Pile plate type roadbed detection and prediction maintenance method and system based on AI analysis

Abstract

The invention discloses a pile-slab roadbed detection and prediction maintenance method and system based on AI analysis. The method comprises the steps that historical record data, multi-source real-time data and project general situation data of a pile-slab roadbed are acquired; analyzing and fusing multi-source real-time data based on an AI intelligent model, extracting real-time structural feature data of the pile-slab roadbed, and obtaining real-time state data of the pile-slab roadbed; establishing an intelligent state prediction model, and predicting the state of the pile plate type roadbed in the next detection period by taking the real-time state data and the historical record data as input; and performing risk assessment on the state of the pile-slab roadbed in the next detection period, and correspondingly selecting to execute a maintenance decision based on a risk assessment result. According to the method, an artificial intelligence model is adopted to efficiently process and fuse real-time data collected by a multi-source sensor, an iterative optimization prediction model is established to predict the state of the pile plate type roadbed in the next stage, risk prediction and pre-estimation are carried out, and automatic detection, prediction and maintenance of the pile plate type roadbed are achieved.

Description

Technical Field

[0001] This invention relates to the field of intelligent monitoring technology for road and bridge engineering, specifically to a method and system for detecting and predicting the maintenance of pile-slab subgrade based on AI analysis. Background Technology

[0002] Pile-slab roadbeds are a common type of highly stable roadbed structure, widely used in engineering scenarios such as soft soil areas and high embankment sections. However, long-term loads, geological changes, and environmental factors can lead to deviations in roadbed verticality and settlement, affecting road safety. Traditional testing methods, such as total stations and levels, rely on manual measurement, which is inefficient, costly, and difficult to implement in real-time monitoring and predictive maintenance.

[0003] In recent years, AI technology has been applied in the field of civil engineering monitoring. However, existing technologies mostly focus on single detection or prediction functions, lacking an integrated solution for verticality detection, settlement prediction, and intelligent maintenance of pile-slab subgrades. Therefore, there is an urgent need for an intelligent, automated, and high-precision monitoring and maintenance system to improve the efficiency and safety of pile-slab subgrade maintenance. Summary of the Invention

[0004] To address the problems existing in the prior art, this invention provides a method and system for detection, prediction and maintenance of pile-slab subgrade based on AI analysis. The purpose is to use an artificial intelligence model to efficiently process and fuse real-time data collected from multiple sources of sensors, establish an iteratively optimized prediction model to predict the state of the pile-slab subgrade in the next stage, and use a neural network for risk assessment to achieve detection, prediction and automated maintenance of the pile-slab subgrade.

[0005] To achieve the above objectives, the present invention provides the following technical solution: In a first aspect, the present invention provides a method for detection and predictive maintenance of pile-slab subgrade based on AI analysis, comprising: In response to a preset detection cycle, historical data, multi-source real-time data, and project overview data of the pile-slab roadbed are acquired; the multi-source real-time data includes inspection image data, laser point cloud data, contact measurement data, and temperature and humidity data. Based on the AI ​​intelligent model analysis and fusion of the multi-source real-time data, real-time structural feature data of the pile-slab roadbed is extracted; real-time status data of the pile-slab roadbed is obtained based on the real-time structural feature data and the project overview data. A state-intelligent prediction model is established, using the real-time state data and the historical data as inputs, to predict the state of the pile-slab subgrade in the next detection cycle; A risk assessment is conducted on the condition of the pile-slab subgrade for the next inspection cycle, and maintenance decisions are selected based on the risk assessment results.

[0006] Preferably, the historical data includes historical testing data, historical maintenance data, and historical stress data; the project overview data includes geological environment data and dimensional parameters of the pile-slab structure.

[0007] Preferably, the step of analyzing and fusing the multi-source real-time data based on the AI ​​intelligent model to extract real-time structural feature data of the pile-slab subgrade includes: The pre-created offline 3D model template of the pile is projected onto the space of the real-time laser point cloud. Point cloud registration is performed based on the iterative nearest point algorithm to obtain the 3D contour description of the pile in the real-time laser point cloud. Based on the projection of the 3D contour description into the inspection image and the gradient change of the inspection image, the laser point cloud and the piles in the inspection image are matched, and the real-time true pose of the piles is obtained through iterative optimization. Real-time pose, contact measurement data, and temperature and humidity data are input into the AI ​​intelligent model for data fusion, and real-time structural characteristic data of pile-slab roadbed are output.

[0008] Preferably, the step of matching the laser point cloud with the piles in the inspection image based on the projection of the 3D contour description into the inspection image and the gradient changes of the inspection image, and iteratively optimizing to obtain the real-time true pose of the piles includes: The 3D contour description is projected onto the inspection image to obtain the predicted pile pose, and a binary mask of the inspection image with respect to the predicted pile pose is generated. Calculate the minimum distance from each pixel in the binary mask to the predicted pose of the pile in the inspection image, wherein the sign of the minimum distance is inverted for different binary value regions; Extract the grayscale gradient magnitude of each pixel in the inspection image based on the gradient operator; The joint observation probability of a pixel is constructed based on the gray-level gradient magnitude and the minimum distance of the pixel, and the feature matching loss function between the predicted pile pose and the inspection image is calculated based on the joint observation probability. Iteratively calculate the minimum value of the feature matching loss function, and output the offset update amount between the predicted pile pose and the inspection image when the feature matching loss function reaches its minimum value; The projection position of the 3D contour description in the inspection image is adjusted based on the offset update amount to obtain the real pose image of the pile, and the real pose of the pile is output.

[0009] Preferably, the joint observation probability of the pixel based on the gray-level gradient magnitude and the minimum distance is specifically as follows: in, Represents pixels in a binary mask The minimum distance to predict the pose of the pile in the inspection image; This represents a smoothing function used to describe the minimum distance. With pixels The probability distribution function belonging to the foreground region; This is a foreground gradient model for inspected images, used to describe the probability distribution function of the gradient in the foreground region of the inspected image. This represents the gradient of pixels in the foreground region. Indicates the foreground region of the image; This is a background gradient model for inspection images, used to describe the probability distribution function of the gradient in the background region of the inspection image. This represents the gradient of pixels in the background region. This represents the background area of ​​the image.

[0010] Preferably, the foreground gradient model is specifically: The background gradient model is specifically as follows: in, This represents the normalized grayscale gradient magnitude; This is a smoothing factor.

[0011] Preferably, the state intelligent prediction model is: In the formula, Indicates based on the first The optimal estimate of the detection cycle for the first Prediction of the condition of pile-slab subgrade during the inspection cycle; Indicates the first Real-time status data during the detection cycle; The control input matrix represents the real-time status data. Indicates based on the first Historical data of the detection cycle for the first Iterative optimization data for detecting the subgrade condition during the detection cycle; Represents the state transition matrix of historical data; Indicates the first Data calculated by the intelligent model after maintenance and repair during the inspection cycle; Represents the calculation matrix; Indicates the first Deviation control matrix for the detection cycle.

[0012] Preferably, the first Real-time status data of the detection cycle ,in, This is the vertical angle deviation. This refers to the verticality displacement deviation. For settlement, Settlement rate; The data calculated by the intelligent model is as follows: In the formula, Indicates the first Real-time status data during the detection cycle; Indicates the first The adjustment matrix for the detection cycle; Indicates the first Deviation control matrix for the detection cycle.

[0013] Preferably, the risk assessment of the pile-slab subgrade status for the next inspection cycle involves using real-time status data of the pile-slab subgrade from different inspection cycles as input to train and optimize a neural network model for short-term and long-term risk trend prediction, and executing maintenance decisions based on the risk level: emergency measures for high-risk levels, planned maintenance for medium-risk levels, and routine maintenance for low-risk levels.

[0014] Secondly, the present invention provides an AI-based system for detecting and predicting the maintenance of pile-slab subgrade, comprising: The data acquisition module is used to acquire historical data, multi-source real-time data, and project overview data of the pile-slab roadbed in response to a preset detection cycle; the multi-source real-time data includes inspection image data, laser point cloud data, contact measurement data, and temperature and humidity data. The AI ​​analysis module is used to analyze and fuse the multi-source real-time data based on the AI ​​intelligent model to extract real-time structural feature data of the pile-slab roadbed; and to obtain real-time status data of the pile-slab roadbed based on the real-time structural feature data and the project overview data. The intelligent prediction module is used to establish a state intelligent prediction model, taking the real-time state data and the historical data as input, to predict the state of the pile-slab subgrade in the next detection cycle. The maintenance decision module is used to conduct a risk assessment of the condition of the pile-slab subgrade in the next inspection cycle, and select and execute maintenance decisions based on the risk assessment results.

[0015] The present invention provides a method and system for detection and predictive maintenance of pile-slab subgrade based on AI analysis, which has the following beneficial effects: This invention utilizes multi-source sensors to collect data in real time and combines it with AI analysis, improving detection efficiency by over 80% compared to traditional manual methods (such as total stations and levels), significantly reducing manual intervention and time costs. It employs deep learning algorithms to integrate detection, prediction, and maintenance decision-making, overcoming the limitations of traditional single-function technologies and forming a closed-loop management system. By using neural networks to predict short-term and long-term risk trends and combining quantitative indicators (such as settlement velocity and tilt angle), risks are categorized into high, medium, and low levels, and differentiated maintenance strategies are matched accordingly. Predictive maintenance reduces unexpected repair costs and lowers long-term maintenance costs. Precise intervention slows structural deterioration and extends the service life of pile-slab subgrades. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating a method for detecting and predicting the maintenance of pile-slab subgrade based on AI analysis, according to the present invention. Figure 2 This is a schematic diagram of the framework of the AI-based pile-slab subgrade detection and predictive maintenance method of the present invention; Figure 3 This is a schematic diagram of the process of iteratively optimizing the acquisition of the real-time true pose of the pile in one embodiment of the present invention; Figure 4 This is a thermal diagram of the verticality of a pile-slab roadbed structure provided in one embodiment of the present invention; Figure 5 This is a schematic diagram of the process for risk assessment of the status of pile-slab subgrade in the next detection cycle in one embodiment of the present invention; Figure 6 This is a structural block diagram of a pile-slab roadbed detection and predictive maintenance system based on AI analysis, according to the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] To facilitate understanding of this embodiment, a detailed description of an AI-based pile-slab subgrade detection and predictive maintenance method disclosed in this embodiment of the invention will be provided first.

[0019] refer to Figure 1 and Figure 2 As shown, the method includes: S1, in response to a preset detection cycle, acquires historical data, multi-source real-time data, and project overview data of the pile-slab roadbed; the multi-source real-time data includes inspection image data, laser point cloud data, contact measurement data, and temperature and humidity data; S2, Based on the AI ​​intelligent model, analyze and integrate the multi-source real-time data to extract the real-time structural feature data of the pile-slab roadbed; based on the real-time structural feature data and the project overview data, obtain the real-time status data of the pile-slab roadbed; S3, Establish a state intelligent prediction model, take the real-time state data and the historical data as input, and predict the state of the pile-slab subgrade in the next detection cycle; S4. Conduct a risk assessment of the condition of the pile-slab subgrade for the next inspection cycle, and select the corresponding maintenance decision based on the risk assessment results.

[0020] This invention utilizes an AI intelligent model to fuse and analyze multi-source real-time data, efficiently and accurately extracting real-time structural features reflecting the health status of the roadbed, providing a high-quality data foundation for condition assessment. By constructing an intelligent condition prediction model, it performs spatiotemporal correlation analysis between real-time conditions and historical data, and obtains extrapolated data based on maintenance decisions made in the previous cycle using a deep learning intelligent model. By summarizing various data from multiple iterations and optimizations, it can accurately predict the roadbed condition evolution trend for the next inspection cycle. Based on this, the system automatically performs risk assessment, significantly improving the scientific rigor, accuracy, and timeliness of maintenance work, and providing intelligent protection for the long-term safety and durability of the roadbed structure.

[0021] Preferably, the historical data includes historical testing data, historical maintenance data, and historical stress data; the project overview data includes geological environment data and dimensional parameters of the pile-slab structure.

[0022] It should be noted that the inspection image data is a two-dimensional image, such as an RGB image captured by a drone during inspection. Laser point cloud data, on the other hand, is a three-dimensional image, a discrete set of points generated by a 3D scanning device. Each point contains at least three-dimensional coordinate information (XYZ), and may also include color (RGB) or reflectance intensity attributes. This image data can macroscopically, intuitively, and quickly cover a large area, directly describing the changes in the position, shape, and orientation of the pile-slab structure in space, and easily detecting visible surface defects such as cracks.

[0023] Contact measurement data can be directly applied to the structural surface or embedded within the structure to directly measure minute structural changes. It is unaffected by light or weather conditions and can be collected in real-time 24 / 7. In this embodiment, the measurement includes physical quantities such as strain and vibration of pile-slab subgrades. This embodiment employs fiber optic grating sensors to directly measure subgrade strain, acquiring micro-strain within the concrete or reinforcing steel to directly reflect the structure's stress level and load response. Vibration sensors are used to directly measure subgrade vibration, monitoring vibrations caused by traffic loads, construction, etc., for analyzing fatigue effects and dynamic responses. Tilt sensors are used to detect changes in the verticality of the pile-slab subgrade. Changes in environmental temperature and humidity cause thermal expansion and contraction of materials and changes in moisture content; temperature and humidity data can reflect changes in structural stress and deformation.

[0024] Geological environmental data, including soil properties and groundwater, are the fundamental geological factors that cause subsidence.

[0025] Image data, being an indirect measurement method, has limited accuracy and is greatly affected by the environment; contact measurements are typically point-based, only acquiring local data from the sensor's installation location, making it difficult to reflect the overall deformation field of the entire cross-section. Therefore, this invention processes and fuses these multi-source, multi-type data through an AI intelligent model. By complementing and verifying multi-level, multi-angle information, the fused data provides both continuous time-series data and the ability to acquire high-precision spatial details when needed, achieving full coverage in the spatiotemporal dimensions, reducing interference from external factors, and lowering the system's false alarm rate.

[0026] Specifically, step S2, which involves analyzing and fusing the multi-source real-time data based on an AI intelligent model to extract real-time structural feature data of the pile-slab subgrade, includes: S21, Project the pre-created offline 3D model template of the pile column onto the space where the real-time laser point cloud is located, perform point cloud registration based on the iterative nearest point algorithm, and obtain the 3D contour description of the pile column in the real-time laser point cloud; First, the location of the pile needs to be determined in the 3D point cloud. In one specific implementation, step S21 can be achieved through the following steps: After the construction of the pile-slab roadbed is completed, the piles are scanned from multiple angles using a ground-based 3D laser scanner or an airborne LiDAR to obtain a multi-site cloud covering the entire surface of the piles.

[0027] The point clouds from multi-station scans are unified into the same coordinate system using an iterative nearest point algorithm, and noise points and outliers generated during the scanning process are removed.

[0028] The Poisson reconstruction algorithm is used to transform the point cloud into a continuous mesh model, creating a high-precision 3D template model of the pile and storing it in system memory. Specifically, an indicator function is estimated from the point cloud, and the isosurface of this function is the reconstructed pile surface.

[0029] During each monitoring cycle, the system needs to process a large amount of real-time data for the current frame. In order to obtain precise details of the pile structure, an initial, roughly accurate estimate of the pile position and attitude is required, followed by further optimization to obtain the true real-time pose.

[0030] The iterative nearest point algorithm is used to register the current sparse laser point cloud with a fine 3D contour description. The iterative calculation minimizes the error between the two point clouds, thus obtaining the 3D contour description of the pile in the real-time laser point cloud.

[0031] S22, based on the projection of the 3D contour description in the inspection image and the gradient change of the inspection image, the laser point cloud and the piles in the inspection image are matched, and the real-time true pose of the piles is obtained through iterative optimization.

[0032] Due to the sparse nature of real-time laser point cloud data (the laser point cloud acquired in the current frame is usually sparse, for example, it may only scan one side of the pile), it cannot fully reflect the structural changes of the pile. At the same time, the surface of concrete piles is usually smooth with few texture features and simple and repetitive geometric shapes, which makes it easy for matching algorithms based on traditional feature points (such as SIFT, ORB) to fail, making it difficult to reconstruct a precise model structure from single-dimensional image data.

[0033] Therefore, this invention considers fusing 3D laser point cloud data and 2D inspection image data. The 3D point cloud provides depth information, and the high-resolution information of the 2D image is fused to improve the recognition accuracy of weak textured piles. This allows for the differentiation of objects with the same geometric shape but different surfaces, and can obtain the accurate three-dimensional deformation of the piles and the corresponding surface appearance.

[0034] In one specific implementation, see Figure 3 Step S22 can be implemented through the following steps: S221, the 3D contour description is projected onto the inspection image to obtain the predicted pile pose, and a binary mask of the inspection image with respect to the predicted pile pose is generated.

[0035] It should be understood that the predicted pile pose involves mapping the 3D contour description of the pile to the generated pile position in the 2D inspection image. The GPS on the inspection drone provides its own latitude, longitude, and altitude, allowing the image-capturing device's own pose and the lidar's own pose to be obtained. Through coordinate transformation between the lidar point cloud and the inspection image, and by using sensor extrinsic parameters, the approximate position and orientation of the pile in the current 2D image sensor coordinate system are calculated. The 3D contour description is then projected onto the inspection image as the predicted pile pose.

[0036] In image processing, a mask is a binary image used to specify a region of interest in an image, for selecting, filtering, or manipulating specific areas of the image.

[0037] In this embodiment, the white area (value 1) of the binary mask image is the inside of the predicted pile pose (i.e., the "foreground"), and the black area (value 0) is the outside of the predicted pile pose (i.e., the "background"). The boundary between the two is the predicted pile pose.

[0038] S222, calculate the minimum distance from each pixel in the binary mask to the predicted pose of the pile in the inspection image, wherein the sign of the minimum distance in different binary value regions is inverted.

[0039] Considering that the pile and its surrounding environment (soil, concrete foundation) may have similar colors and weak texture, this invention extracts the geometric features of the foreground provided in the 3D laser point cloud image through step S222. Specifically, in the minimum distance calculation, the minimum distance value outside the contour is set to positive, and the minimum distance value inside is set to negative.

[0040] The geometry of the piles is regular, typically cylindrical or square, with continuous edges. To ensure the continuity of the overall edges, global geometric information of the entire profile is encoded through a minimum distance.

[0041] S223, extracts the grayscale gradient magnitude of each pixel in the inspection image based on the gradient operator.

[0042] The acquired 2D inspection images are preprocessed to calculate their grayscale gradient magnitude. In one specific implementation, the Sobel operator can be used to extract the intensity of the true edges provided in the 2D inspection images. Given the global geometric information, the contour information of the piles is then comprehensively determined by combining local edge intensity information.

[0043] S224, construct the joint observation probability of the pixel based on the gray-level gradient magnitude of the pixel and the minimum distance, and calculate the feature matching loss function between the predicted pile pose and the inspection image based on the joint observation probability.

[0044] The joint observation probability of a pixel is the probability that a pixel belongs to the foreground or background, calculated by combining three-dimensional point cloud data and two-dimensional inspection image data.

[0045] As a preferred implementation, the joint observation probability of pixels can be expressed by the following formula: in, Represents pixels in a binary mask The minimum distance to predict the pose of the pile in the inspection image; This represents a smoothing function that can smoothly determine pixel points based on the minimum distance value. The probability of belonging to the "prospect".

[0046] In this embodiment, the smoothing function Adopt the following form: It can adapt to edge blurring caused by subtle jitter, enabling accurate foreground and background determination of laser point cloud contour descriptions. Among these features, Represents the hyperbolic tangent function. For amplitude, The slope.

[0047] This indicates the probability that the pixel is both foreground and background in both 3D and 2D. This indicates the probability that the pixel is both background in 3D and 2D. The joint observation probability is maximized when the predicted stake pose in the point cloud perfectly coincides with the actual edge in the inspection image, which is the required precise matching state.

[0048] The foreground gradient model for the inspected image is as follows: in, This represents the normalized grayscale gradient magnitude; This is a smoothing factor.

[0049] The foreground gradient model is used to describe the probability distribution function of the gradient in the foreground region of an inspected image. This represents the gradient of pixels in the foreground region. This represents the foreground region of the image.

[0050] When gradient When the gradient approaches 1 (strong edges), the function value is close to 1, meaning that in the foreground region (near the edge of the post), the gradient should be as large as possible. The gradient at a certain point in the image... When the foreground gradient model is zero (e.g., on a uniform wall), log(0) is undefined and tends towards negative infinity when calculating the loss function, which can cause program crashes or result in meaningless values ​​in numerical computation. Therefore, a smoothing factor is set. To prevent the foreground gradient model from becoming zero, a small probability value is maintained even in regions without gradient, smoothly representing the possibility that each pixel might be a foreground pillar. That is, when the gradient... When the function value is 0, the function value is very small but not zero. .

[0051] The background gradient model for the inspected images is as follows: in, This represents the normalized grayscale gradient magnitude, ranging from [0, 1]. This is a smoothing factor.

[0052] The background gradient model is used to describe the probability distribution function of the gradient in the background region of an inspected image. This represents the gradient of pixels in the background region. This represents the background region of the image. When the gradient... When the gradient approaches 0 (in the smooth region), the function value approaches 1. When the gradient is 1, the function value is minimized, meaning that in the background region, the smaller the gradient, the better.

[0053] In this embodiment, negative log-likelihood estimation is used to describe the feature matching loss function, i.e. in, This refers to the predicted pile pose obtained by projecting the 3D contour description onto the inspection image. Indicates the predicted pile position The minimum distance is calculated below. The meanings of other parameters are the same as those in the joint observation probability formula for pixels, and will not be repeated here.

[0054] When the 3D contour description perfectly coincides with the real edges in the inspection image, for each pixel... : Pixels near the outline (foreground) Approaching 1, and gradient The probability is relatively large, so the foreground probability plays a dominant role in the joint observation probability.

[0055] In pixels far from the outline (background). Approaching 0, and gradient Since the probability is relatively small, the background probability plays a dominant role in the joint observation probability.

[0056] To maximize the overall likelihood probability, the true pose that best matches the 3D contour description with the real edge is the negative log-likelihood estimate, at which point the contour description best coincides with the edge in the image.

[0057] This invention integrates the distance geometric features of 3D laser point cloud data and the physical gradient features of 2D inspection image data to measure whether a pixel is foreground (i.e., a pile) or background, and measures the degree of matching between the edges in the inspection image and the contour description obtained from the point cloud. The feature matching loss function is to minimize the distance between the actual edge and the contour description, so that the 3D image and the 2D image can be accurately matched.

[0058] S225, iteratively calculate the minimum value of the feature matching loss function, and output the offset update amount of the predicted pile pose in the inspection image when the feature matching loss function reaches the minimum value.

[0059] Optimize prediction of pile position The process involves continuously adjusting the projection position of the 3D contour description in the inspection image so that its contour coincides with the actual edge of the pile in the image.

[0060] In practical implementation, iteratively calculating the minimum value of the feature matching loss function allows the feature matching loss function to be expressed in least squares form. The regression parameters are then iteratively solved using a nonlinear regression model, such as the Levenberg-Marquardt (LM) method. Real-world pile and column engineering images inevitably contain noise (such as variations in lighting and sensor noise) and outliers (such as weeds or cracks mistakenly detected as edges). The damping mechanism of the LM method makes it less sensitive to outliers, resulting in a more stable optimization process.

[0061] The offset update amount can be obtained based on the LM method. .

[0062] In the formula, This represents the Hessian matrix obtained by the LM method. Indicates the damping coefficient. Represents the coefficient matrix. This represents the Jacobian matrix obtained by the LM method. For the specific solution process, please refer to existing LM methods; this invention is not limited thereto.

[0063] S226, Adjust the projection position of the 3D contour description in the inspection image based on the offset update amount to obtain the real pose image of the pile, and output the real pose of the pile in real time.

[0064] The predicted pile pose is adjusted by translation and rotation based on the offset update, so that the features of the 3D image data can match the features of the 2D image data, thereby obtaining the true real-time pile pose. It should be understood that the output real-time true pose of the pile is the three-dimensional coordinates of the pile feature points.

[0065] S23 inputs real-time pose, contact measurement data, and temperature and humidity data into the AI ​​intelligent model for data fusion, and outputs real-time structural characteristic data of pile-slab roadbed.

[0066] This invention integrates the advantages of both 3D and 2D images, closely linking the 3D geometric shape information of the pile with the edge information of the 2D image. Minimum distance information provides absolute geometric constraints, clearly defining the internal / external position information of the pile, limiting the algorithm's search range to a reasonable geometric space, and preventing the image algorithm from losing its way due to texture loss. Gradient magnitude provides local, high-resolution edge responses, accurately indicating the true edge strength information in the image. The combination of these two methods effectively solves the recognition accuracy problem caused by weak texture on concrete surfaces, avoiding interference from changes in lighting, surface contamination, and motion blur in single-dimensional data. By pursuing the minimization of the matching loss function through an iterative optimization process, it can provide higher accuracy pose estimation than traditional feature point matching, thus keenly capturing subtle vertical displacement deviations and settlement.

[0067] The real-time structural feature data includes pile top coordinates, pile bottom coordinates, stress conditions, verticality data, strain data, settlement data, and structural deformation data. This data is then compared with historical data and the dimensional parameters of the pile-slab structure to perform verticality and settlement detection. Verticality detection includes angle deviation detection and displacement deviation detection; specifically, the angle deviation is measured by the pile inclination angle, and the displacement deviation is measured by the pile top coordinate deformation. Using the pile top coordinates (x0, y0, z0) as marker points, assuming a total of n pile marker points, the initial parameters for each pile i are... , For tilt sensor data, This represents the pile length under healthy baseline conditions. The parameters for this pile point at the corresponding testing time t-th testing cycle are: ;when and When the inclination angle is 1, only settlement occurs for pile i; when the inclination angle is 1, the settlement occurs for pile i. And the deformation of the pile top coordinates At that time, the corresponding pile i has a tilt; when the tilt angle And the deformation of the pile top coordinates At that time, pile i simultaneously experienced settlement and tilting. The deformation of the pile top coordinates... .

[0068] Thus, the first... Real-time status data of the detection cycle ,in, This is the vertical angle deviation. This refers to the verticality displacement deviation, which is the aforementioned deformation of the pile top coordinates. For settlement, Settlement rate. Precise location visualization of the pile-slab structure is achieved using computer vision, such as... Figure 4 As shown, the system can output results such as a heat map of the verticality of the pile-slab structure and an early warning of verticality exceeding limits. In the figure, the four piles of the pile-slab foundation are numbered 1#, 2#, 3#, and 4#, respectively. Indicates the inclination angle of the pile. This refers to the verticality displacement deviation.

[0069] As a preferred implementation, the state intelligent prediction model in step S3 is: In the formula, Indicates based on the first The optimal estimate of the detection cycle for the first Prediction of the condition of pile-slab subgrade during the inspection cycle; Indicates the first Real-time status data during the detection cycle; This represents the control input matrix for real-time status data. Indicates based on the first Historical data of the detection cycle for the first The iterative optimization data of the roadbed state during the detection cycle, that is, based on the optimal state estimate of the previous cycle, a theoretical prediction value of the current cycle state is calculated through the physical evolution model of the system. This prediction value is the best guess of the current state before the fusion of new sensor data. Represents the state transition matrix of historical data; Indicates the first Data calculated by the intelligent model after maintenance and repair during the inspection cycle; Represents the calculation matrix; Indicates the first Deviation control matrix for the detection cycle.

[0070] This invention integrates information from three different sources and time scales: real-time detection data, historical iterative optimization data, and intelligent model-derived data. This significantly overcomes the limitations of single data sources or single models, greatly improving the accuracy and reliability of predicting the future state of complex roadbed systems. By embedding the long-term trend prediction capability of the AI ​​intelligent model into short-term state prediction based on the physical model, it can actively receive external input (real-time detection data) and utilize auxiliary information derived from the intelligent model (deep learning prediction) to correct predictions, achieving an effective combination of short-term physical models and long-term data-driven models.

[0071] As a preferred implementation, the data calculated by the intelligent model is: In the formula, Indicates the first Real-time status data during the detection cycle. Indicates the first The adjustment matrix of the detection cycle is used to learn and extract features from historical maintenance data, and to estimate the system state after the previous maintenance cycle. Indicates the first The deviation control matrix for the detection cycle is used to correct for deviations and uncertainties in the calculation process.

[0072] Roadbed verticality detection and settlement prediction involve multiple highly nonlinear factors such as soil creep and groundwater changes, which traditional physical models struggle to accurately describe. This invention, by adjusting the AI ​​model represented by the matrix, can better capture and fit these complex nonlinear relationships, providing more accurate trend-based projection data. The AI ​​model can be independently optimized and updated without altering the core state prediction framework, improving the system's maintainability.

[0073] As a preferred embodiment, see [link to previous document]. Figure 5 The risk assessment of the pile-slab subgrade status for the next inspection cycle involves using real-time status data of the pile-slab subgrade from different inspection cycles as input to train and optimize a neural network model for short-term and long-term risk trend prediction. Maintenance decisions are made based on the assessment results, providing corresponding measures: high risk requires emergency measures (48 hours), medium risk requires planned maintenance (within 2 weeks), and low risk requires routine maintenance. For different risk levels, the system can provide information on required manpower, equipment, budget, and construction period based on the project scale for user selection and reference. In this embodiment, the risk assessment levels are shown in Table 1. It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0074] Based on the same inventive concept, this application also provides a system for implementing the AI-based pile-slab subgrade detection and predictive maintenance described above. The solution provided by this system is similar to the implementation scheme described in the above method. Therefore, the specific limitations in the embodiments of the AI-based pile-slab subgrade detection and predictive maintenance system provided below can be found in the limitations of the method described above, and will not be repeated here.

[0075] like Figure 6 As shown, the present invention also provides an AI-based pile-slab subgrade detection and predictive maintenance system, comprising: The data acquisition module is used to acquire historical data, multi-source real-time data, and project overview data of the pile-slab roadbed in response to a preset detection cycle; the multi-source real-time data includes inspection image data, laser point cloud data, contact measurement data, and temperature and humidity data. The AI ​​analysis module is used to analyze and fuse the multi-source real-time data based on the AI ​​intelligent model to extract real-time structural feature data of the pile-slab roadbed; and to obtain real-time status data of the pile-slab roadbed based on the real-time structural feature data and the project overview data. The intelligent prediction module is used to establish a state intelligent prediction model, taking the real-time state data and the historical data as input, to predict the state of the pile-slab subgrade in the next detection cycle. The maintenance decision module is used to conduct a risk assessment of the condition of the pile-slab subgrade in the next inspection cycle, and select and execute maintenance decisions based on the risk assessment results.

[0076] It should be noted that the AI-based pile-slab subgrade detection and predictive maintenance system provided in this embodiment is only illustrated with the above-mentioned functional module division when dealing with subgrade maintenance problems. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the system can be divided into different functional modules. Each functional module can be composed of a single execution unit, or two or more execution units can be integrated into one functional module to realize all the functions of that module.

[0077] Those skilled in the art will understand that the above modules can be implemented in whole or in part through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0078] This invention is not limited to the specific embodiments described above. Any modifications made by those skilled in the art based on the above concept without creative effort are within the scope of protection of this invention.

Claims

1. A method for detection and predictive maintenance of pile-slab subgrade based on AI analysis, characterized in that, Includes the following steps: In response to a preset detection cycle, historical data, multi-source real-time data, and project overview data of the pile-slab roadbed are acquired; the multi-source real-time data includes inspection image data, laser point cloud data, contact measurement data, and temperature and humidity data. Based on the AI ​​intelligent model analysis and fusion of the multi-source real-time data, real-time structural feature data of the pile-slab roadbed is extracted; real-time status data of the pile-slab roadbed is obtained based on the real-time structural feature data and the project overview data. A state-intelligent prediction model is established, using the real-time state data and the historical data as inputs, to predict the state of the pile-slab subgrade in the next detection cycle; A risk assessment is conducted on the condition of the pile-slab subgrade for the next inspection cycle, and maintenance decisions are selected based on the risk assessment results.

2. The method for detection and predictive maintenance of pile-slab subgrade based on AI analysis according to claim 1, characterized in that, The historical data includes historical testing data, historical maintenance data, and historical stress data; the project overview data includes geological environment data and dimensional parameters of the pile-slab structure.

3. The method for detection and predictive maintenance of pile-slab subgrade based on AI analysis according to claim 2, characterized in that, The real-time structural feature data of the pile-slab subgrade extracted by analyzing and fusing the multi-source real-time data based on the AI ​​intelligent model includes: The pre-created offline 3D model template of the pile is projected onto the space of the real-time laser point cloud. Point cloud registration is performed based on the iterative nearest point algorithm to obtain the 3D contour description of the pile in the real-time laser point cloud. Based on the projection of the 3D contour description into the inspection image and the gradient change of the inspection image, the laser point cloud and the piles in the inspection image are matched, and the real-time true pose of the piles is obtained through iterative optimization. Real-time pose, contact measurement data, and temperature and humidity data are input into the AI ​​intelligent model to perform data fusion and output real-time structural feature data of pile-slab roadbed.

4. The method for detection and predictive maintenance of pile-slab subgrade based on AI analysis according to claim 3, characterized in that, The process of matching the laser point cloud with the piles in the inspection image based on the projection of the 3D contour description into the inspection image and the gradient changes of the inspection image, and iteratively optimizing to obtain the real-time true pose of the piles includes: The 3D contour description is projected onto the inspection image to obtain the predicted pile pose, and a binary mask of the inspection image with respect to the predicted pile pose is generated. Calculate the minimum distance from each pixel in the binary mask to the predicted pose of the pile in the inspection image, wherein the sign of the minimum distance is inverted for different binary value regions; Extract the grayscale gradient magnitude of each pixel in the inspection image based on the gradient operator; The joint observation probability of a pixel is constructed based on the gray-level gradient magnitude and the minimum distance of the pixel, and the feature matching loss function between the predicted pile pose and the inspection image is calculated based on the joint observation probability. Iteratively calculate the minimum value of the feature matching loss function, and output the offset update amount of the predicted pile pose in the inspection image when the feature matching loss function reaches the minimum value; The projection position of the 3D contour description in the inspection image is adjusted based on the offset update amount to obtain the real pose image of the pile, and the real pose of the pile is output.

5. The method for detection and predictive maintenance of pile-slab subgrade based on AI analysis according to claim 4, characterized in that, The joint observation probability of a pixel constructed based on the grayscale gradient magnitude of the pixel and the minimum distance is specifically as follows: in, Represents pixels in a binary mask The minimum distance to predict the pose of the pile in the inspection image; This represents a smoothing function used to describe the minimum distance. With pixels The probability distribution function belonging to the foreground region; This is a foreground gradient model for inspected images, used to describe the probability distribution function of the gradient in the foreground region of the inspected image. This represents the gradient of pixels in the foreground region. Indicates the foreground region of the image; This is a background gradient model for inspection images, used to describe the probability distribution function of the gradient in the background region of the inspection image. This represents the gradient of pixels in the background region. This represents the background area of ​​the image.

6. The method for detection and predictive maintenance of pile-slab subgrade based on AI analysis according to claim 5, characterized in that, The foreground gradient model is specifically as follows: The background gradient model is specifically as follows: in, This represents the normalized grayscale gradient magnitude; This is a smoothing factor.

7. The method for detection and predictive maintenance of pile-slab subgrade based on AI analysis according to claim 1, characterized in that, The state intelligent prediction model is as follows: In the formula, Indicates based on the first The optimal estimate of the detection cycle for the first Prediction of the condition of pile-slab subgrade during the inspection cycle; Indicates the first Real-time status data during the detection cycle; The control input matrix represents the real-time status data. Indicates based on the first Historical data of the detection cycle for the first Iterative optimization data for detecting the subgrade condition during the detection cycle; Represents the state transition matrix of historical data; Indicates the first Data calculated by the intelligent model after maintenance and repair during the inspection cycle; Represents the calculation matrix; Indicates the first Deviation control matrix for the detection cycle.

8. The method for detection and predictive maintenance of pile-slab subgrade based on AI analysis according to claim 7, characterized in that, The first Real-time status data of the detection cycle ,in, This is the vertical angle deviation. This refers to the verticality displacement deviation. For settlement, Settlement rate; The data calculated by the intelligent model is as follows: In the formula, Indicates the first Real-time status data during the detection cycle; Indicates the first The adjustment matrix for the detection cycle; Indicates the first Deviation control matrix for the detection cycle.

9. The method for detection and predictive maintenance of pile-slab subgrade based on AI analysis according to claim 1, characterized in that, The risk assessment of the pile-slab subgrade status for the next inspection cycle involves using real-time status data of the pile-slab subgrade from different inspection cycles as input to train and optimize a neural network model for short-term and long-term risk trend prediction, and then implementing maintenance decisions based on the risk level: emergency measures for high-risk levels, planned maintenance for medium-risk levels, and routine maintenance for low-risk levels.

10. A pile-slab subgrade detection and predictive maintenance system based on AI analysis, characterized in that, include: The data acquisition module is used to acquire historical data, multi-source real-time data, and project overview data of the pile-slab roadbed in response to a preset detection cycle; the multi-source real-time data includes inspection image data, laser point cloud data, contact measurement data, and temperature and humidity data. The AI ​​analysis module is used to analyze and fuse the multi-source real-time data based on the AI ​​intelligent model to extract real-time structural feature data of the pile-slab roadbed; and to obtain real-time status data of the pile-slab roadbed based on the real-time structural feature data and the project overview data. The intelligent prediction module is used to establish a state intelligent prediction model, taking the real-time state data and the historical data as input, to predict the state of the pile-slab subgrade in the next detection cycle. The maintenance decision module is used to conduct a risk assessment of the condition of the pile-slab subgrade in the next inspection cycle, and select and execute maintenance decisions based on the risk assessment results.

Images