Let x be the stress-weighted variance within a local area of ​​the x-th monitoring point. Let x be the stress value at the x-th monitoring point. , ...

[0026] Optionally, the step of processing the target monitoring points based on the LOF algorithm to obtain the pile-slab wall bearing capacity test results includes:

[0027] The comprehensive score of the target monitoring point is obtained based on the material property anomaly factor and the structural response anomaly factor of the target monitoring point.

[0028] Obtain the comprehensive scores of all target monitoring points in descending order, and select a preset number of target monitoring points as candidate points from the comprehensive score descending order;

[0029] Calculate the LOF value of the candidate points, calculate the anomaly threshold based on the mean and standard deviation of the LOF values ​​of all candidate points, and obtain the abnormal bearing points in the pile-slab wall bearing capacity test results based on the comparison between the LOF values ​​of the candidate points and the anomaly threshold.

[0030] Optionally, the calculation of the LOF value of the candidate point includes:

[0031] The candidate point is located at a distance of K from its neighboring points in the neighborhood. The local density of the candidate point is obtained, and the average of the ratios of the local densities of the neighboring points to that of the candidate point is recorded as the LOF value of the candidate point.

[0032] Optionally, after obtaining the test results of the pile-slab wall bearing capacity, the method further includes:

[0033] Multiple clusters are obtained by clustering the abnormal load-bearing points, and the mean values ​​of the material property anomaly factor and the mean value of the structural response anomaly factor of the abnormal load-bearing points in the clusters are obtained.

[0034] If the mean value of the material property anomaly factor of a cluster is greater than the mean value of the structural response anomaly factor, then the anomaly of the cluster is a material defect; if the mean value of the material property anomaly factor of a cluster is less than the mean value of the structural response anomaly factor, then the anomaly of the cluster is a structural stress anomaly; if the mean value of the material property anomaly factor of a cluster is equal to the mean value of the structural response anomaly factor, then the anomaly of the cluster is a combination of material defects and structural stress anomalies.

[0035] Furthermore, to achieve the above objectives, the present invention also provides a pile-slab wall bearing capacity testing system, the system comprising:

[0036] processor;

[0037] Memory used to store the processor's executable instructions;

[0038] The processor is configured to execute the instructions to implement the steps of the pile-slab wall bearing capacity detection method as described in any of the above-mentioned methods.

[0039] This invention provides a method for detecting the bearing capacity of pile-slab walls. The method constructs a two-dimensional anomaly identification mechanism by comprehensively analyzing the multidimensional characteristics of material parameters and structural response. This mechanism includes both material property anomaly factors and structural response anomaly factors, effectively capturing potential damage in pile-slab walls caused by material performance degradation (such as insufficient concrete strength and steel corrosion) and abnormal mechanical behavior (such as stress concentration and deformation incoordination). By introducing spatial evolution characteristics such as gradient amplitude, coefficient of variation, response difference, and gradient direction angle within the neighborhood of the monitoring point, the sensitivity to subtle local changes is enhanced. Weighted variance and normalized nonlinear functions are used to improve the discrimination ability of anomaly factors, overcoming the problems of strong dependence on single indicators and weak anti-interference ability of traditional methods. Furthermore, the LOF algorithm is combined to perform local density analysis on high-risk target points, achieving accurate identification of anomalies and quantification of outlier degree. Ultimately, this method not only improves the accuracy and reliability of bearing capacity assessment but also allows for the identification of anomaly causes through subsequent cluster analysis, providing strong technical support for the diagnosis and maintenance decisions of pile-slab walls. It has good engineering practicality and promotional value. Attached Figure Description

[0040] Figure 1 This is a schematic flowchart of an embodiment of the pile-slab wall bearing capacity testing method of the present invention.

[0041] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0042] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0043] Reference Figure 1 , Figure 1 This is a flowchart illustrating an embodiment of the pile-slab wall bearing capacity testing method of the present invention, which presents an embodiment of the pile-slab wall bearing capacity testing method of the present invention.

[0044] In one embodiment, the method for testing the bearing capacity of a pile-slab wall includes the following steps:

[0045] Step S100: Obtain the material parameters and structural response data corresponding to each monitoring point in the pile-slab wall structure monitoring data. Calculate the material property anomaly factor for the monitoring point based on the gradient magnitude of the material parameters at the monitoring point, the coefficient of variation in the eight neighborhoods of the monitoring point, and the difference between the material parameters of the monitoring point and the design reference parameters.

[0046] The monitoring data for the pile-slab wall structure can be a multi-source data set reflecting the service status of the pile-slab wall, collected by sensors or detection equipment, and can be used to provide raw input for material parameter and structural response analysis. In this embodiment, the monitoring data for the pile-slab wall structure can be collected in real time or periodically by sensing devices such as strain gauges, displacement gauges, rebar corrosion probes, and ultrasonic strength meters deployed on the pile-slab wall. Each monitoring point can be a preset discrete observation location with spatial coordinates on the pile-slab wall structure, which can be used as a spatial sampling unit for material parameter and structural response data to support local feature calculation. For example, each monitoring point can include, but is not limited to, one or more of the following: monitoring point at the middle of the pile, monitoring point at the top of the pile, and monitoring point at the connection of the retaining plate.

[0047] Material parameters can be quantitative indicators characterizing the physical and mechanical properties of the materials constituting the pile-slab wall, reflecting material degradation states such as concrete strength and steel reinforcement corrosion. Further, material parameters can include, but are not limited to, one or more of concrete compressive strength, steel reinforcement corrosion rate, and elastic modulus. Structural response data can be measured values ​​of the mechanical behavior of the pile-slab wall under load or environmental action, reflecting the actual working state of the structure, such as stress, strain, and displacement. In an exemplary embodiment, structural response data can include, but is not limited to, one or more of horizontal displacement, bending moment, and axial strain. The gradient magnitude of the material parameters at the monitoring point can be the absolute value of the rate of change of the material parameters at the monitoring point and its neighboring points in space, used to identify regions of abrupt changes in material properties and indicate potential degradation boundaries. In this embodiment, the gradient magnitude of the material parameters at the monitoring point can be calculated by the spatial difference between the material parameters at the monitoring point and its neighboring areas.

[0048] The eight-neighborhood of a monitoring point can be a set of eight spatial locations directly adjacent to the current monitoring point in a two-dimensional grid. This set can be used to define a local spatial range and to calculate local statistical characteristics such as the coefficient of variation. Furthermore, the eight-neighborhood of a monitoring point can work in conjunction with the coefficient of variation to provide a local sample set for assessing material homogeneity. The coefficient of variation can be the ratio of the standard deviation to the mean of the material parameters within the eight-neighborhood, and can be used to measure the dispersion of local material properties, reflecting homogeneity degradation. In one specific embodiment, the coefficient of variation can be obtained by calculating the ratio of the standard deviation and mean of the material parameters within the eight-neighborhood. The design benchmark parameters can be the standard values ​​of material properties specified during the pile-slab wall design phase, and can be used as a reference benchmark for the degree of deviation from the service condition. For example, design benchmark parameters can include, but are not limited to, one or more of the following: design concrete strength grade, design rebar diameter, and design protective layer thickness. The material characteristic anomaly factor can be a comprehensive index that integrates gradient amplitude, the eight-neighborhood coefficient of variation, and deviation from the design benchmark parameters, and can be used to quantify the degree of anomaly caused by material performance degradation. In this embodiment, the material characteristic anomaly factor can be generated by weighted combination of the above three types of features and processed by a normalized nonlinear function. Furthermore, material property anomaly factors can be used together with structural response anomaly factors to screen target monitoring points.

[0049] Obtaining material parameters and structural response data corresponding to each monitoring point in the pile-slab wall structure monitoring data can be achieved by extracting the material and structural response data of each monitoring point from the monitoring system database or real-time sensor network. Furthermore, this operation can be implemented by synchronously collecting and uploading data to a cloud platform via a wireless sensor network, or by batch exporting data from a historical monitoring database by timestamp, thereby establishing a basic dataset for multi-dimensional feature analysis. Based on the gradient magnitude of the material parameters at the monitoring point, the coefficient of variation in the eight neighborhoods of the monitoring point, and the difference between the material parameters of the monitoring point and the design baseline parameters, the material characteristic anomaly factor of the monitoring point is calculated. This can be achieved by inputting three types of features into a weighted model and outputting a single anomaly factor value through a normalized nonlinear function. In an exemplary embodiment, this operation can be achieved by using an exponential decay function to nonlinearly amplify the deviation term, or by using a sigmoid function to saturate the gradient magnitude to suppress noise, thereby enabling multi-dimensional fusion characterization of material degradation.

[0050] Step S200: Construct a local range of the monitoring point with the monitoring point as the center. Calculate the weighting coefficients of the other monitoring points in the local range to the monitoring point based on the absolute value of the structural response difference between the monitoring point and other monitoring points in the local range and the gradient direction angle. Obtain the weighted variance of the structural response in the local range of the monitoring point.

[0051] The local range can be a spatial neighborhood centered on a certain monitoring point, used for structural response analysis. It can be used to define the calculation area for the weighted variance of the structural response, reflecting the local correlation of the mechanical field. For example, the local range can include, but is not limited to, one or more of the following: a 3×3 monitoring point window, a circular area with a 5-meter radius, or an interval of ±2 monitoring points along the pile axis. The absolute value of the structural response difference can be the absolute value of the difference between the structural response values ​​of the target monitoring point and other monitoring points within the local range. It can be used to reflect the degree of difference in local mechanical response and for weight allocation. In this embodiment, the absolute value of the structural response difference can be obtained by calculating the difference between the structural response data of the target point and the neighboring points point by point and then taking the absolute value.

[0052] The gradient direction angle can be the spatial angle between the structural response gradient vectors of the target monitoring point and its neighboring points. It can be used to measure the consistency of the local mechanical field direction and assist in judging deformation compatibility. Furthermore, the gradient direction angle can be obtained by calculating the inverse cosine of the ratio of the dot product of the gradient vectors of two points to their magnitudes. The weighting coefficient can be the contribution ratio of each neighboring monitoring point to the weighted variance of the target point's structural response within a local range. It can be used to make the weighted variance more closely match the spatial distribution characteristics of the actual mechanical field. In a specific embodiment, the weighting coefficient can be generated based on the absolute value of the structural response difference and the gradient direction angle through a normalized nonlinear function mapping. The structural response weighted variance can be the variance of the structural response data within a local range calculated using the weighting coefficient. It can be used to characterize the discreteness of the local mechanical response and enhance sensitivity to subtle anomalies. In this embodiment, the structural response weighted variance can be calculated using the weighting coefficient as weights on the local structural response data.

[0053] Constructing a local range around a monitoring point can be achieved by determining a set of neighboring monitoring points based on preset spatial rules (such as a fixed radius or a fixed number of points). Furthermore, this operation can be implemented by using Manhattan distance to select the N nearest monitoring points, or by defining all monitoring points within a fixed radius using Euclidean distance, thereby limiting the spatial scale of structural response analysis and reflecting local mechanical correlations.

[0054] Based on the absolute value of the structural response difference between a monitoring point and other monitoring points within its local area, and the angle between their gradient directions, calculate the weighting coefficients of the other monitoring points within that local area. This can be achieved by mapping the absolute value of the difference to the angle, with smaller differences and angles generally resulting in higher weights. Obtain the weighted variance of the structural response within the local area of ​​that monitoring point, which can be calculated using the weighting coefficients as weights.

[0055] Step S300: Calculate the structural response anomaly factor of the monitoring point based on the gradient magnitude of the structural response data at the monitoring point, the weighted variance of the structural response in the local area of ​​the monitoring point, and the difference between the structural response data at the monitoring point and the minimum value of the structural response in the monitoring data of the pile-slab wall structure.

[0056] The gradient magnitude of the structural response data can be the absolute value of the rate of change of the structural response at the monitoring point in space, which can be used to identify areas of stress or deformation concentration. In this embodiment, the gradient magnitude of the structural response data can be calculated by the spatial difference between the monitoring point and its neighboring structural response data. The minimum structural response value in the pile-slab wall structure monitoring data can be the minimum value of the structural response data of all monitoring points, which can be used as a benchmark reference for calculating the structural response anomaly factor, reflecting the overall lowest response level. The structural response anomaly factor can be a comprehensive index that integrates the structural response gradient magnitude, weighted variance, and difference from the global minimum value, which can be used to quantify the degree of abnormal mechanical behavior and capture stress concentration or deformation incoordination. Furthermore, the structural response anomaly factor can be generated by weighted combination of three types of features and processed by a normalized nonlinear function. In a specific embodiment, the structural response anomaly factor can participate in the target monitoring point screening together with the material property anomaly factor.

[0057] Based on the gradient magnitude of the structural response data at the monitoring point, the weighted variance of the structural response within the local area of ​​the monitoring point, and the difference between the structural response data at the monitoring point and the minimum structural response value in the pile-slab wall structure monitoring data, the structural response anomaly factor at the monitoring point is calculated. This can be achieved by weighting and fusing the three types of features and outputting a single anomaly factor through a normalized nonlinear function. For example, this operation can be implemented by taking the logarithm of the difference from the minimum value to compress the dynamic range, or by smoothing the gradient magnitude using a moving average before participating in the calculation, thus comprehensively reflecting the multidimensional characteristics of abnormal mechanical behavior.

[0058] In step S400, the target monitoring point is obtained when both the material property anomaly factor and the structural response anomaly factor of the monitoring point are greater than the corresponding thresholds. The target monitoring point is then processed using the LOF algorithm to obtain the pile-slab wall bearing capacity test results.

[0059] The corresponding threshold can be a pre-set judgment boundary between material property anomaly factors and structural response anomaly factors, which can be used to screen high-risk monitoring points and control the false alarm rate. In an exemplary embodiment, the corresponding threshold may include, but is not limited to, one or more of the following: material anomaly threshold, structural response anomaly threshold, and dynamic adaptive threshold. The target monitoring point can be a monitoring point where both the material property anomaly factor and the structural response anomaly factor exceed the corresponding threshold, which can be used as the input object of the LOF algorithm to limit high-risk areas for detailed analysis. In this embodiment, the target monitoring point can be obtained through a two-factor joint judgment logic. The LOF algorithm can be a Local Outlier Factor algorithm, used to identify outliers with local density significantly lower than their neighbors, which can be used to perform local density analysis on the target monitoring point and quantify its outlier degree. Furthermore, the LOF algorithm can calculate the local reachability density based on the reachability distance between the target point and its k nearest neighbors, and then obtain the LOF value.

[0060] The bearing capacity detection results can be the output of pile-slab wall anomalies and their outlier quantification values ​​after processing by the LOF algorithm, which can be used to provide a basis for bearing capacity assessment and disease diagnosis. Determining whether both the material property anomaly factor and the structural response anomaly factor of the monitoring point are greater than the corresponding threshold can be done by performing a logical AND judgment; only when both conditions are met simultaneously is it considered an anomaly. In a specific embodiment, this operation can use a fixed threshold for hard decision-making, or dynamically adjust the threshold based on historical data (such as the 3σ principle), thereby reducing false positives caused by misjudgment of a single indicator. Obtaining the target monitoring point can be achieved by marking the monitoring points that meet the dual-factor threshold conditions as target monitoring points. Furthermore, this operation can narrow the scope of subsequent LOF analysis, improving computational efficiency and noise resistance. Processing the target monitoring point based on the LOF algorithm can be done by performing LOF calculation within the target monitoring point set and outputting the outlier factor of each point. For example, this operation can be performed within the target point and its original neighborhood, or only by constructing a k-nearest neighbor graph within the target point set for LOF calculation, thereby identifying local density anomalies in high-risk subsets and improving anomaly location accuracy. The bearing capacity test results for pile-slab walls can be obtained by identifying target points whose output LOF values ​​exceed a set outlier threshold as final outlier points. Furthermore, this operation can provide interpretable and quantifiable assessment results of bearing capacity degradation.

[0061] Taking long-term health monitoring of highway slope pile-slab walls as an example, the bearing capacity testing method for pile-slab walls in this embodiment can be to set up 50 monitoring points on a 10-year-old highway slope pile-slab wall, simultaneously collecting concrete strength (material parameters) and horizontal displacement (structural response). The system calculates the material property anomaly factor (integrating strength gradient, eight-neighbor coefficient of variation, and deviation from C30 design strength) and structural response anomaly factor (integrating displacement gradient, weighted variance, and difference from the global minimum displacement) for each point. It was found that monitoring point No. 23 exceeded the threshold for both factors and was listed as a target monitoring point. Subsequently, the LOF algorithm was applied to all 8 target monitoring points, and the LOF value of point No. 23 was found to be significantly higher than that of other points, and it was determined to be a local outlier. Combined with subsequent cluster analysis, it was confirmed that this point is located at the joint of the retaining slab, and there is coupled damage of concrete cracking and steel corrosion, so priority maintenance is recommended.

[0062] In one embodiment, the material parameters and structural response data corresponding to each monitoring point in the pile-slab wall structure monitoring data are obtained, including:

[0063] After applying graded loads to the pile-slab wall, material parameter datasets and structural response datasets are collected from the structural monitoring data. The material parameter datasets and structural response datasets are preprocessed to obtain the material parameters and structural response data corresponding to each monitoring point. The preprocessing method includes at least normalization of the material parameter datasets and structural response datasets. The material parameters include the elastic modulus of concrete and the yield strength of steel bars, and the structural response data includes stress, strain, and displacement.

[0064] The graded load can be a sequence of external loads applied to the pile-slab wall in a predetermined manner, which can be used to systematically stimulate the material and mechanical responses of the structure under different stress states and obtain monitoring data with clear physical meaning. In an exemplary embodiment, the graded load can be one or more of the following: static graded loading, cyclic graded loading, and monotonically increasing graded loading. The material parameter dataset can be a set of original multidimensional data reflecting material properties collected under graded loads, which can be used as the original input for calculating material property anomaly factors, including indicators such as the elastic modulus of concrete and the yield strength of steel bars. Furthermore, the material parameter dataset can be collected synchronously under each level of load using devices such as ultrasonic sensors, rebound hammers, and electrochemical sensors. The structural response dataset can be a set of original multidimensional data reflecting the mechanical behavior of the structure collected under graded loads, which can be used as the original input for calculating structural response anomaly factors, including indicators such as stress, strain, and displacement. For example, the structural response dataset can be recorded synchronously under each level of load using strain gauges, displacement gauges, and fiber optic grating sensors.

[0065] The elastic modulus of concrete can be the ratio of stress to strain in the elastic stage of concrete, characterizing its stiffness characteristics and reflecting the degree of aging or damage. In a specific embodiment, the elastic modulus of concrete may include, but is not limited to, one or more of the following: static elastic modulus, dynamic elastic modulus, and secant elastic modulus. The yield strength of reinforcing steel can be the stress value corresponding to the onset of plastic deformation in the reinforcing steel, and can be used to characterize the degradation state of the reinforcing steel's load-bearing capacity and identify corrosion or fatigue damage. Further, the yield strength of reinforcing steel may include, but is not limited to, one or more of the following: upper yield strength, lower yield strength, and nominal yield strength (0.2% residual strain). Stress can be the intensity of internal forces per unit area, reflecting the stress state inside the structure and can be used to identify areas of stress concentration. Strain can be the relative deformation of a structure under load, and can be used to characterize the degree of local deformation and assist in determining cracking or yielding states. Displacement can be the change in position of a structural point in space, and can be used to reflect overall or local stiffness degradation and assess deformation compatibility.

[0066] After applying graded loads to the pile-slab wall, the material parameter dataset and structural response dataset in the structural monitoring data are collected. This can be achieved by loading the pile-slab wall step by step according to preset load levels in the laboratory or on-site, and simultaneously recording the raw material and structural response data under each load level. Furthermore, this operation can be achieved by using hydraulic jacks to apply loads in 5% design load steps and collecting data while maintaining the load, or by using servo actuators to apply cyclic graded loads and collect steady-state data after each peak value. This allows for the acquisition of high-quality data with a clear load-response correspondence, enhancing the ability to capture nonlinear evolution patterns. Preprocessing the material parameter dataset and structural response dataset can involve performing a standardization process on the raw collected data, including denoising, interpolation, alignment, and normalization. For example, this operation can be achieved by first performing moving average filtering for denoising, then performing Min-Max normalization, or by using spline interpolation to fill missing values ​​and then performing Z-score standardization. This improves data quality and consistency, providing reliable input for subsequent anomaly factor calculations.

[0067] Normalizing the material parameter dataset and structural response dataset can be achieved by mapping data of different physical quantities to the range [0, 1] or a standard normal distribution. In a specific embodiment, this operation can be accomplished by using Min-Max normalization to [0, 1] for the elastic modulus of concrete and the yield strength of steel bars, and by uniformly using Z-score normalization to make the mean 0 and the standard deviation 1 for stress, strain, and displacement. This eliminates differences in dimensions and magnitudes, ensuring the comparability of the calculation of features such as gradient magnitude and weighted variance. Obtaining the material parameter and structural response data corresponding to each monitoring point can be achieved by organizing the preprocessed data according to the spatial index of the monitoring points to form a structured data table. This allows for the establishment of an input format compatible with the original scheme steps, supporting the subsequent calculation of two-dimensional anomaly factors.

[0068] Taking the bearing capacity verification test of a newly constructed pile-slab wall as an example, the bearing capacity testing method of this embodiment can be to apply five levels of static loading (0→30%→60%→90%→100% of the design load) to the newly constructed pile-slab wall during the acceptance stage of a railway subgrade support project. Each loading level is maintained for 10 minutes, and the concrete elastic modulus (ultrasonic method), steel yield strength (electrochemical impedance spectroscopy), stress (fiber grating), strain (resistance strain gauge), and displacement (total station) are simultaneously collected at 50 monitoring points. After Z-score normalization, the raw data are reorganized according to the monitoring point number and input into step S100 of the original scheme. Because the data characterizes the material-response co-evolution at a unified scale, the subsequently calculated material property anomaly factor can effectively identify the low elastic modulus of pile No. 18 due to insufficient curing, while the structural response anomaly factor simultaneously captures the nonlinear displacement increase that occurs at this point after 80% load, ultimately being identified as a high-risk point by the LOF algorithm.

[0069] This embodiment provides a method for detecting the bearing capacity of a pile-slab wall. After applying graded loads to the pile-slab wall, material parameter datasets and structural response datasets are collected from structural monitoring data. These datasets are then preprocessed to obtain material parameter and structural response data corresponding to each monitoring point. The preprocessing method includes at least normalizing the material parameter and structural response datasets. By applying graded loads under controlled conditions to actively stimulate the material properties and mechanical responses of the pile-slab wall under different stress states, material parameter and structural response datasets with clear physical meaning and covering normal to critical states are obtained. Normalization preprocessing of the original datasets eliminates the dimensional and magnitude differences between multi-source heterogeneous data, providing high-quality, standardized input for two-dimensional anomaly identification and enhancing the characterization ability of the material-mechanical coupling degradation process. Simultaneously, the graded loading strategy effectively captures the nonlinear response evolution law, fundamentally improving the technical effect of screening high-risk target monitoring points.

[0070] In one embodiment, calculating the material property anomaly factor at the monitoring point includes:

[0071] Obtain the design material parameter data of the pile-slab wall and perform normalization processing to obtain the material reference parameters; calculate the material property anomaly factor at the x-th monitoring point. :

[0072] ;

[0073] in, Let x be the gradient magnitude of the material parameters at the x-th monitoring point. , , Let $x$ be the standard deviation, mean, and coefficient of variation of the material parameters in the eight neighborhood of the $x$-th monitoring point, respectively. , Let be the measured material parameters and the material reference parameters at the x-th monitoring point, respectively. ln() is the logarithmic function with base e, || is the absolute value sign, and norm() is the linear normalization function.

[0074] The material parameter data for the pile-slab wall design can be a set of material performance indicators specified in the original design documents of the pile-slab wall. This data can be used as the basis for normalization processing to generate material reference parameters of a uniform scale. In an exemplary embodiment, the material parameter data for the pile-slab wall design can be extracted from engineering design drawings, material specifications, or BIM models. The material reference parameter can be a standardized reference value E0 obtained after normalizing the material parameter data for the pile-slab wall design, which can be used as a measured material parameter. The comparison benchmark eliminates the dimensional differences between different material types. Furthermore, the material benchmark parameter can be obtained by linearly normalizing the original design material parameters (such as concrete strength, modulus of elasticity, etc.) to their maximum value or standard value. In a specific embodiment, the material benchmark parameter can be directly normalized to 1.0 according to the standard value (e.g., 30MPa for C30 concrete), or it can be obtained by weighted summation of multiple material parameters (strength, modulus, etc.) after separate normalization. The xth monitoring point can be a specific spatial observation location numbered x in the pile-slab wall structure monitoring network, and can be used as a material property anomaly factor. The calculation unit. For example, the xth monitoring point may be one or more of the following, including but not limited to the monitoring point at the top of the pile, the monitoring point in the middle of the retaining plate, and the monitoring point at the pile-slab connection node.

[0075] Material property anomaly factor It can be by , and The comprehensive anomaly index, obtained by normalizing the product of the three factors using a linear normalization function, can be used to quantify the overall degree of material performance degradation at the x-th monitoring point. In a specific embodiment, the material characteristic anomaly factor... It can be done through formula = Calculations yielded the following: Further, the material property anomaly factor... It can be used as a material-side input in a two-dimensional anomaly identification mechanism, and can be used in conjunction with structural response anomaly factors to determine target monitoring points. (The gradient magnitude of the material parameter at the x-th monitoring point) can be the absolute value of the maximum rate of change of the material parameter at the x-th monitoring point within its spatial neighborhood. It can be used to characterize the degree of spatial abrupt change in material properties, indicating localized deterioration such as cracks and corrosion boundaries. In an exemplary embodiment, The gradient magnitude can be calculated based on the difference vector magnitude between point x and its eight neighboring material parameters. Furthermore, the gradient magnitude of the material parameters at the x-th monitoring point can be calculated. The two-dimensional gradient magnitude can be approximated by using the Sobel operator, or by taking the maximum value of the absolute value of the eight-directional difference as... This allows us to capture the spatially abrupt changes in material properties.

[0076] (Standard deviation of material parameters in the eight neighborhood of the xth monitoring point) can be a measure of the dispersion of material parameter values ​​in the eight neighborhood of the xth point. It can be used to reflect local material performance fluctuations and help assess uniformity degradation. (The mean of material parameters in the eight neighborhoods of the xth monitoring point) can be the arithmetic mean of the material parameters in the eight neighborhoods of the xth point. It can be used as the denominator in the calculation of the coefficient of variation to characterize the average level of local material performance. The coefficient of variation (CV) can be the ratio of the standard deviation to the mean of the material parameters in the eight neighborhoods. It can be used as a dimensionless measure of local material homogeneity and is sensitive to small inhomogeneities. Further, the CV in the eight neighborhoods of the x-th monitoring point is calculated. It can be done when By introducing a smoothing term to prevent division by zero when the coefficient of variation is close to zero, or by imposing an upper limit truncation on the coefficient of variation to suppress extreme noise, a dimensionless local discreteness index can be obtained. In this embodiment, Clearly participate as a multiplicative factor Construct and strengthen its coupling effect with gradients and biases.

[0077] (Measured material parameters at the xth monitoring point) can be measured material properties at point x obtained through on-site non-destructive testing (such as rebound strength, elastic modulus calculated by ultrasonic velocity, etc.), which can be used to reflect the material properties under actual service conditions. (Material reference parameters) can be normalized design material parameter reference values, which can be used to calculate relative deviations and eliminate the influence of absolute dimensions. In this embodiment, It is clearly derived from the normalization of design parameters, rather than directly using the original design values. It can be a nonlinear function output that applies a natural logarithmic transformation to the relative deviation between measured and reference material parameters. It can be used to nonlinearly amplify small deviations, suppress the dominant effect of large deviations, and improve anomaly resolution. The linear normalization function can be a linear scaling function that maps composite anomaly indicators to [0, 1] or a fixed interval. It can be used to unify the dimensions of anomaly factors, facilitating the fusion of multiple indicators and threshold setting.

[0078] Obtain the material parameter data for pile-slab wall design and normalize it to obtain material reference parameters. This can be done by extracting original material parameters from design documents and converting them to a unified reference through linear scaling. Furthermore, this operation can be performed by directly normalizing to 1.0 according to the standard value (e.g., 30MPa for C30 concrete), or by normalizing multiple material parameters (strength, modulus, etc.) separately and then weighting them to synthesize a single value. This allows for the elimination of dimensional differences caused by different material types or unit systems, establishing a comparable benchmark. The gradient magnitude of the material parameter at the x-th monitoring point is calculated. This can be based on the spatial difference between point x and the material parameters of its eight neighbors, taking the magnitude of the gradient vector. Furthermore, this operation can be performed by approximating the two-dimensional gradient magnitude using the Sobel operator, or by taking the maximum absolute value of the eight-directional differences as... This allows us to capture the spatial abrupt changes in material properties.

[0079] For example, in the scenario of evaluating the service performance of pile-slab walls in mountainous railway subgrades, the pile-slab wall bearing capacity testing method in this embodiment could be: A railway pile-slab wall has been in service for 8 years, with 40 monitoring points set up. The system obtains the C35 concrete strength (35MPa) and the steel reinforcement elastic modulus (200GPa) from the design documents, and after normalization, obtains... =1.0. The value at point 15 was measured using the on-site ultrasonic rebound method. =0.68. Its eight-neighbor coefficient of variation is calculated to be 0.22, and the gradient magnitude is... =0.15. Relative deviation |0.68−1.0| / 1.0=0.32, after ln(1+0.32)=0.278. Composite index=0.15×0.22×0.278≈0.0092. The composite index of the entire point set after normalization by the linear normalization function, =0.87, exceeding the material anomaly threshold of 0.75. Combined with the structural response anomaly factor also exceeding the standard, this point was listed as the target for LOF analysis. Ultimately, it was confirmed that steel reinforcement rust expansion caused concrete spalling, verifying the high sensitivity of this method to coupled damage.

[0080] This embodiment provides a method for detecting the bearing capacity of pile-slab walls. The method obtains material baseline parameters by acquiring and normalizing the design material parameter data of the pile-slab wall, and then calculates the material property anomaly factor at the x-th monitoring point. By multiplicatively coupling spatial gradient amplitude, local coefficient of variation, and logarithmic relative deviation to form a composite anomaly index, and then outputting a standardized factor through linear normalization, it is possible to fuse three pieces of information: spatial abrupt change characteristics of material properties, local homogeneity deterioration, and degree of deviation from design benchmarks. High response is triggered only when the three types of anomalies co-occur, effectively suppressing misjudgments caused by single features, and outputting a standardized anomaly factor with cross-point comparability. This provides a highly robust and interpretable material-side input for the two-dimensional anomaly identification mechanism, significantly improving the accuracy and engineering applicability of pile-slab wall bearing capacity degradation detection.

[0081] In one embodiment, constructing a local range of the monitoring point centered on the monitoring point includes:

[0082] The local range of a preset monitoring point is n×n. Taking the monitoring point as the center, n×n-1 spatially adjacent monitoring points are obtained around the monitoring point to construct the local range of the monitoring point.

[0083] The n×n local area size can be the side length parameter of a square local neighborhood constructed with the monitoring point as the center. It represents the grid size containing n monitoring points in both the row and column directions, and can be used to control the spatial scale of the local area, balancing sensitivity to local details and noise resistance. Furthermore, the n×n-1 spatially adjacent monitoring points can be the set of all neighboring monitoring points remaining in the n×n grid after removing the central monitoring point. This can be used to form the neighborhood sample set of the local area for calculating spatial evolution characteristics such as the weighted variance of the structural response. In a specific embodiment, the n×n-1 spatially adjacent monitoring points can be a fixed number of neighboring points systematically selected in the directions above, below, left, right, and diagonally opposite the central monitoring point, according to a preset value of n.

[0084] For example, spatially adjacent monitoring points can be other monitoring points that are physically adjacent to the target monitoring point in the pile-slab wall monitoring point grid layout. These can provide reference information on local mechanical or material states, supporting spatial gradient and variation analysis. Furthermore, spatially adjacent monitoring points can include, but are not limited to, one or more of four-neighbor monitoring points, eight-neighbor monitoring points, and extended cross-neighbor monitoring points. The preset local range size of the monitoring point is n×n, which can be achieved by setting the grid dimension parameter n of the local neighborhood during the algorithm initialization phase, forming an n×n square analysis window. Furthermore, the preset local range size of the monitoring point is n×n, which can be achieved by specifying the n value (e.g., n=3) through a configuration file or by adaptively selecting the minimum n based on the monitoring point density to ensure that the neighborhood contains at least valid data points. This allows for the establishment of a unified, structured spatial analysis unit, ensuring the consistency and reproducibility of subsequent feature calculations.

[0085] Centered on a monitoring point, a local range is constructed by acquiring n×n-1 spatially adjacent monitoring points around that monitoring point. This can be based on a pre-defined n×n grid, extracting all neighboring monitoring points (excluding the center point) as members of the local range within the monitoring point's coordinate system. Furthermore, this local range can be achieved by directly indexing neighboring points by row and column offsets in a regularly arranged two-dimensional monitoring grid, or by selecting n×n−1 points based on the Euclidean distance principle to simulate an n×n structure in the case of irregular point arrangement. This forms a neighborhood set with clear topological relationships, providing a computational basis for spatial features such as weighted variance and gradient direction angles.

[0086] Taking the anomaly detection of pile-slab walls in foundation pit support as an example, the pile-slab wall bearing capacity detection method in this embodiment can be as follows: In a deep foundation pit support project, monitoring points are arranged on the surface of the pile-slab wall at 2-meter intervals to form an approximately regular grid. The system presets the local range size to be 5×5 (i.e., n=5), and constructs a local range containing 24 spatially adjacent monitoring points for each monitoring point. When calculating the weighted variance of the structural response of the 18th monitoring point, the algorithm assigns weights based on the absolute value of the displacement difference between the point and its 24 neighboring points and the gradient direction angle, effectively identifying the abrupt change in local deformation in the area where the point is located. This structured neighborhood mechanism avoids variance fluctuations caused by arbitrarily selecting neighboring points and improves the stability of anomaly detection.

[0087] This embodiment provides a method for detecting the bearing capacity of pile-slab walls. It uses a pre-defined local range of n×n for each monitoring point. Centered on the monitoring point, n×n-1 spatially adjacent monitoring points are acquired to construct the local range of that monitoring point. By establishing a spatial analysis unit with a fixed geometric structure and consistent neighborhood relationships, a unified calculation basis is provided for spatial evolution characteristics such as gradient amplitude, coefficient of variation, response difference, and gradient direction angle. This eliminates deviations caused by ambiguous neighborhood definitions and suppresses isolated noise interference while ensuring the ability to capture local details. Serving as a spatial support domain for calculating the weighted variance and anomaly factors of the structural response, this method effectively improves the accuracy of identifying local mechanical behavior anomalies in pile-slab walls (such as stress concentration and deformation inconsistency), laying a spatial topological foundation for the reliability of a two-dimensional anomaly detection mechanism.

[0088] In one embodiment, calculating the weighting coefficients of other monitoring points within the local area for that monitoring point includes:

[0089] ;

[0090] in, Let a be the weighting coefficient of the a-th monitoring point within the local range of the x-th monitoring point, and let a be the weighting coefficient of the a-th monitoring point within the x-th monitoring point. Let be the cosine of the angle between the x-th monitoring point and the a-th monitoring point within the local range of the x-th monitoring point. Let x be the absolute value of the stress difference between the x-th monitoring point and the a-th monitoring point. , These represent the maximum and minimum stress values ​​in the monitoring data of the pile-slab wall structure, respectively.

[0091] Among them, the weighting coefficient It can be the contribution ratio of the a-th monitoring point to the x-th monitoring point in the weighted variance calculation of the structural response, which is determined by the difference in strain gradient direction and stress difference. It can be used to dynamically quantify the influence of neighboring monitoring points on the center point, making the weighted variance more consistent with the spatial non-uniformity and directionality of the mechanical field. It can be the cosine of the angle between the strain gradient directions of the x-th monitoring point and the a-th monitoring point, which can be used to characterize the consistency of local deformation directions between the two points and to measure the degree of coordination in mechanical behavior. Furthermore, It can be calculated by dividing the dot product of the two strain gradient vectors by their magnitudes.

[0092] The strain gradient direction angle can be the spatial angle between the strain gradient vectors of two monitoring points. It can be used to reflect the directional differences of the local deformation field. The larger the angle, the worse the deformation compatibility. For example, the strain gradient direction angle can be one or more of the following: same-direction deformation angle (0°~30°), orthogonal deformation angle (80°~100°), and opposite-direction deformation angle (150°~180°). This can be the absolute value of the stress difference between the x-th monitoring point and the a-th monitoring point, which can be used to measure the degree of local stress abrupt change and serve as the basis for intensity difference in weight allocation. In an exemplary embodiment, The absolute value of the difference between the measured stress values ​​at two monitoring points can be taken. The absolute value of the stress difference can be the absolute value of the difference between the stress measurements at two monitoring points, which can be used to reflect the discontinuity of the local stress field and to identify potential stress concentration areas.

[0093] It can be the maximum stress value at all monitoring points in the pile-slab wall structure monitoring data, which can be used as the upper limit benchmark for stress difference normalization to eliminate the influence of dimensions. In one specific embodiment... It can be the highest value of the measured stress data from all monitoring points, and participate in the global normalization of stress difference. It can be the minimum stress value at all monitoring points in the pile-slab wall structure monitoring data, and can be used as the lower limit benchmark for stress difference normalization, ensuring that the normalization denominator is non-zero and representative. Furthermore, It can be the lowest value of the measured stress data from all monitoring points, and participate in the global normalization of stress difference.

[0094] Obtain the cosine value of the angle between the strain gradient directions of the x-th monitoring point and the a-th monitoring point. This can be achieved by calculating the cosine of the angle between the strain gradient vectors of two points. For example, this operation can be performed by estimating the gradient vector from neighborhood strain data using the finite difference method and then calculating the dot product, or by directly acquiring the principal strain direction using a strain rose sensor and constructing the gradient direction vector. This allows for quantifying the consistency of local deformation directions and providing geometric feature input for weight calculation. The absolute value of the stress difference between the x-th monitoring point and the a-th monitoring point is then calculated. The difference can be the absolute value of the difference between the measured stress values ​​at two monitoring points. In one specific embodiment, this operation can be performed by directly measuring the stress values ​​using a fiber optic grating sensor and then calculating the difference, or by inverting the stress from strain data and material constitutive relations and then calculating the difference. This allows for the extraction of local stress abrupt change information as a strength difference feature. The maximum stress value in the monitoring data of the pile-slab wall structure is determined. and minimum stress This can involve iterating through stress data from all monitoring points and extracting the global extremum. Furthermore, this operation can be dynamically updated at the start of each monitoring cycle. and Alternatively, historical extreme values ​​within a sliding time window can be used as a normalization benchmark to suppress abnormal extreme value interference, thereby providing a normalization benchmark for stress difference and ensuring the stability of weight calculation under different working conditions.

[0095] For example, in the scenario of identifying anomalies in deep foundation pit support pile-slab walls, the pile-slab wall bearing capacity detection method in this embodiment can be implemented in a subway deep foundation pit project, where 30 stress and strain synchronous monitoring points are set up on the pile-slab wall. When calculating the structural response anomaly factor of monitoring point 15, the system performs weight allocation on its local range (including 7 neighboring points): the angle between the strain gradient directions of point 18 and point 15 is 160° (cosθ≈-0.94), and the stress difference reaches 2.8MPa, while the global... =5.2MPa =0.4MPa. Substituting into the formula, we get... = (1 - (-0.94)) + 2.8 / (5.2 - 0.4) ≈ 1.94 + 0.58 = 2.52, significantly higher than other neighboring points. This high weight makes point 18 contribute significantly to the weighted variance, ultimately leading to an increase in the structural response anomaly factor of point 15, which was then identified as the target monitoring point. Subsequent LOF analysis confirmed the presence of stress concentration caused by a sudden increase in local soil pressure in this area, verifying the high sensitivity of this weighting mechanism to mechanical anomalies.

[0096] This embodiment provides a method for detecting the bearing capacity of pile-slab walls. By calculating the weight coefficients of other monitoring points in a local area for that monitoring point, and by converting the cosine value of the strain gradient direction angle into a direction inconsistency penalty term, and normalizing the absolute value of the stress difference after global extremum as a strength difference term, and linearly superimposing the two to form a physically driven dynamic weight, it can achieve the integration of two-dimensional information of deformation direction coordination and stress field continuity. This makes the weighted variance of the structural response more accurately reflect the characteristics of defects such as stress concentration and deformation incoordination, improves the resolution of abnormal factors, and ultimately achieves high sensitivity and low false alarm detection of minor damage to pile-slab walls. This overcomes the technical defects of traditional uniform or distance-weighted methods that ignore the directionality and nonlinearity of the mechanical field.

[0097] In one embodiment, calculating the structural response anomaly factor at the monitoring point includes:

[0098] Calculate the structural response anomaly factor at the x-th monitoring point. :

[0099] ;

[0100] Let x be the gradient magnitude of the stress value at the x-th monitoring point. Let x be the stress-weighted variance within a local area of ​​the x-th monitoring point. Let x be the stress value at the x-th monitoring point. , ...

[0101] Among them, the structural response anomaly factor at the xth monitoring point It can be a normalized comprehensive anomaly index calculated based on stress gradient amplitude, local stress weighted variance and relative stress location index enhancement term. It can be used to quantify the risk of bearing capacity degradation at the xth monitoring point due to abnormal mechanical behavior (such as stress concentration and local instability). (The gradient magnitude of the stress value at the x-th monitoring point) can be the absolute value of the rate of change of the stress value at the x-th monitoring point within its spatial neighborhood. It can be used to identify regions of abrupt changes in the local stress field, indicating potential stress concentrations or crack initiation locations. In an exemplary embodiment, The modulus can be obtained by performing a spatial difference calculation on the stress values ​​of the x-th monitoring point and its neighboring monitoring points, and then taking the modulus. Furthermore, Can be with Together with the exponential enhancement term, these constitute the input features of the anomaly factor, reflecting spatial non-uniformity. For example, the gradient magnitude of the stress value at the x-th monitoring point is calculated. The Euclidean norm can be calculated by using the central difference method to calculate the two-dimensional gradient vector, or by using image gradient operators such as the Sobel operator to approximate the calculation.

[0102] (The stress-weighted variance within the local area of ​​the x-th monitoring point) can be the variance of stress data calculated with spatial correlation weights within the local area of ​​the x-th monitoring point. It can be used to reflect the dispersion of the local stress field and enhance sensitivity to non-coordinated deformation or local failure. In one specific embodiment, The weighted variance of the local stress values ​​can be calculated based on the weighting coefficients generated in the previous steps, using the absolute value of the structural response difference and the angle between the gradient direction. Furthermore, Can be with The complexity of the local mechanical state is characterized collaboratively. s_x (the stress value at the x-th monitoring point) can be the current stress state value obtained by actual measurement or inversion at the x-th monitoring point, and can be used as a specific physical quantity of structural response data, forming the basic input for constructing anomaly factors.

[0103] (The exponential enhancement term based on the relative position of stress) can map the position of the stress at the monitoring point relative to the global stress range into an exponentially growing function output. This can be used for nonlinear amplification of high-stress areas, allowing points approaching the limit state to receive higher anomaly scores. In an exemplary embodiment, this exponential enhancement term can be calculated first... The relative stress level within the interval [0, 1] is obtained, and then enhanced using the natural exponential function exp(). Furthermore, this exponential enhancement term can be combined with... and Multiplication amplifies the anomalous response in the critical state region. For example, for... The exponential function exp() with base e can be applied in the form of exp(k·r) (where k is the adjustment parameter and r is the relative stress), or a truncated exponential function can be used to prevent extreme amplification.

[0104] Calculate the gradient magnitude of the stress value at the x-th monitoring point. This can be achieved by calculating the spatial gradient vector and taking its magnitude based on the stress values ​​of the x-th monitoring point and its spatial neighborhood. Furthermore, this operation can be implemented by calculating the Euclidean norm after calculating the two-dimensional gradient vector using the central difference method, or by approximating it using image gradient operators such as the Sobel operator. This allows for the capture of spatial non-uniformity in the local stress field and the identification of potentially high-risk areas. The stress-weighted variance in the local area of ​​the x-th monitoring point is then obtained. This can be achieved by using the weighting coefficients calculated in previous steps to calculate the weighted variance of the stress values ​​at each point within a local area. Furthermore, this operation can be implemented by using the weighted sample variance formula or by directly calculating the weighted second-order central moments as an approximation of the variance, thus allowing the variance calculation to reflect the spatial correlation of the actual mechanical field and suppressing irrelevant noise interference.

[0105] For example, in the scenario of real-time safety assessment of pile-slab walls for deep foundation pit support, the pile-slab wall bearing capacity testing method in this embodiment could be: In a deep foundation pit project for a subway in a certain city, 30 stress monitoring points are set up on the pile-slab wall. The system collects the stress values ​​at each point in real time. And calculate the global =2.8MPa =0.3MPa. For monitoring point No. 15, its =2.6MPa, ∇s_x =0.45MPa / m (indicating a significant stress gradient in the vicinity). =0.18 (the weighted variance is high, reflecting local stress inconsistency). The calculated relative stress is (2.6-0.3) / (2.8-0.3) = 0.92, exp(0.92) ≈ 2.51. The product of the three terms is 0.45×0.18×2.51 ≈ 0.203, which is normalized by norm(). =0.87, exceeding the threshold of 0.75, and was marked as a target monitoring point. Subsequent LOF analysis confirmed it as a local outlier, and combined with location information, it was determined to be caused by concrete cracking near the support node, triggering an early warning.

[0106] This embodiment provides a method for detecting the bearing capacity of a pile-slab wall, which calculates the structural response anomaly factor at the x-th monitoring point. By fusing stress gradient amplitude to identify local stress abrupt changes, introducing stress weighted variance to reflect local discreteness, employing an exponential function based on global stress extremum normalization to nonlinearly enhance high-stress areas, and then linearly normalizing the product of these three factors to output a unified anomaly score, we can achieve the technical effect of explicitly constructing anomaly factors with physical meaning and engineering interpretability, improving the sensitivity and discrimination ability of bearing capacity degradation areas, especially critical state points, enhancing the reproducibility of the algorithm, and providing high-quality input for subsequent anomaly detection.

[0107] In one embodiment, the target monitoring points are processed based on the LOF algorithm to obtain the pile-slab wall bearing capacity test results, including:

[0108] The comprehensive score of the target monitoring point is obtained based on the material property anomaly factor and the structural response anomaly factor of the target monitoring point.

[0109] The comprehensive score can be a single quantitative index generated by fusing material property anomaly factors and structural response anomaly factors of the target monitoring point. It can be used to rank the overall risk level of the target monitoring point and prioritize the identification of high-risk areas. In this embodiment, the comprehensive score can map the two types of anomaly factors to a unified score through weighted summation, product, or other normalized fusion functions. For example, the comprehensive score can adopt a weighted linear combination form, where the weight coefficients are multiplied by the material property anomaly factor and the structural response anomaly factor respectively and then summed. Alternatively, a geometric mean or product form can be used to emphasize points with simultaneously high values ​​for both factors. Obtaining the comprehensive score of the target monitoring point based on its material property anomaly factors and structural response anomaly factors can be achieved by generating a single score value from the two calculated anomaly factors through a fusion function. Furthermore, this operation can be implemented using a weighted linear combination or product form, thereby achieving unified quantification of two-dimensional anomaly information and supporting priority ranking.

[0110] Obtain the comprehensive scores of all target monitoring points in descending order, and select a preset number of target monitoring points as candidate points from the comprehensive score descending order.

[0111] The descending order of the comprehensive scores of all target monitoring points can be a sequence formed by sorting all target monitoring points from highest to lowest comprehensive score. This sequence can be used to provide a priority basis for candidate point selection, ensuring that high-risk points participate in subsequent analysis first. In an exemplary embodiment, this sorting can use a quicksort algorithm to sort the score array while retaining the original index to associate spatial location information. The preset number of target monitoring points can be the top N target monitoring points selected from the descending order of comprehensive scores. This can be used to limit the input size of the LOF algorithm and focus on the most potentially dangerous areas. Furthermore, the preset number of target monitoring points can include, but is not limited to, one or more of the following: the top 5 high-scoring points, the top 10% of high-scoring points, and a dynamic number (such as all points with a score greater than 0.8).

[0112] Candidate points can be a subset of high-risk target monitoring points selected for LOF calculation after comprehensive scoring and ranking. These can serve as input for LOF local density analysis, improving computational efficiency and noise resistance. In one specific embodiment, candidate points can participate in LOF calculation together with their neighboring points in the original monitoring network. Obtaining a preset number of target monitoring points as candidate points from the descending order of comprehensive scores can be achieved by truncating the first N items of the ranking sequence. Furthermore, this operation can be implemented by using a fixed number (e.g., the first 10), selecting by proportion (e.g., the first 20%), or truncating by a scoring threshold (e.g., a score greater than 0.75), thereby narrowing the scope of LOF analysis and improving computational efficiency and engineering relevance.

[0113] Calculate the LOF value of the candidate points, calculate the anomaly threshold based on the mean and standard deviation of the LOF values ​​of all candidate points, and obtain the abnormal bearing points in the pile-slab wall bearing capacity test results based on the comparison between the LOF values ​​of the candidate points and the anomaly threshold.

[0114] The LOF value of a candidate point can be the local outlier factor obtained by applying the LOF algorithm to each candidate point, which can be used to quantify the outlier degree of the candidate point in its local neighborhood. In this embodiment, the LOF value of a candidate point can be calculated by performing the standard LOF algorithm within the candidate point set or in the extended neighborhood. The mean of the LOF values ​​of all candidate points can be the arithmetic mean of the LOF value set of candidate points, which can be used to characterize the typical outlier level of the current candidate point group and serve as a basic parameter for calculating the anomaly threshold. The standard deviation of the LOF values ​​of all candidate points can be a measure of the dispersion of the LOF value set of candidate points, which can be used to reflect the volatility of the LOF value distribution and to dynamically set the anomaly judgment boundary.

[0115] The outlier threshold can be a dynamically calculated outlier determination boundary based on the mean and standard deviation of the LOF values ​​of candidate points. This can be used to achieve adaptive anomaly identification, avoiding the failure of fixed thresholds under non-stationary data. In an exemplary embodiment, the outlier threshold can be calculated by multiplying the mean by an adjustment coefficient by the standard deviation. Abnormal bearing capacity points in the pile-slab wall bearing capacity test results can be candidate points whose LOF values ​​exceed the outlier threshold. These can be used as key locations of bearing capacity degradation in the final output for maintenance decisions.

[0116] The LOF value of candidate points can be calculated by performing the LOF algorithm within the candidate point set or within a spatial range including its original neighborhood. Furthermore, this operation can be achieved by constructing a k-nearest neighbor graph only among candidate points to calculate LOF, or by including it in the LOF calculation along with neighborhood points from the original monitoring network, thus identifying local density anomalies within high-risk subsets. The anomaly threshold is calculated based on the mean and standard deviation of the LOF values ​​of all candidate points, and the outlier determination boundary can be dynamically set using statistical parameters. Furthermore, this operation can be achieved by setting the anomaly threshold to the mean plus twice the standard deviation, or the mean plus 1.5 times the interquartile range, thus adapting anomaly identification to the current data distribution and enhancing robustness. Based on the comparison between the LOF values ​​of candidate points and the anomaly threshold, anomalous bearing capacity points in the pile-slab wall bearing capacity detection results are obtained. If the LOF value of a candidate point is greater than the anomaly threshold, it is marked as an anomalous bearing capacity point. Furthermore, this operation can be achieved by comparing the LOF values ​​point-by-point with the dynamic threshold, thus outputting final anomalies with high outlier prevalence and high overall risk, improving the reliability of the results.

[0117] Taking the post-flood season assessment of pile-slab walls in mountainous railway subgrades as an example, the pile-slab wall bearing capacity testing method in this embodiment can be as follows: After the flood season, the system calculates a comprehensive score from 32 target monitoring points and sorts them in descending order, selecting the top 8 as candidate points. The LOF value of these 8 points is calculated, with a mean of 1.2 and a standard deviation of 0.3. An anomaly threshold is set as 1.2 plus twice 0.3, which equals 1.8. Among them, the LOF values ​​of candidate points No. 3 and No. 7 are 2.1 and 1.9, respectively, exceeding the threshold, and are judged as abnormal bearing capacity points. On-site verification confirmed that the two locations were near drainage ditches, with concrete spalling and exposed rebar, verifying the effectiveness of the method.

[0118] This embodiment provides a method for detecting the bearing capacity of pile-slab walls. It obtains a comprehensive score for each target monitoring point based on material property anomaly factors and structural response anomaly factors. All target monitoring points are then sorted in descending order of their comprehensive scores. A predetermined number of target monitoring points are selected as candidate points from this descending order. The Loose-of-Flight (LOF) value of the candidate points is calculated. An anomaly threshold is calculated based on the mean and standard deviation of the LOF values ​​of all candidate points. Anomaly bearing points are identified in the pile-slab wall bearing capacity detection results by comparing the LOF values ​​of the candidate points with the anomaly threshold. High-risk points are quantified and sorted by fusing material property anomaly factors and structural response anomaly factors to narrow the LOF analysis scope. Anomaly thresholds are dynamically set based on the statistical distribution of the LOF values ​​of the candidate points to achieve adaptive outlier identification. This method effectively combines physics-driven two-factor screening with data-driven local density analysis, improving the robustness and decision support value of anomaly detection while maintaining engineering interpretability.

[0119] In one embodiment, calculating the LOF value of a candidate point includes:

[0120] The candidate point is located at a distance of K from its neighboring points in the neighborhood. The local density of the candidate point is obtained, and the average of the ratios of the local densities of the neighboring points to that of the candidate point is recorded as the LOF value of the candidate point.

[0121] The K-distance neighborhood of a candidate point can be a set of all monitoring points centered on the candidate point and encompassing its Kth nearest neighbor. This can be used to dynamically determine the local context and adapt to non-uniform monitoring point distribution. In this embodiment, the K-distance neighborhood of a candidate point can be determined by calculating the distances from the candidate point to other points, taking the Kth smallest distance as the K-distance, and including all points whose distance does not exceed this K-distance. For example, the K-distance neighborhood of a candidate point can be determined using Euclidean distance in a two-dimensional monitoring grid, or by selecting the K / 2 points before and after the candidate point in the one-dimensional space along the pile axis according to the monitoring point index order.

[0122] Neighboring points can be other monitoring points located within the K-distance neighborhood of a candidate point, and can be used to form the base sample for local density calculation and LOF ratio analysis. Furthermore, neighboring points can participate in the LOF value calculation together with their local densities. The local density of a candidate point can be the reciprocal of the average reachable distance from all points within the K-distance neighborhood to that point, and can be used to quantify the data density of the area where the candidate point is located. In an exemplary embodiment, the local density of a candidate point can be obtained by first calculating the reachable distance from each neighboring point to the candidate point (taking the maximum of the two K-distances), then averaging and taking the reciprocal. For example, the local density can be expressed as 1 divided by the quotient of the sum of all reachable distances within the K-distance neighborhood and the number of neighboring points, or by using a harmonic mean instead of an arithmetic mean to enhance sensitivity to minimal reachable distances.

[0123] The local density ratio of a neighboring point to the candidate point can be obtained by dividing the local density of each neighboring point by the local density of the candidate point. This ratio reflects the density difference of the candidate point relative to its neighborhood; a ratio greater than 1 indicates a lower density for the candidate point. In one specific embodiment, the local density ratio of a neighboring point to the candidate point can be obtained by calculating the quotient of the local density of each neighboring point to the local density of the candidate point. The LOF value of the candidate point can be the arithmetic mean of the local density ratios of all neighboring points to the candidate point. This value can be used to quantify the degree to which the candidate point is a local outlier; a value greater than 1 indicates a potential anomaly. Further, the LOF value of the candidate point can be obtained by averaging the density ratios of all neighboring points within a K-distance neighborhood. For example, the LOF value can be truncated by adjusting the density ratios (e.g., setting an upper limit of 3) to suppress the influence of extreme values.

[0124] The neighboring points in the K-distance neighborhood of the preset candidate point can be determined by setting a preset K value, collecting all monitoring points whose distance does not exceed this value as neighboring points. Furthermore, the neighboring points in the K-distance neighborhood of the preset candidate point can be determined by using Euclidean distance in a two-dimensional monitoring grid, or by selecting K / 2 points before and after the pile axis in a one-dimensional space according to the monitoring point index order as the neighborhood. This allows for the construction of a local neighborhood that adapts to the point distribution density, avoiding the failure of a fixed radius in sparse regions.

[0125] The local density of a candidate point can be obtained by calculating the average reachable distance from each point in the K-distance neighborhood to the candidate point and taking its reciprocal. Alternatively, the local density can be obtained by dividing the local density by 1 and the average reachable distance in the neighborhood, or by using a harmonic mean instead of an arithmetic mean, thus generating a density index reflecting the local data density. The mean of the ratios of the local densities of neighboring points to the candidate point is recorded as the LOF value of the candidate point. This can be achieved by calculating the ratio of the local density of each neighboring point to the local density of the candidate point, and then averaging these ratios. Furthermore, the mean of the ratios of the local densities of neighboring points to the candidate point can be recorded as the LOF value of the candidate point by taking the arithmetic mean of the density ratios of all neighboring points, or by truncating the density ratios, thus outputting a standardized measure of outlier severity, supporting the identification of anomalous carrying points.

[0126] Taking the service evaluation of pile-slab walls in deep foundation pits of urban subways as an example, the pile-slab wall bearing capacity detection method in this embodiment can be to set K=5 for each of the 6 selected candidate points. Taking candidate point No. 3 as an example, its K distance neighborhood includes 5 neighboring points. Its local density is calculated to be 0.82, while the local densities of the 5 neighboring points are 1.05, 0.98, 1.12, 0.94, and 1.01, respectively, with corresponding density ratios of 1.28, 1.20, 1.37, 1.15, and 1.23, and the LOF value is the average of 1.25. Because it is greater than 1 and exceeds the dynamic threshold of 1.20, it is judged as an abnormal bearing point. On-site testing revealed that there were micro-cracks in the concrete and thinning of the steel reinforcement protective layer at the corresponding location, verifying the effectiveness of the LOF value in capturing local sparse anomalies.

[0127] This embodiment provides a method for detecting the bearing capacity of a pile-slab wall. By using neighboring points in the K-distance neighborhood of a preset candidate point, the local density of the candidate point is obtained. The mean of the ratio of the local density of the neighboring points to that of the candidate point is recorded as the LOF value of the candidate point. By using the K-distance neighborhood to adaptively determine the local context to avoid distortion under a fixed radius in a non-uniform monitoring layout, defining the local density based on the reachability distance to accurately reflect the local density of the data point distribution, and generating the LOF value by the mean of the ratio of the local density of the neighboring points to that of the candidate point to give the outlier metric a clear mathematical and physical meaning, this method can effectively avoid the interference of the original monitoring noise on the density estimation, accurately identify potential damage points in relatively sparse areas, and embed the traditional LOF algorithm into a multi-stage anomaly identification process to significantly improve the accuracy, noise resistance, and engineering interpretability of anomaly bearing point identification.

[0128] In one embodiment, after obtaining the test results of the pile-slab wall bearing capacity, the process further includes:

[0129] Multiple clusters are obtained by clustering the abnormal carrying points.

[0130] In this context, a cluster can be a set of points with internal similarity formed by spatially or feature-based clustering of anomalous load-bearing points. It can be used to group discrete anomalous points into diseased areas of engineering significance, facilitating causal analysis and maintenance location. In this embodiment, clusters can be grouped based on the spatial coordinates or two-factor feature vectors of the anomalous load-bearing points, and their acquisition can employ clustering algorithms (such as DBSCAN, K-means, or hierarchical clustering). For example, clusters can include, but are not limited to, one or more of the following: spatial proximity clusters, material-structure two-factor similarity clusters, and hybrid feature-driven clusters.

[0131] Clustering anomalous load-bearing points yields multiple clusters. This can be achieved by using anomalous load-bearing points as input and applying clustering algorithms to group them based on spatial location or two-factor feature similarity. Furthermore, this operation can be implemented by automatically identifying density-connected anomalous regions through DBSCAN clustering based on spatial coordinates, or by grouping them according to anomalous patterns through K-means clustering based on two-dimensional feature vectors of [material factor, structural factor]. This allows isolated anomalous points to be integrated into diseased areas with engineering significance, improving diagnostic stability and interpretability.

[0132] Obtain the mean values ​​of material property anomaly factors and structural response anomaly factors for anomalous load-bearing points in the cluster.

[0133] The mean of the material property anomaly factor at anomalous load-bearing points within a cluster can be the arithmetic mean of the material property anomaly factors at all anomalous load-bearing points within a cluster. It can be used to characterize the average level of material degradation for the entire cluster and is compared with the structural response factor to identify the dominant anomaly type. The mean of the structural response anomaly factor at anomalous load-bearing points within a cluster can be the arithmetic mean of the structural response anomaly factors at all anomalous load-bearing points within a cluster. It can be used to characterize the average level of mechanical behavior anomalies for the entire cluster and is compared with the material factor to identify the dominant anomaly type.

[0134] Obtaining the mean values ​​of material property anomaly factors and structural response anomaly factors for anomalous load-bearing points within a cluster can be achieved by calculating the arithmetic mean of the two anomaly factors for all points within each cluster. In one specific embodiment, this operation makes it possible to quantify cluster-level anomaly characteristics, providing numerical evidence for causal identification.

[0135] If the mean value of the material property anomaly factor of a cluster is greater than the mean value of the structural response anomaly factor, then the anomaly of the cluster is a material defect.

[0136] Material defects can be types of pile-slab wall defects dominated by material performance degradation such as insufficient concrete strength and steel corrosion. These defects can be used as one of the results for identifying the cause of anomalies, guiding targeted material repair measures (such as deep carbonization treatment and cathodic protection). The system determines whether the mean value of the material characteristic anomaly factor for a cluster is greater than the mean value of the structural response anomaly factor. If so, the anomaly of that cluster is determined to be a material defect. This can be achieved by performing numerical comparisons and outputting corresponding defect type labels. In an exemplary embodiment, this operation enables the automatic identification of material-dominated defects.

[0137] If the mean value of the material property anomaly factor of a cluster is less than the mean value of the structural response anomaly factor, then the anomaly of the cluster is a structural stress anomaly.

[0138] Among these, structural stress anomalies can be pile-slab wall defects dominated by abnormal mechanical responses such as stress concentration and deformation incoordination. These anomalies can be used as one of the results for determining the cause of the anomalies, guiding structural reinforcement or load adjustment measures (such as adding supports or unloading). The system determines whether the mean value of the material property anomaly factor for a cluster is less than the mean value of the structural response anomaly factor. If so, the anomaly of that cluster is determined to be a structural stress anomaly, which can be achieved by performing numerical comparisons and outputting corresponding defect type labels. Furthermore, this operation enables the automatic identification of mechanically dominated defects.

[0139] If the mean of the material property anomaly factor of a cluster is equal to the mean of the structural response anomaly factor, then the anomaly of the cluster is caused by the combined effect of material defects and structural stress anomalies.

[0140] Among these, the combined effect of material defects and structural stress anomalies can result in a complex condition caused by the coupling of similar degrees of material degradation and mechanical anomalies. This can be used as one of the results for identifying the cause of anomalies, indicating the need for a comprehensive maintenance strategy that simultaneously implements material repair and structural reinforcement. The system determines whether the mean of the material characteristic anomaly factor for a cluster is equal to the mean of the structural response anomaly factor. If so, the anomaly of that cluster is determined to be caused by the combined effect of material defects and structural stress anomalies. This can be done within an allowable error range by determining whether the two means are equal, and a composite condition label is output. For example, this operation identifies material-structure coupled damage areas, supporting comprehensive maintenance decisions.

[0141] Taking the service evaluation of pile-slab walls in deep foundation pits of urban subways as an example, the pile-slab wall bearing capacity detection method in this embodiment can identify 12 abnormal bearing points and form 3 clusters through DBSCAN clustering. Cluster A (5 points) has a material factor mean of 0.82 and a structural factor mean of 0.45, which is judged as material defects. On-site, it was found that the area is adjacent to a sewage pipe and there is serious steel corrosion. Cluster B (4 points) has a material factor mean of 0.38 and a structural factor mean of 0.79, which is judged as abnormal structural stress. The corresponding location is exactly the temporary loading area, with obvious outward tilt. Cluster C (3 points) has a mean of 0.65 for both types of factors, which is judged as the combined effect of material defects and abnormal structural stress. Verification found that there was both concrete cracking and overload construction, which verified the effectiveness of the discrimination logic.

[0142] This embodiment provides a method for detecting the bearing capacity of pile-slab walls. By clustering abnormal bearing points, multiple clusters are obtained. The mean values ​​of material property anomaly factors and structural response anomaly factors of abnormal bearing points in the clusters are obtained. Based on the relative magnitude of the two, a discrimination rule is established: the material factor dominates and corresponds to material defects, the structural factor dominates and corresponds to structural stress anomalies, and when the two are equal, it is considered coupled damage. This method can achieve the technical effects of summarizing discrete anomaly points into disease areas with engineering significance, transforming numerical detection results into interpretable engineering diagnostic conclusions, improving diagnostic stability and noise resistance, and clarifying the causes of anomalies to guide differentiated maintenance strategies.

[0143] Furthermore, this invention also proposes a pile-slab wall bearing capacity testing system, the system comprising:

[0144] processor;

[0145] Memory used to store the processor's executable instructions;

[0146] The processor is configured to execute the instructions to implement the steps of the pile-slab wall bearing capacity detection method as described in any of the above-mentioned methods.

[0147] Other embodiments or specific implementations of the pile-slab wall bearing capacity testing system of the present invention can be referred to the above-described method embodiments, and will not be repeated here.

[0148] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A method for detecting the load bearing capacity of a pile wall, characterized in that, The method comprises: Obtaining material parameters and structure response data corresponding to each monitoring point in the pile-slab wall structure monitoring data; calculating the material characteristic anomaly factor of the monitoring point according to the gradient amplitude of the material parameters of the monitoring point, the coefficient of variation in the eight-neighborhood of the monitoring point, and the difference between the material parameters of the monitoring point and the design reference parameters; Constructing the local range of the monitoring point with the monitoring point as the center, calculating the weight coefficient of other monitoring points in the local range of the monitoring point to the monitoring point according to the absolute value of the structure response difference between the monitoring point and other monitoring points in the local range of the monitoring point and the included angle of the gradient direction, and obtaining the structure response weighted variance in the local range of the monitoring point; Calculating the structure response anomaly factor of the monitoring point according to the gradient amplitude of the structure response data of the monitoring point, the structure response weighted variance in the local range of the monitoring point, and the difference between the structure response data of the monitoring point and the minimum value of the structure response in the pile-slab wall structure monitoring data; In response to the material characteristic anomaly factor and the structure response anomaly factor of the monitoring point being greater than the corresponding threshold value, obtaining the target monitoring point; processing the target monitoring point based on the LOF algorithm to obtain the pile-slab wall bearing capacity detection result.

2. The pile wall bearing capacity detection method according to claim 1, wherein The obtaining of the material parameters and structure response data corresponding to each monitoring point in the pile-slab wall structure monitoring data comprises: Collecting the material parameter data set and the structure response data set in the structure monitoring data after applying a graded load to the pile-slab wall, and pre-processing the material parameter data set and the structure response data set to obtain the material parameters and structure response data corresponding to each monitoring point, wherein the pre-processing mode at least includes normalizing the material parameter data set and the structure response data set; the material parameters include concrete elastic modulus and steel yield strength, and the structure response data includes stress, strain and displacement.

3. The pile wall bearing capacity detection method according to claim 1, wherein The calculation of the material characteristic anomaly factor of the monitoring point comprises: The material design parameter data of the pile board wall is acquired and normalized to obtain material benchmark parameters; and the material characteristic abnormality factor of the xth monitoring point is calculated : ； wherein, is the gradient amplitude of the material parameter at the xth monitoring point, , , is the standard deviation of the material parameter, the mean value of the material parameter, the coefficient of variation in the octant of the xth monitoring point, respectively, , is the measured material parameter, the material reference parameter at the xth monitoring point, respectively, ln() is the logarithm function with base e, || is the absolute value symbol, norm() is the linear normalization function.

4. The pile wall bearing capacity detection method according to claim 1, wherein The construction of the local range of the monitoring point with the monitoring point as the center comprises: The local range size of the monitoring point is preset as n*n, and the local range of the monitoring point is constructed by obtaining n*n-1 spatially adjacent monitoring points around the monitoring point with the monitoring point as the center.

5. The pile wall load capacity detection method according to claim 1, wherein The calculation of the weight coefficient of other monitoring points in the local range of the monitoring point to the monitoring point comprises: ； wherein, is the weight coefficient of the a-th monitoring point to the x-th monitoring point in the local range of the x-th monitoring point, is the cosine value of the included angle between the strain gradient direction of the x-th monitoring point and the a-th monitoring point in the local range of the x-th monitoring point, is the absolute value of the stress difference between the x-th monitoring point and the a-th monitoring point, , respectively are the maximum stress value and the minimum stress value in the pile board wall structure monitoring data.

6. The pile wall bearing capacity detection method according to claim 1, wherein The calculation of the structure response anomaly factor of the monitoring point comprises: calculating a structural response anomaly factor for the xth monitoring point : ； the gradient amplitude of the stress value of the xth monitoring point, the stress weighted variance in the local range of the xth monitoring point, the stress value of the xth monitoring point, , respectively the maximum stress value and the minimum stress value in the monitoring data of the pile board wall structure, exp() is an exponential function with e as the base, and norm() is a linear normalization function.

7. The pile wall bearing capacity detection method according to claim 1, wherein The processing of the target monitoring point based on the LOF algorithm to obtain the pile-slab wall bearing capacity detection result comprises: Obtaining the comprehensive score of the target monitoring point according to the material characteristic anomaly factor and the structure response anomaly factor of the target monitoring point; Obtaining the comprehensive scores of all target monitoring points in descending order, and obtaining a preset number of target monitoring points in the comprehensive scores in descending order as candidate points; Calculating the LOF value of the candidate point, calculating the anomaly threshold value according to the mean and standard deviation of the LOF values of all candidate points, and obtaining the abnormal bearing point in the pile-slab wall bearing capacity detection result according to the comparison result of the LOF value of the candidate point and the anomaly threshold value.

8. The pile wall bearing capacity detection method according to claim 7, wherein The calculation of the LOF value of the candidate point comprises: Obtaining the local density of the candidate point by presetting the neighbor points in the K-distance neighborhood of the candidate point, and taking the mean value of the ratio of the neighbor points to the local density of the candidate point as the LOF value of the candidate point.

9. The pile wall bearing capacity detection method according to claim 7, wherein The pile-slab wall bearing capacity detection result is obtained, and then the method further comprises the following steps: clustering the abnormal bearing points to obtain a plurality of clustering clusters, and obtaining a material property abnormal factor mean value and a structural response abnormal factor mean value of the abnormal bearing points in the clustering clusters; if the material property abnormal factor mean value of the clustering cluster is greater than the structural response abnormal factor mean value, the abnormality of the clustering cluster is a material defect; if the material property abnormal factor mean value of the clustering cluster is less than the structural response abnormal factor mean value, the abnormality of the clustering cluster is a structural stress abnormality; and if the material property abnormal factor mean value of the clustering cluster is equal to the structural response abnormal factor mean value, the abnormality of the clustering cluster is a joint action of a material defect and a structural stress abnormality.

10. A system for detecting the load bearing capacity of a shoring wall, characterized by The system comprises: a processor; a memory for storing instructions executable by the processor; wherein the processor is configured to execute the instructions to implement the steps of the pile-slab wall bearing capacity detection method according to any one of claims 1-9.

Images