Uplift pile bearing capacity nondestructive testing system and method

Abstract

The invention provides a nondestructive testing system and method for the bearing capacity of an uplift pile, and the system comprises a strain sensing unit which comprises a distributed annular fiber grating array disposed on a target uplift pile and is used for detecting the strain information of the target uplift pile on the whole pile length; the displacement monitoring unit comprises an MEMS inclinometer and a laser range finder which are arranged at the top of the pile, and an LVDT displacement meter pre-embedded in a pile-soil interface, and is used for synchronously monitoring absolute displacement and relative slippage of the target uplift pile; and the processor unit is used for carrying out inversion calculation on the bearing capacity of the target uplift pile by adopting an FPGA accelerated LSTM-GRU hybrid neural network based on the strain information, the absolute displacement and the relative slippage. According to the method, the differential coupling model of the axial strain of the pile body, the relative displacement of the pile soil and the side friction resistance is established, accurate mapping from the distributed optical fiber strain to the bearing capacity subitem index is achieved, therefore, the nondestructive testing target is achieved, the detection cost is low, and the detection accuracy and reliability are high.

Description

Technical Field

[0001] This invention relates to the field of civil engineering technology, and in particular to a non-destructive testing system and method for the bearing capacity of pull-out piles. Background Technology

[0002] In civil engineering construction, tension piles are key components of deep foundation engineering. Their performance testing, especially bearing capacity testing, has become a core issue in ensuring the buoyancy safety of super high-rise buildings, the reliability of anchorage systems for cross-sea bridges, and the stability of offshore wind power platforms. Traditional static load testing methods were once considered the "gold standard" for bearing capacity testing, but their destructive nature does not conform to the concept of sustainable development in modern engineering, and the testing cost is high, making it difficult to quantify bearing capacity without affecting the normal use of the structure.

[0003] The rise of distributed fiber optic sensing technology once brought revolutionary hope to pile foundation testing, but its engineering applications in tensile pile scenarios have revealed deep-seated technical defects. First, strain measurements based on Brillouin scattering (BOTDR) are susceptible to interference from multiple environmental factors. Second, existing technologies directly equate fiber strain with pile displacement, ignoring the physical nature of relative slippage at the pile-soil interface; therefore, the accuracy and reliability of the test results cannot be guaranteed. Summary of the Invention

[0004] This invention provides a non-destructive testing system and method for the bearing capacity of pull-out piles, which addresses the shortcomings of existing technologies in terms of accuracy and reliability, and aims to effectively improve the accuracy and reliability of testing.

[0005] This invention provides a non-destructive testing system for the bearing capacity of pull-out piles, comprising: a strain sensing unit, a displacement monitoring unit, and a processor unit, wherein the processor unit is connected to the strain sensing unit and the displacement monitoring unit respectively; The strain sensing unit includes a distributed annular fiber Bragg grating (FBG) array arranged on the target tensile pile according to a preset arrangement scheme, used to detect the strain information of the target tensile pile along the entire pile length. The displacement monitoring unit includes a MEMS inclinometer and a laser rangefinder arranged on the top of the target pull-out pile, and an LVDT displacement meter pre-embedded at a preset density at the pile-soil interface, used to simultaneously monitor the absolute displacement and relative slippage of the target pull-out pile. The processor unit is used to calculate the bearing capacity of the target pull-out pile by using an FPGA-accelerated LSTM-GRU hybrid neural network based on the strain information, the absolute displacement, and the relative slip.

[0006] According to the present invention, a non-destructive testing system for the bearing capacity of pull-out piles is provided, wherein the processor unit is further configured to: Based on the strain information and the absolute displacement, the variable cross-section elastic mechanical equation corresponding to the target tensile pile is established to characterize the strain-displacement conversion relationship. Based on the absolute displacement and the relative slippage, a pile-soil interface contact mechanics model is established to separate the absolute displacement of the target pull-out pile from the relative slippage between the pile and the soil. Based on the relationship between the relative slip and the layered skin friction of the target pull-out pile, an optimization model with L1-L2 mixed norm constraint is established by constructing an optimization objective function with L1-L2 mixed norm constraint and combining dynamic correction of soil parameters. The L1 norm penalty term represents the bearing layer, and the L2 norm smoothing term is used to suppress spurious fluctuations. Based on the variable cross-section elasticity equation, the pile-soil interface contact mechanics model, and the optimization model with L1-L2 mixed norm constraints, an LSTM-GRU hybrid neural network model is initialized. The strain gradient time series features of the finite element simulation data are extracted through the initialized LSTM-GRU hybrid neural network model. The model is pre-trained and then transferred to FPGA parallel computing using field measured data to construct the FPGA-accelerated LSTM-GRU hybrid neural network.

[0007] According to the present invention, a non-destructive testing system for the bearing capacity of a tensile pile is provided, wherein the processor unit, when used for the inversion calculation of the bearing capacity of the target tensile pile, is used for: The temporal features of the strain gradient corresponding to the strain information are extracted by an LSTM network, and based on the temporal features of the strain gradient, combined with a GRU network, the spatial distribution law of the displacement derivative is captured. Based on the spatial distribution pattern, the layered frictional resistance of the target pull-out pile is calculated using an optimization model with L1-L2 mixed norm constraints and an FPGA parallel computing architecture, and the bearing capacity is determined based on the layered frictional force.

[0008] According to the present invention, a non-destructive testing system for the bearing capacity of pull-out piles is provided, wherein the processor unit is further configured to: Based on the continuous strain information, the cumulative sum statistic (CUSUM) is calculated using the sliding window technique, and the abrupt change characteristics of the strain gradient are dynamically analyzed based on the CUSUM and a preset threshold. Based on the aforementioned mutation characteristics and preset judgment conditions, the location of potential soil layer interfaces is identified.

[0009] The non-destructive testing system for the bearing capacity of pull-out piles provided by the present invention further includes a temperature and humidity sensing unit; The temperature and humidity sensing unit includes multiple temperature sensors and humidity sensors arranged alternately at a preset density along the axis of the target tensile pile, used to collect ambient temperature and humidity information around the axis of the target tensile pile. The processor unit is further configured to decouple temperature noise based on the ambient temperature information and the ambient humidity information, using wavelet envelope analysis and an improved Tikhonov regularization algorithm, and to correct the bearing capacity based on the temperature noise.

[0010] According to the present invention, a non-destructive testing system for the bearing capacity of a pull-out pile is provided, wherein the distributed ring fiber Bragg grating (FBG) array includes multiple fiber Bragg grating rings distributed along the axial direction of the target pull-out pile, and each fiber Bragg grating ring includes multiple fiber Bragg gratings symmetrically and uniformly distributed along the circumferential direction of the target pull-out pile.

[0011] According to the present invention, in a non-destructive testing system for the bearing capacity of a tensile pile, the distribution density of the fiber Bragg grating ring along the axial direction of the target tensile pile is increased at the soil layer interface, and / or, if the target tensile pile is a variable cross-section pile, the distribution density of the fiber Bragg grating in the fiber Bragg grating ring is increased along the cross-section change location of the target tensile pile.

[0012] According to the present invention, a non-destructive testing system for the tensile bearing capacity of a pile is provided, wherein the fiber Bragg grating in the distributed ring fiber Bragg grating (FBG) array is protected by a double-layer protective sleeve, and the double-layer protective sleeve is made by nano-injection molding process. The waterproof rating of the double-layer protective sleeve is not lower than IP68, and the tensile strength of the double-layer protective sleeve is ≥200MPa.

[0013] The non-destructive testing system for the bearing capacity of pull-out piles provided by the present invention further includes a LoRa / 5G dual-mode communication channel for transmitting feature data packets and / or full data streams based on the displacement state of the target pull-out pile.

[0014] The non-destructive testing system for the bearing capacity of pull-out piles provided by the present invention further includes a visualization unit, which is used to generate one or more of the following based on the feature data package and / or the full data stream: strain cloud map, frictional resistance bar chart, bearing capacity-displacement curve, and three-dimensional safety factor cloud map, and to perform three-dimensional visualization display.

[0015] According to the present invention, a non-destructive testing system for the bearing capacity of a pull-out pile is provided. If the target pull-out pile is an existing pile, the distributed ring fiber Bragg grating (FBG) array is installed using a minimally invasive implantation process. The minimally invasive implantation process includes: implanting the fiber Bragg grating through a 30mm borehole and injecting a nano-modified cement-based grout with matching elastic modulus to fill and seal the borehole.

[0016] The present invention also provides a non-destructive testing method for the tensile bearing capacity of piles based on any of the above-described non-destructive testing systems for tensile pile bearing capacity, comprising: The strain sensing unit is used to detect the strain information of the target pull-out pile along its entire length. The displacement monitoring unit is used to simultaneously monitor the absolute displacement and relative slippage of the target pull-out pile; Using the processor unit, based on the strain information, the absolute displacement, and the relative slip, the bearing capacity of the target tensile pile is calculated by inverting the LSTM-GRU hybrid neural network accelerated by FPGA.

[0017] The non-destructive testing system and method for tensile pile bearing capacity provided by this invention achieves accurate mapping from distributed optical fiber strain data to bearing capacity sub-indicators by establishing a differential coupling model of pile axial strain, pile-soil relative displacement and side friction, thereby achieving the goal of non-destructive testing of tensile pile bearing capacity, effectively reducing testing costs and improving testing accuracy and reliability. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0019] Figure 1 A schematic diagram of the non-destructive testing system for the tensile bearing capacity of piles provided by the present invention; Figure 2 This is a schematic diagram of the hardware architecture of the non-destructive testing system for the tensile pile bearing capacity provided by the present invention. Figure 3 This is a schematic diagram of the distribution of a distributed ring fiber Bragg grating array in the non-destructive testing system for the tensile bearing capacity of piles provided by the present invention. Figure 4 This is a schematic diagram of the data processing flow of the cascade inversion model of "strain-displacement-friction" in the system provided by the present invention; Figure 5 This is a framework diagram of the monitoring system in the non-destructive testing system for tensile pile bearing capacity provided by the present invention; Figure 6 This is a flowchart illustrating the non-destructive testing method for the tensile bearing capacity of piles provided by the present invention. Figure 7 This is a schematic diagram of the drilling and grouting reinforcement evaluation process for existing piles in the non-destructive testing method for the tensile bearing capacity of piles provided by the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of this invention. Obviously, the described embodiments are only some embodiments of this invention, not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0021] This invention addresses the shortcomings of existing technologies in terms of accuracy and reliability. By establishing a differential coupling model of pile axial strain, pile-soil relative displacement, and side skin friction, it achieves a precise mapping from distributed fiber optic strain data to bearing capacity sub-indicators. This enables non-destructive testing of the bearing capacity of tension piles, effectively reducing testing costs and improving accuracy and reliability. The invention will be further illustrated and described below through several specific embodiments.

[0022] Figure 1 This is a schematic diagram of the non-destructive testing system for the uplift pile bearing capacity provided by the present invention. This system can be used to achieve non-destructive testing of the uplift pile bearing capacity, such as... Figure 1 As shown, the system includes a strain sensing unit 101, a displacement monitoring unit 102, and a processor unit 103, wherein the processor unit 103 is connected to both the strain sensing unit 101 and the displacement monitoring unit 102. The strain sensing unit 101 includes a distributed ring fiber Bragg grating (FBG) array arranged on the target tensile pile according to a preset arrangement scheme, which is used to detect the strain information of the target tensile pile along the entire pile length. The displacement monitoring unit 102 includes a MEMS inclinometer and a laser rangefinder arranged on the top of the target pull-out pile, and an LVDT displacement meter pre-embedded at a preset density at the pile-soil interface, used to simultaneously monitor the absolute displacement and relative slippage of the target pull-out pile. The processor unit 103 is used to calculate the bearing capacity of the target pull-out pile by using an FPGA-accelerated LSTM-GRU hybrid neural network based on the strain information, the absolute displacement and the relative slip.

[0023] This can be understood as, for example Figure 1 As shown, the non-destructive testing system for the bearing capacity of pull-out piles of the present invention should include at least three components: a strain sensing unit 101, a displacement monitoring unit 102, and a processor unit 103, which are used to realize the processing flow of target pull-out pile strain information acquisition, displacement information acquisition, and bearing capacity calculation.

[0024] Specifically, such as Figure 2The diagram shows the hardware architecture of the non-destructive testing system for the bearing capacity of pull-out piles provided by the present invention. The sensing system includes a strain sensing unit 101 for strain sensing and a displacement monitoring unit 102 for displacement sensing. The specific hardware includes a distributed ring fiber Bragg grating (FBG) array arranged on the target pull-out pile according to a preset arrangement, a MEMS inclinometer and a laser rangefinder arranged on the top of the target pull-out pile, and an LVDT displacement meter pre-embedded at the pile-soil interface according to a preset density.

[0025] The strain sensing unit 101 monitors the strain along the entire length of the pile through a distributed ring-shaped fiber Bragg grating (FBG) array arranged on the target tensile pile. Optionally, the distributed ring-shaped FBG array includes multiple fiber Bragg grating rings distributed along the axial direction of the target tensile pile, and each fiber Bragg grating ring includes multiple fiber Bragg gratings symmetrically and uniformly distributed along the circumferential direction of the target tensile pile.

[0026] This can be understood as, for example Figure 3 The diagram shows the distribution of a distributed ring-shaped fiber Bragg grating array in the non-destructive testing system for tensile pile bearing capacity provided by the present invention. The distributed ring-shaped fiber Bragg grating array in the strain sensing layer is arranged symmetrically around the pile axis. That is, along the pile axis, multiple fiber Bragg gratings (3 in the diagram) are arranged at corresponding intervals (3 meters in the diagram), forming a ring on the same horizontal line, i.e., a fiber Bragg grating ring. The multiple fiber Bragg gratings in each fiber Bragg grating ring are arranged symmetrically around the pile, which can be a symmetrical and uniform arrangement. For example... Figure 3 The strain sensing unit in the system includes a double-layer ring fiber grating array with three circumferentially distributed nodes arranged every 3 meters. These nodes eliminate the influence of eccentric loads through differential measurement.

[0027] For example, for newly built piles, monitoring rings (i.e., fiber Bragg grating rings, which may contain four 90° evenly distributed FBGs) can be set every 500 mm; for the renovation of existing piles, monitoring rings (which may contain three 120° evenly distributed FBGs) can be set every 3 m.

[0028] Optionally, the distribution density of the fiber Bragg grating ring along the axial direction of the target tensile pile is increased at the soil layer interface, and / or, if the target tensile pile is a variable cross-section pile, the distribution density of the fiber Bragg grating in the fiber Bragg grating ring is increased along the cross-section variation of the target tensile pile.

[0029] This can be understood as follows: to further improve accuracy, the system in this invention optimizes the characteristics of pull-out piles. Specifically, it densifies the arrangement of monitoring rings at the soil layer interface, such as reducing the spacing between adjacent monitoring rings to 1.5m. For piles with variable cross-sections, dense points can be arranged along the cross-section change points, that is, increasing the number of fiber Bragg gratings contained in the monitoring rings to improve the distribution density.

[0030] Optionally, if the target pull-out pile is an existing pile, the distributed ring fiber Bragg grating (FBG) array is installed using a minimally invasive implantation process. The minimally invasive implantation process includes: implanting the fiber Bragg grating through a 30mm borehole and injecting a nano-modified cement-based grout with matching elastic modulus to fill and seal the borehole.

[0031] This invention can be understood as follows: when installing fiber Bragg gratings on existing piles, the present invention can employ either pre-embedded conduits or minimally invasive drilling implantation techniques. In the minimally invasive drilling implantation technique, the sensor chain is implanted through a 30mm borehole, and a nano-modified cement-based grout with matching elastic modulus (±5%) is injected. Simultaneously, to prevent corrosion, the fiber Bragg grating can be encapsulated and protected with a double-layer stainless steel sheath (e.g., IP68 protection).

[0032] In other words, for existing piles as the target tensile strength, this invention employs a minimally invasive sensor network implantation process to install fiber Bragg gratings. First, a 30mm diameter vertical hole is drilled using a diamond drill bit, penetrating the pile bottom and reaching at least 500mm into the bearing stratum. During drilling, three-dimensional laser positioning technology is used to avoid the main reinforcement bars, and the hole quality is monitored in real-time using an endoscopic imaging system. Second, after drilling, a distributed sensor chain pre-encapsulated in a flexible sheath is slowly implanted into the hole, followed by the injection of a cement-based grout with nano-silica modified properties. The elastic modulus of this grout is precisely adjusted to maintain a ±5% match with the pile concrete, ensuring coordinated deformation between the sensors and the pile structure.

[0033] The displacement monitoring unit 102 consists of a MEMS inclinometer (accuracy of 0.001°) integrated at the pile top, a laser rangefinder (resolution of ±0.05mm), and LVDTs pre-embedded every 5m at the pile-soil interface. Through data fusion, it achieves synchronous monitoring of absolute displacement and relative slippage. In other words, by integrating the MEMS inclinometer (0.001° accuracy) and the laser rangefinder, combined with the pre-embedded LVDT displacement gauges, it achieves synchronous capture of absolute and relative displacement.

[0034] The edge computing terminal includes a processor unit 103, which mainly uses information collected by the sensing system to perform real-time cascaded inversion calculations of strain-displacement-friction resistance using a built-in FPGA accelerator and an LSTM-GRU hybrid neural network. This ultimately determines the pile-soil displacement field and calculates the layered friction resistance. In other words, based on a deep learning architecture using an LSTM-GRU hybrid neural network, a transfer learning mechanism combines limited field test data with a large amount of numerical simulation data to obtain an FPGA-accelerated LSTM-GRU hybrid neural network model. This model is then used to perform cascaded inversion calculations of strain-displacement-friction resistance to obtain the final bearing capacity value, breaking through the accuracy limitations of traditional empirical formulas for predicting layered friction resistance.

[0035] The non-destructive testing system for the tensile strength of piles provided by this invention achieves precise mapping from distributed fiber optic strain data to bearing capacity sub-indicators by establishing a differential coupling model of pile axial strain, pile-soil relative displacement, and side skin friction. This achieves the goal of non-destructive testing of tensile strength pile bearing capacity while effectively reducing testing costs and improving accuracy and reliability. Furthermore, through a built-in FPGA accelerator, real-time analysis of a 20-meter pile length can be completed within 2 seconds, thus efficiently and accurately determining the real-time bearing capacity of the target tensile strength pile.

[0036] Optionally, in the non-destructive testing system for the bearing capacity of tension piles provided in the above embodiments, the processor unit is further configured to: establish, based on the strain information and the absolute displacement, a variable cross-section elastic mechanical equation corresponding to the target tension pile, to characterize the strain-displacement conversion relationship; establish a pile-soil interface contact mechanical model based on the absolute displacement and the relative slippage, to separate the absolute displacement of the target tension pile from the relative slippage between the pile and the soil; and establish, based on the relationship between the relative slippage and the layered skin friction of the target tension pile, an optimization objective function constrained by L1-L2 mixed norm is constructed, combined with dynamic correction of soil parameters, to establish... An optimization model with L1-L2 hybrid norm constraints is used, where the L1 norm penalty term represents the bearing layer and the L2 norm smoothing term is used to suppress spurious fluctuations. Based on the variable cross-section elasticity equation, the pile-soil interface contact mechanics model, and the optimization model with L1-L2 hybrid norm constraints, an LSTM-GRU hybrid neural network model is initialized. The strain gradient time series features of the finite element simulation data are extracted through the initialized LSTM-GRU hybrid neural network model. The model is pre-trained, and transfer learning is performed using FPGA parallel computing in combination with field measured data to construct the FPGA-accelerated LSTM-GRU hybrid neural network.

[0037] It can be understood that before the present invention is put into practical application, it is necessary to first establish and train an LSTM-GRU hybrid neural network model for calculation. Specifically, through the cross-integration of multiple disciplines, an innovative solution is proposed, namely, to construct a cascaded inversion LSTM-GRU hybrid neural network model of "strain-displacement-friction resistance" to realize the accurate conversion of distributed strain to the real-time bearing capacity of the tensile pile.

[0038] Based on the cascade inversion concept of "strain-displacement-friction," the governing equations considering the variable cross-section of the pile can be derived theoretically first. This involves constructing the initialization network for the LSTM-GRU hybrid neural network model, such as... Figure 4 The diagram shown illustrates the data processing flow of the cascade inversion model of "strain-displacement-friction" in the system provided by the present invention. The operation process is as follows: First, a strain-displacement conversion is performed. Based on the variable cross-section elasticity equation (Equation a), a mathematical mapping relationship between the strain of the distributed optical fiber and the axial displacement field of the pile is established.

[0039] (a) In the formula, ρ Represents the density gradient of the material. dA / dz Indicates the rate of change of cross section. E Indicates the elastic modulus. ν Represents Poisson's ratio. r 0 represents the radius of the pile cross-section. τ ( z ) indicates a distance of z Shear stress at that time.

[0040] This invention is the first to incorporate material density gradient and cross-sectional change rate into displacement field calculations, significantly improving the modeling accuracy of piles with non-uniform cross-sections. The elastic modulus E and Poisson's ratio ν are dynamically calibrated through pile material experiments, solving the model mismatch problem caused by parameter fixation in traditional methods.

[0041] Secondly, displacement-relative slippage decoupling is performed. By introducing a pile-soil interface contact mechanics model, the absolute displacement of the pile body and the relative slippage between the pile and the soil are separated.

[0042] Optionally, the processor unit is further configured to: calculate the cumulative sum statistic CUSUM based on the continuous strain information using a sliding window technique, and dynamically analyze the abrupt change characteristics of the strain gradient based on the cumulative sum statistic CUSUM and a preset threshold; and identify the location of potential soil layer interfaces based on the abrupt change characteristics and preset judgment conditions.

[0043] This invention can be understood as employing an improved CUSUM mutation detection algorithm to identify soil interface layers in tension pile monitoring. Based on continuous strain data acquired by distributed fiber optic sensors, the algorithm dynamically analyzes the abrupt change characteristics of the strain gradient using a sliding window technique. The identification process is as follows: First, the pile strain sequence is divided into analysis windows of fixed length (e.g., 0.5m-1m), and the cumulative sum statistic (i.e., CUSUM value) is calculated within each window. The calculation expression is as follows: ; In the formula, Δϵk is the strain gradient change value in the current window, and μ and σ are the mean and standard deviation of the strain gradient in the historical window, respectively.

[0044] Secondly, when the cumulative sum Sk exceeds a preset threshold H (usually 3-5 times the standard deviation), abrupt change point detection is triggered. The algorithm mainly judges based on the abrupt change amplitude and duration of the strain gradient. For example, when the strain gradient change rate exceeds 0.15με / m for three consecutive windows, it is determined to be a suspected location of the soil interface, i.e., a potential soil interface location.

[0045] The preset threshold H adopts a dynamic adjustment strategy, that is, firstly, the initial value is preset according to the soil mechanical parameters (such as shear modulus and Poisson's ratio) in the geological survey report, and then the Bayesian update algorithm is used to optimize it in combination with real-time strain data during the monitoring process.

[0046] This invention employs the CUSUM mutation detection algorithm to automatically identify soil interfaces, thereby controlling the error to below 0.3 meters.

[0047] Next, slip-friction inversion is performed. By designing an optimization objective function with L1-L2 mixed norm constraints and combining it with dynamic correction of soil parameters, an optimization model with L1-L2 mixed norm constraints is constructed to achieve refined calculation of layered skin friction. The optimization objective function with L1-L2 mixed norm constraints is as follows: ; In the formula: L 1 represents the norm penalty term; L 2 represents the norm smoothing term; λ 1 is L 1. The weighting coefficient of the regularization term is used to control the sparsity of the solution and highlight the contribution of the main bearing layer; λ 2 is L 2. The weighting coefficients of the regularization term are used to control the smoothness of the solution and suppress spurious fluctuations in the inversion results; i meas For the first i Sensor strain measured in actual operation of the pile segment; imodel For the first i Strain of the segment pile model; τ ( z ) indicates a distance of z Shear stress at that time.

[0048] This invention highlights the main bearing layer by using an L1 norm penalty term and suppresses spurious fluctuations by using an L2 norm smoothing term, so that the layered friction inversion results have both physical rationality and numerical stability.

[0049] After completing the initial model setup described above, this invention optimizes the initial model through transfer learning. Specifically, it first pre-trains the model using finite element simulation data (e.g., >100,000 sets), and then performs transfer learning using field measured data (e.g., >500 sets), overcoming the limitations of traditional empirical formulas in adaptability to soil layers. During training and learning, the temporal characteristics of strain gradients are first extracted using an LSTM network, and combined with a GRU network to capture the spatial distribution of displacement derivatives. Simultaneously, based on an FPGA parallel computing architecture, computational efficiency is effectively improved, accelerating the model training process.

[0050] Optionally, in the non-destructive testing system for the bearing capacity of the pull-out pile provided by the above embodiments, the processor unit, when used for the inversion calculation of the bearing capacity of the target pull-out pile, is configured to: extract the temporal characteristics of the strain gradient corresponding to the strain information through an LSTM network, and capture the spatial distribution law of the displacement derivative based on the temporal characteristics of the strain gradient and in combination with a GRU network; calculate the layered frictional resistance of the target pull-out pile based on the spatial distribution law through an optimization model constrained by L1-L2 mixed norm and an FPGA parallel computing architecture, and determine the bearing capacity based on the layered frictional resistance.

[0051] As described in the above embodiments, this invention constructs a full-chain non-destructive testing scheme of "perception-decoupling-inversion-optimization," develops a deep learning architecture based on an LSTM-GRU hybrid neural network, and achieves the fusion of multi-scale features. Meanwhile, to improve efficiency, data processing employs an FPGA-accelerated LSTM-GRU hybrid neural network.

[0052] In practical applications, after real-time strain information of the target tension pile is acquired using a strain sensing unit, the temporal characteristics of the strain gradient are extracted through an LSTM network. Combined with a GRU network, the spatial distribution of the displacement derivative is captured. A hybrid L1-L2 norm optimization model is then used to calculate the layered skin friction. The L1 norm highlights the bearing layer, while the L2 norm suppresses fluctuations. Simultaneously, during data processing, real-time inversion acceleration is achieved based on an FPGA parallel computing architecture. This enables not only real-time analysis of strain information from 0.5 to 50 Hz with an inversion error ≤5%, but also real-time analysis of the skin friction distribution of a 20m pile length within 2 seconds, significantly improving efficiency.

[0053] Optionally, the non-destructive testing system for the bearing capacity of pull-out piles provided according to the above embodiments may also include a temperature and humidity sensing unit; the temperature and humidity sensing unit includes multiple temperature sensors and humidity sensors arranged alternately at a preset density along the axial direction of the target pull-out pile, used to collect ambient temperature information and ambient humidity information around the axis of the target pull-out pile; the processor unit is further used to decouple temperature noise based on the ambient temperature information and the ambient humidity information, using wavelet envelope analysis and an improved Tikhonov regularization algorithm, and to correct the bearing capacity based on the temperature noise.

[0054] To eliminate strain artifacts caused by temperature changes, the non-destructive testing system for the tensile strength of piles in this invention also includes a temperature and humidity sensing unit for environmental compensation. This unit consists of multiple temperature and humidity sensors arranged on the target tensile pile. For example, temperature and humidity sensors can be staggered every 10m along the axial direction of the target tensile pile. The temperature and humidity sensors can be embedded within the concrete protective layer 20-30mm from the pile surface, dynamically adjusting the thermal expansion coefficient based on the concrete's age to compensate for strain data in real time.

[0055] Specifically, the environmental compensation of this invention establishes a thermal expansion correction model through embedded temperature and humidity sensors, and its temperature decoupling formula is as follows: ,in ε ture This indicates the strain after temperature compensation. ε raw Represents the original strain, Δ T The coefficient of thermal expansion α represents the temperature change, and is dynamically corrected based on the concrete age. Simultaneously, by performing wavelet envelope analysis for noise reduction and an improved Tikhonov regularization algorithm, temperature noise is decoupled, the influence of temperature drift noise is eliminated, the variable cross-section elasticity equation is solved, and the pile displacement field is inverted. This effectively suppresses measurement drift during the hydration heat stage and dynamically corrects temperature interference.

[0056] Optionally, according to the above embodiments, to improve the robustness of monitoring and identification, the present invention establishes a comprehensive judgment mechanism, in which the algorithm integrates multi-dimensional data, including: (1) Strain-displacement joint analysis, that is, when potential strain mutation points are identified by the algorithm, the overlap between them and the pile-soil relative slip anomaly area measured by LVDT is judged. If they overlap, the confidence of the interface identification is enhanced.

[0057] (2) Environmental compensation verification, that is, by using temperature and humidity sensor data, strain artifacts caused by temperature changes are eliminated, and the temperature fluctuation at the abrupt change point is required to be <0.5℃ / h.

[0058] (3) Historical data association, that is, combining the soil interface database established by the previous static load test, the test results are corrected by the pattern matching algorithm so that the final interface position error is ≤ ±1.5m.

[0059] As the application scenarios of tension piles extend from land to sea, such as a deep-sea floating wind power project requiring the pile foundation testing system to have long-term self-powered monitoring capabilities at a water depth of 200 meters, higher requirements are placed on the waterproof rating and data transmission reliability of the testing equipment. Optionally, the fiber Bragg gratings in the distributed ring fiber Bragg grating (FBG) array are protected by a double-layer protective sleeve, which is manufactured using a nano-injection molding process. The waterproof rating of the double-layer protective sleeve is not lower than IP68, and the tensile strength of the double-layer protective sleeve is ≥200MPa.

[0060] In essence, to meet higher protection and transmission requirements, the fiber Bragg grating used for strain detection in this invention is encapsulated and protected with a double-layer protective sleeve. This protective sleeve can be made of materials with high tensile strength (e.g., tensile strength ≥200MPa) and a high protection rating, such as stainless steel, and is manufactured using a nano-injection molding sealing process. This waterproof shell, i.e., the double-layer protective sleeve, can achieve an IP68 or higher protection rating and supports operation in a wide temperature range from -20℃ to 60℃.

[0061] Optionally, the non-destructive testing system for the tensile pile bearing capacity provided in the above embodiments may further include a wireless transmission unit, which may be, for example, a LoRa / 5G dual-mode communication channel, for transmitting feature data packets and / or full data streams based on the displacement state of the target tensile pile.

[0062] This invention can be understood as follows: the entire monitoring system can be built on a B / S architecture IoT platform. By integrating low-power wide-area IoT (LPWAN) technology, i.e., low-power wireless transmission units, the front-end sensor network can upload data packets to the base station processor. Optionally, the base station equipment can be equipped with a dual-power redundant system to ensure continuous operation. The low-power wireless transmission unit of this invention can support LoRa / 5G dual-mode communication, realizing the transmission of feature data packets and / or full data streams, thereby forming a complete monitoring system from fiber optic sensor arrays, edge computing nodes to cloud platforms.

[0063] Optionally, the LoRa / 5G dual-mode communication channel supports a basic transmission mode and an emergency transmission mode. In the basic transmission mode, compressed feature data packets are uploaded at preset time intervals; in the emergency transmission mode, when a sudden displacement change in the target uplift pile is detected, the full data stream is uploaded in real time.

[0064] In other words, the LoRa / 5G dual-mode communication channel in this invention uploads local data based on the displacement state of the target uplift pile. When no displacement anomaly is detected, local feature data packets are uploaded regularly at preset time intervals, i.e., uploaded in the basic transmission mode. When a sudden change in the displacement of the target uplift pile is detected, a full local data stream related to the target uplift pile's state is uploaded to further analyze the cause of the displacement change, i.e., uploaded in the emergency transmission mode. For example, in the basic mode, compressed feature data packets are uploaded every 30 minutes; in the emergency mode, when a sudden change in displacement occurs, a full data stream is transmitted in real time.

[0065] Optionally, the non-destructive testing system for the tensile pile bearing capacity provided in the above embodiments may further include a visualization unit, which is used to generate one or more of the following based on the feature data package and / or the full data stream: strain cloud map, frictional resistance bar chart, bearing capacity-displacement curve, and three-dimensional safety factor cloud map, and to perform three-dimensional visualization display.

[0066] In order to facilitate the observation of the real-time status of the target tension pile and to conduct dynamic analysis, the system of the present invention also includes a visualization unit. The visualization unit includes a visualization interface that supports multi-dimensional data display, including real-time strain field animation demonstration, historical data trend comparison, and hierarchical push of early warning information. It can also display strain cloud map, frictional resistance bar chart and safety factor curve in real time, and can also generate bearing capacity-displacement curve and three-dimensional safety factor cloud map. It is suitable for the long-term operation and maintenance of existing piles and provides an intelligent solution for the full life cycle management of tension piles under complex geological conditions.

[0067] Optionally, the non-destructive testing system for the tensile bearing capacity of the pile of the present invention further includes a storage unit for storing historical time-series data of the target tensile pile and automatically generating an ISO18629 standard test report based on the historical time-series data.

[0068] This can be understood as, for example Figure 5 The diagram shown illustrates the monitoring system framework of the non-destructive testing system for tensile pile bearing capacity provided by this invention. Historical data is stored in a time-series database, and a pile foundation performance degradation prediction model is established using machine learning algorithms. All monitoring data conforms to the ISO 18629 standard and can automatically generate test report templates that meet engineering acceptance specifications.

[0069] This invention's technical system organically combines precast pile lifecycle quality monitoring with intelligent assessment of existing piles, achieving full-process digital control of pile foundation engineering from construction and acceptance to later operation and maintenance. Engineering verification shows that the detection accuracy of newly constructed piles reaches ±2% of the load measurement value, and the error in assessing the bearing capacity of existing piles is controlled within ±5%, significantly better than the ±10% technical indicator of traditional detection methods. Particularly in the diagnosis of pile foundation quality under complex geological conditions, the system can identify local defects as small as 5mm, improving detection sensitivity by more than three times compared to conventional methods.

[0070] Optionally, the non-destructive testing system for the tensile strength of piles of the present invention further includes a self-powered unit, which integrates a photovoltaic thin film and a vibration power generation device.

[0071] It can be understood that, in order to facilitate the normal operation of the drive system, the system of the present invention also integrates a self-powered unit, which integrates a photovoltaic thin film and a vibration power generation device, and can support photovoltaic + vibration dual-mode power generation. The low power consumption design enables the battery to last for more than 3 years, meeting the long-term monitoring needs of harsh environments such as marine platforms and ensuring long-term stability.

[0072] It is understood that the embodiments of the system described above are merely illustrative, and the units described as separate components may or may not be physically separated; they may be located in one place or distributed across different network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any inventive effort.

[0073] Based on the same inventive concept, this invention also provides a non-destructive testing method for the tensile bearing capacity of piles according to the above embodiments. This method achieves non-destructive testing of the tensile bearing capacity of piles by applying the non-destructive testing system for tensile bearing capacity provided in the above embodiments. Therefore, the descriptions and definitions in the non-destructive testing system for tensile bearing capacity of piles in the above embodiments can be used to understand the relevant processing steps in this invention. For details, please refer to the above embodiments, which will not be repeated here.

[0074] According to an embodiment of the present invention, the processing flow of the non-destructive testing method for the bearing capacity of tension piles is as follows: Figure 6 The diagram shown is a flowchart illustrating the non-destructive testing method for the tensile bearing capacity of piles provided by this invention. This method can be implemented by applying the non-destructive testing system for the tensile bearing capacity of piles described in the above embodiments, and specifically includes the following processing steps: S601, using the strain sensing unit, detect the strain information of the target pull-out pile along the entire pile length.

[0075] S602, using the displacement monitoring unit, the absolute displacement and relative slippage of the target pull-out pile are monitored simultaneously.

[0076] S603, using the processor unit, based on the strain information, the absolute displacement and the relative slip, an FPGA-accelerated LSTM-GRU hybrid neural network is used to inversely calculate the bearing capacity of the target pull-out pile.

[0077] The non-destructive testing method for the bearing capacity of pull-out piles provided by this invention achieves accurate mapping from distributed optical fiber strain data to bearing capacity sub-indicators by establishing a differential coupling model of pile axial strain, pile-soil relative displacement and side friction, thereby achieving the goal of non-destructive testing of pull-out pile bearing capacity, effectively reducing testing costs and improving testing accuracy and reliability.

[0078] In one embodiment of the present invention, the implementation scheme for the reinforcement assessment technology of the target uplift pile being an existing pile is as follows: Figure 7 The diagram shown illustrates the process for evaluating the bored pile reinforcement of existing piles using the non-destructive testing method for tensile pile bearing capacity provided by this invention. The process includes: First, a minimally invasive sensor network implantation technique was used to assess the reinforcement of existing pile foundations in service.

[0079] Secondly, a vertical hole with a diameter of 30mm was drilled using a diamond drill bit, and the drilling depth needed to penetrate the pile bottom and enter the bearing layer at least 500mm. During the drilling process, three-dimensional laser positioning technology was used to avoid the main reinforcement positions, and the hole quality was monitored in real time using an endoscopic imaging system.

[0080] Next, after drilling is completed, a distributed sensor chain pre-encapsulated in a flexible sheath is slowly implanted into the borehole, followed by the injection of a cement-based grout with nano-silica modified properties. The elastic modulus of this grout is precisely adjusted to maintain a ±5% match with the pile concrete, ensuring coordinated deformation between the sensors and the pile structure.

[0081] Secondly, after grouting is completed, a quality verification test must be conducted, using core drilling to extract no fewer than three verification samples. The grout density is then tested using CT scanning, requiring the grout porosity to be less than 2% and the interfacial bonding strength to reach at least 85% of the original pile strength.

[0082] Then, after acceptance, the system enters a fully automatic long-term monitoring mode, which includes three operating mechanisms: the basic mode performs full parameter acquisition every 30 days and generates a pile foundation service status report; when the pile top displacement rate exceeds the warning value of 0.1 mm / day, it automatically switches to the high-frequency acquisition mode, and the sampling frequency is increased to 1 time / hour; in the event of special working conditions such as earthquakes and foundation pit excavation, the event trigger mode is activated to continuously track data.

[0083] Then, after the long-term monitoring data is processed by the cloud platform, the decay curve of the remaining bearing capacity of the pile foundation is calculated using the improved Burgers creep model.

[0084] Finally, the model input parameters include 12 indicators such as concrete carbonation depth, steel corrosion rate, and soil creep coefficient. The output results can predict the bearing capacity change trend over the next 5 years. When the system determines that the remaining safety factor is less than 1.3, it automatically pushes a graded early warning signal and generates reinforcement scheme suggestions, including specific repair parameters such as the grouting reinforcement range and the number of carbon fiber cloth wrapping layers.

[0085] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a USB flash drive, mobile hard drive, ROM, RAM, magnetic disk, or optical disk, etc., and includes several instructions to cause a computer device (such as a personal computer, server, or network device, etc.) to execute the methods described in the above method embodiments or some parts of the method embodiments.

[0086] Furthermore, those skilled in the art should understand that in the application documents of this invention, the terms "comprising," "including," or any other variations thereof are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0087] Numerous specific details are set forth in this specification. However, it should be understood that embodiments of the invention may be practiced without these specific details. In some instances, well-known methods, structures, and techniques have not been shown in detail so as not to obscure the understanding of this specification. Similarly, it should be understood that, in order to simplify the disclosure of this invention and aid in the understanding of one or more aspects of the invention, various features of the invention are sometimes grouped together in a single embodiment, figure, or description thereof in the above description of exemplary embodiments of the invention.

[0088] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A non-destructive testing system for the bearing capacity of pull-out piles, characterized in that, include: The system includes a strain sensing unit, a displacement monitoring unit, and a processor unit, wherein the processor unit is connected to the strain sensing unit and the displacement monitoring unit, respectively. The strain sensing unit includes a distributed ring fiber Bragg grating (FBG) array arranged on the target tensile pile according to a preset arrangement scheme, which is used to detect the strain information of the target tensile pile along the entire pile length. The displacement monitoring unit includes a MEMS inclinometer and a laser rangefinder arranged on the top of the target tensile pile, and an LVDT displacement meter pre-embedded at a preset density at the pile-soil interface, used to simultaneously monitor the absolute displacement and relative slippage of the target tensile pile. The processor unit is used to calculate the bearing capacity of the target pull-out pile by using an FPGA-accelerated LSTM-GRU hybrid neural network based on the strain information, the absolute displacement, and the relative slip.

2. The non-destructive testing system for the bearing capacity of pull-out piles according to claim 1, characterized in that, The processor unit is also used for: Based on the strain information and the absolute displacement, the variable cross-section elastic mechanical equation corresponding to the target tensile pile is established to characterize the strain-displacement conversion relationship. Based on the absolute displacement and the relative slippage, a pile-soil interface contact mechanics model is established to separate the absolute displacement of the target pull-out pile from the relative slippage between the pile and the soil. Based on the relationship between the relative slip and the layered skin friction of the target pull-out pile, an optimization model with L1-L2 mixed norm constraint is established by constructing an optimization objective function with L1-L2 mixed norm constraint and combining dynamic correction of soil parameters. The L1 norm penalty term represents the bearing layer, and the L2 norm smoothing term is used to suppress spurious fluctuations. Based on the variable cross-section elasticity equation, the pile-soil interface contact mechanics model, and the optimization model with L1-L2 mixed norm constraints, an LSTM-GRU hybrid neural network model is initialized. The strain gradient time series features of the finite element simulation data are extracted through the initialized LSTM-GRU hybrid neural network model. The model is pre-trained and then transferred to FPGA parallel computing using field measured data to construct the FPGA-accelerated LSTM-GRU hybrid neural network.

3. The non-destructive testing system for the bearing capacity of pull-out piles according to claim 1 or 2, characterized in that, When the processor unit is used for the inversion calculation of the bearing capacity of the target uplift pile, it is used for: The temporal features of the strain gradient corresponding to the strain information are extracted by an LSTM network, and based on the temporal features of the strain gradient, combined with a GRU network, the spatial distribution law of the displacement derivative is captured. Based on the spatial distribution pattern, the layered frictional resistance of the target pull-out pile is calculated using an optimization model with L1-L2 mixed norm constraints and an FPGA parallel computing architecture, and the bearing capacity is determined based on the layered frictional force.

4. The non-destructive testing system for the bearing capacity of pull-out piles according to claim 3, characterized in that, The processor unit is also used for: Based on the continuous strain information, the cumulative sum statistic (CUSUM) is calculated using the sliding window technique, and the abrupt change characteristics of the strain gradient are dynamically analyzed based on the CUSUM and a preset threshold. Based on the aforementioned mutation characteristics and preset judgment conditions, the location of potential soil layer interfaces is identified.

5. The non-destructive testing system for the bearing capacity of pull-out piles according to claim 1, 2, or 4, characterized in that, It also includes a temperature and humidity sensing unit; The temperature and humidity sensing unit includes multiple temperature sensors and humidity sensors arranged alternately at a preset density along the axis of the target tensile pile, used to collect ambient temperature and humidity information around the axis of the target tensile pile. The processor unit is further configured to decouple temperature noise based on the ambient temperature information and the ambient humidity information, using wavelet envelope analysis and an improved Tikhonov regularization algorithm, and to correct the bearing capacity based on the temperature noise.

6. The non-destructive testing system for the bearing capacity of pull-out piles according to claim 1, characterized in that, The distributed ring fiber Bragg grating (FBG) array includes multiple fiber Bragg grating rings distributed along the axial direction of the target uplift pile, and each fiber Bragg grating ring includes multiple fiber Bragg gratings symmetrically and uniformly distributed along the circumferential direction of the target uplift pile.

7. The non-destructive testing system for the bearing capacity of pull-out piles according to claim 6, characterized in that, The distribution density of the fiber Bragg grating ring along the axial direction of the target tensile pile increases at the soil layer interface, and / or, if the target tensile pile is a variable cross-section pile, the distribution density of the fiber Bragg grating in the fiber Bragg grating ring increases along the cross-section variation of the target tensile pile.

8. The non-destructive testing system for the bearing capacity of pull-out piles according to claim 1, characterized in that, The fiber Bragg gratings in the distributed ring fiber Bragg grating (FBG) array are protected by a double-layer protective sleeve, which is manufactured using a nano-injection molding process. The waterproof rating of the double-layer protective sleeve is not lower than IP68, and the tensile strength of the double-layer protective sleeve is ≥200MPa.

9. The non-destructive testing system for the bearing capacity of pull-out piles according to claim 1, characterized in that, It also includes a LoRa / 5G dual-mode communication channel, used to transmit feature data packets and / or full data streams based on the displacement state of the target pull-out pile.

10. The non-destructive testing system for the bearing capacity of pull-out piles according to claim 9, characterized in that, It also includes a visualization unit, which is used to generate one or more of the following based on the feature data package and / or the full data stream: strain cloud map, frictional resistance bar chart, bearing capacity-displacement curve and three-dimensional safety factor cloud map, and to perform three-dimensional visualization display.

11. The non-destructive testing system for the bearing capacity of pull-out piles according to claim 1, characterized in that, If the target pull-out pile is an existing pile, the distributed ring fiber Bragg grating (FBG) array is installed using a minimally invasive implantation process. The minimally invasive implantation process includes: implanting the fiber Bragg grating through a 30mm borehole and injecting a nano-modified cement-based grout with matching elastic modulus to fill and seal the borehole.

12. A method for non-destructive testing of the bearing capacity of an uplift pile based on the non-destructive testing system for uplift pile bearing capacity as described in any one of claims 1-11, characterized in that, include: The strain sensing unit is used to detect the strain information of the target pull-out pile along its entire length. The displacement monitoring unit is used to simultaneously monitor the absolute displacement and relative slippage of the target pull-out pile; Using the processor unit, based on the strain information, the absolute displacement, and the relative slip, the bearing capacity of the target tensile pile is calculated by inverting the LSTM-GRU hybrid neural network accelerated by FPGA.

Images