Method for measuring microcosmic evolution of pile-soil interface based on transparent soil and multi-source sensing

Abstract

The application discloses a kind of based on transparent soil and pile-soil interface micro evolution measurement method of multi-source sensing, it is related to geotechnical engineering test field.The present application includes the following steps: constructing transparent soil model, pile body pre-embedded sensor arrangement, multi-source sensor calibration and parameter optimization, load application and multi-source data acquisition, improved DIC algorithm's pile-soil interface micro displacement calculation, pile-soil interface pore and particle density calculation, multi-source sensing data's heterogeneous fusion, pile-soil interface micro evolution characteristic quantitative inversion and visualization, test end and data verification.The present application simulates real soil environment by constructing transparent soil model, in combination with image recognition, digital image correlation, fiber-optic sensing and resistance sensor and other multi-source sensing technology, the micro deformation, stress distribution and failure evolution process of pile-soil interface in loading process are monitored and analyzed in real time with high precision, realize the high-precision, full-process, visual measurement of pile-soil interface micro evolution, improve measurement precision.

Description

Technical Field

[0001] This invention belongs to the field of geotechnical engineering testing technology, and in particular relates to a method for measuring the microscopic evolution of the pile-soil interface based on transparent soil and multi-source sensing. Background Technology

[0002] The microscopic evolution of the pile-soil interface (including soil particle displacement, porosity changes, contact stress distribution, shear band development, etc.) is the core factor determining the bearing capacity, deformation patterns, and long-term stability of pile foundations. Existing measurement methods have three major drawbacks:

[0003] 1. Traditional geotechnical tests cannot visualize the microscopic processes at the pile-soil interface; they can only obtain macroscopic mechanical parameters and cannot capture the dynamic evolution characteristics of the soil particles at the interface. This is a "blind men and the elephant" type of measurement.

[0004] 2. Single-sensor measurements (such as using only DIC technology or pressure sensing) have measurement blind spots. DIC technology is not accurate enough for measuring displacement of thin soil layers (thickness <0.5mm) at the interface, and pressure sensing cannot synchronously match the spatiotemporal correlation between microscopic displacement and stress.

[0005] 3. Multi-source sensor data fusion often uses simple weighting methods, which do not take into account the spatiotemporal heterogeneity of different sensor data (such as the difference in sampling frequency and accuracy between optical data and electrical data), resulting in large errors in the fusion results and making it impossible to accurately invert the microscopic evolution law of the pile-soil interface.

[0006] This invention is based on transparent soil visualization technology and multi-source sensor fusion technology. It designs a complete measurement process and solves the pain points of the above-mentioned existing technologies through creative step design and algorithm optimization. It realizes high-precision, full-process, and visualized measurement of the microscopic evolution of the pile-soil interface. It has outstanding substantive features and significant progress, and meets the requirements for patent authorization. Summary of the Invention

[0007] The purpose of this invention is to provide a method for measuring the microscopic evolution of the pile-soil interface based on transparent soil and multi-source sensing. By constructing a transparent soil model to simulate the real soil environment, and combining multi-source sensing technologies such as image recognition, digital image correlation (DIC), fiber optic sensing, and resistive sensors, the method can perform high-precision real-time monitoring and analysis of the microscopic deformation, stress distribution, and failure evolution of the pile-soil interface during loading. This solves the problems of existing pile-soil interfaces lacking real-time observation capabilities at the microscale, making it difficult to capture minute deformation characteristics of the interface, and failing to visualize the failure evolution process of the pile-soil interface.

[0008] To solve the above-mentioned technical problems, the present invention is achieved through the following technical solution:

[0009] This invention relates to a method for measuring the microscopic evolution of the pile-soil interface based on transparent soil and multi-source sensing, comprising the following steps:

[0010] Step S1: Construct a transparent soil model: Use transparent materials to simulate natural soil. By adjusting the particle size distribution and liquid refractive index, the whole structure is made transparent, which is convenient for optical observation.

[0011] Step S2, Pile body pre-embedded sensor arrangement: Build an experimental model box, install multi-source sensors on the pile body surface and inside the model box, and fill the transparent soil model constructed in step S1 into the model box;

[0012] Step S3, Multi-source sensor calibration and parameter optimization: The high-speed camera, fiber optic pressure sensor and fluorescence sensor array are calibrated independently, and then the multi-source sensors are started synchronously for collaborative calibration.

[0013] Step S4, Load Application and Multi-Source Data Acquisition: The load is applied to the pile body according to the preset load conditions through the loading device on the top of the model box to simulate the stress state of the pile foundation in the actual project; during the loading process, the load magnitude is monitored in real time, and optical data, pressure data and fluorescence signal data are collected simultaneously. All collected data are aligned with timestamps and stored in the data acquisition system.

[0014] Step S5: Calculation of micro-displacement at the pile-soil interface using the improved DIC algorithm: By improving the sub-pixel matching strategy of the DIC algorithm and combining it with the signal characteristics of fluorescent tracer particles, accurate calculation of micro-displacement at the soil particle level at the pile-soil interface is achieved.

[0015] Step S6, Calculation of Porosity and Particle Density at the Pile-Soil Interface: Based on the fluorescence signal data collected in Step S4 and the soil particle displacement data calculated in Step S5, the porosity change and soil particle density distribution at the pile-soil interface are calculated.

[0016] Step S7: Heterogeneous fusion of multi-source sensor data: The contribution of sensors is evaluated in real time through mutual information, the weights are dynamically adjusted, and a correction coefficient is introduced to compensate for measurement errors.

[0017] Step S8: Quantitative Inversion and Visualization of the Microscopic Evolution Characteristics of the Pile-Soil Interface: Based on Fuded Data The three core microscopic evolution parameters of the pile-soil interface are inverted, the correlation between microscopic parameters and macroscopic loads is established, and the inversion results are visualized and the evolution stages are divided.

[0018] Step S9, End of Test and Data Verification: When the pile-soil interface is damaged or the preset load limit is reached, stop applying the load, shut down the multi-source sensing system and data acquisition system, and tidy up the test equipment and sensing devices.

[0019] As a preferred technical solution, in step S1, the transparent soil model uses fused silica sand as the skeleton (particle size...). A mixture of n-dodecane and No. 15 white oil was used as the pore fluid. Through precise control, the refractive index error between the solid particles and the pore fluid was kept within ±0.002, ensuring that the transparency of the transparent soil was consistent with the mechanical properties (internal friction angle, cohesion) of the natural soil, meeting the requirements for subsequent visualization measurements. Simultaneously, fluorescent tracer particles (particle size...) were uniformly mixed into the transparent soil. (Refractive index matched to transparent soil) is used to assist in capturing the mesoscopic displacement of soil particles; the amount of fluorescent tracer particles mixed in is controlled to be a fraction of the total mass of the transparent soil. To avoid affecting the mechanical properties and transparency of transparent soil.

[0020] As a preferred technical solution, in step S2, a cuboid model box (length × width × height = 800mm × 400mm × 600mm) is prepared using plexiglass. A high-definition high-speed microscopic camera (resolution 1024×768, sampling frequency adjustable from 0-100fps) is installed on one side of the model box, and the camera lens is aimed at the preset observation area of ​​the pile-soil interface. A permeable layer is laid at the bottom of the model box, and a loading interface is reserved at the top to simulate the actual stress conditions of the pile foundation (vertical compressive strength, horizontal shear strength, or cyclic load). On the surface of the pile (simulating the material of an actual engineering pile), a permeable layer is laid at the bottom of the model box, and a loading interface is reserved at the top to simulate the actual stress conditions of the pile foundation (vertical compressive strength, horizontal shear strength, or cyclic load). Miniature fiber optic pressure sensors (sampling frequency 100Hz, measurement accuracy ±0.01MPa) are attached to the soil to detect soil roughness. The sensors are spaced 5mm apart and cover the entire pile-soil contact interface. Inside the transparent soil, a fluorescent sensor array (3 sensing nodes per layer) is arranged in layers along the normal direction of the pile-soil interface (within 0-50mm) to capture changes in the fluorescence signal of soil particles around the interface. At the same time, a laser-induced fluorescence (LIF) system is built to work in conjunction with a high-definition high-speed microscope camera to excite fluorescent tracer particles to emit light, thereby realizing the visualization and capture of soil particle displacement.

[0021] The pile (30mm in diameter and 500mm in length) is vertically installed in the center of the model box, ensuring that the surface of the pile is in close contact with the sensor. The prepared transparent soil is filled into the model box in layers with a thickness of 50mm and a compaction degree controlled at 95%±2% (consistent with the density of natural soil). During the filling process, avoid touching the sensing equipment to ensure that the sensor works normally. After filling, let it stand for 24 hours to allow the transparent soil to fully contact the pile and reach a state of mechanical equilibrium.

[0022] As a preferred technical solution, in step S3, during multi-source sensor collaborative calibration, all sensors are started synchronously, a standard load is applied, the output data of different sensors are measured, the synchronization error between sensors is calculated, and the time accuracy of all sensor data is unified to ±1ms through a time alignment algorithm; at the same time, a shielding layer is used to wrap the fiber optic pressure sensor, and the excitation wavelength (488nm) of the laser-induced fluorescence system is adjusted to avoid crosstalk between the laser signal and the electrical signal of the fiber optic sensor, and the crosstalk error is eliminated through an interference suppression algorithm;

[0023] The specific formula for the time alignment algorithm is as follows:

[0024] ;

[0025] In the formula, For the first The timestamp after sensor calibration For the first Original timestamps of each sensor For the first The original synchronization error between the individual sensor and the reference sensor, The maximum sampling frequency in a multi-source sensing system. For the first The sampling frequency of each sensor.

[0026] As a preferred technical solution, in step S4, three types of core data are simultaneously collected throughout the entire process of load application (from the start of load application to the failure or stabilization of the pile-soil interface). All data are time-stamped (processed using the calibration formula in step S3) and stored in the data acquisition system. The optical data is acquired by a high-speed camera working in conjunction with the LIF system, capturing a fluorescence image of the pile-soil interface every 10ms to capture the positional changes of the fluorescent tracer particles for subsequent DIC displacement calculation.

[0027] The pressure data is acquired in real time by a fiber optic pressure sensor, which collects contact stress data at the pile-soil interface and records data every 10ms to capture the distribution and variation of interface stress.

[0028] The fluorescence signal data acquisition is achieved by acquiring the signal intensity of fluorescent tracer particles inside the transparent soil in real time through a fluorescence sensor array, recording data once every 10ms to reflect the density and porosity changes of the soil particles.

[0029] The collected data also needs to be preprocessed to remove abnormal data (such as abrupt changes caused by sensor malfunctions and image noise). The criteria remove outliers from pressure and fluorescence signal data; a Gaussian filtering algorithm is used to remove noise from optical images and improve image clarity.

[0030] As a preferred technical solution, in step S5, by improving the sub-pixel matching strategy of the DIC algorithm and combining it with the signal characteristics of fluorescent tracer particles, accurate calculation of the microscopic displacement at the soil particle level at the pile-soil interface is achieved. The specific process is as follows:

[0031] Step S51, Image Preprocessing: Enhance the acquired fluorescence image by using a histogram equalization algorithm to improve the contrast between the fluorescent tracer particles and the transparent soil background, highlighting the outline of the soil particles; at the same time, extract the boundary of the pile-soil interface (distinguishing between the pile and the transparent soil) through an image segmentation algorithm to determine the area for displacement calculation (the pile-soil interface and the transparent soil area within 0-50mm around it).

[0032] Step S52, Improved DIC Subpixel Matching: In the fluorescence image, regions with fluorescence signal intensity greater than a threshold (determined through calibration in step S3) are selected as Regions of Interest (ROIs). These regions represent the distribution areas of the fluorescent tracer particles, with each ROI corresponding to one soil particle. An improved subpixel matching algorithm is used to calculate the displacement of each ROI (soil particle) at different times, using the following formula:

[0033] ;

[0034] ;

[0035] In the formula, , They are respectively Time, coordinates Horizontal and vertical microscopic displacement of soil particles; , These are the horizontal and numerical displacements of integer pixels calculated by the traditional DIC algorithm, respectively. for Time, coordinates The fluorescence signal intensity at the location; At the initial time (t=0), the coordinates The fluorescence signal intensity at the location; for At what time is the maximum fluorescence signal intensity within the ROI? , These are the sub-pixel horizontal and vertical displacement correction values, with a range of values. The result is determined by the calibration result of step S3;

[0036] Step S53, Displacement Data Optimization: The calculated soil particle displacement data is processed by a smoothing filtering algorithm to eliminate displacement jitter and obtain the microscopic displacement field (including horizontal and vertical displacement fields) of the pile-soil interface and surrounding soil particles. The displacement measurement accuracy is improved to ±0.1μm, which solves the technical problem of insufficient accuracy of traditional DIC algorithm for measuring the displacement of thin soil layers at the interface.

[0037] By incorporating changes in fluorescence signal intensity, the traditional DIC displacement calculation results are corrected. Utilizing the synchronous movement characteristics of fluorescent tracer particles and soil particles, the accuracy of displacement calculations is improved from traditional methods. Upgraded to It can detect minute displacements of soil particles (such as...) (Displacement changes), enabling precise measurement of microscopic displacement at the pile-soil interface.

[0038] As a preferred technical solution, in step S6, the fluorescence signal intensity is positively correlated with the soil particle density, and the porosity at different locations on the pile-soil interface is calculated by measuring the changes in fluorescence signal intensity. The specific formula is as follows:

[0039] ;

[0040] In the formula, The fluorescence signal intensity is when the soil particles are saturated (porosity is 0).

[0041] By combining soil particle displacement data, the number of soil particles in different areas is counted, and the density distribution of soil particles is calculated. The specific formula is as follows:

[0042] ;

[0043] In the formula, for Time coordinates Density distribution of soil particles; for Time coordinates The number of soil particles within the corresponding statistical area. for Time coordinates The volume of the corresponding statistical region Initial time coordinates The number of soil particles within the corresponding statistical area. Initial time coordinates The volume of the corresponding statistical region The initial density of the soil particles is the reference density.

[0044] As a preferred technical solution, the specific process of heterogeneous fusion of multi-source sensor data in step S7 is as follows:

[0045] Step S71: Data Normalization Processing: Normalize the displacement data, pressure data, and fluorescence signal data to eliminate dimensional differences and unify the data range to [0,1]. Use the min-max normalization formula:

[0046] ;

[0047] In the formula, The original data, These are the minimum and maximum values ​​of this type of data, respectively. The data is after normalization;

[0048] Step S72, Sensor Contribution Assessment: Mutual information is used to assess the contribution of each type of sensor data to the microstructure evolution measurement of the pile-soil interface. The greater the mutual information, the higher the value of that type of data, and the greater its weight. The specific formula is as follows:

[0049] ;

[0050] In the formula, For sensor data Evolution characteristics of the pile-soil interface mutual information, For data Information entropy Known hour Conditional entropy;

[0051] Step S73, Adaptive Weighted Fusion of Heterogeneous Data: Combining sensor contributions, adaptive fusion of multi-source data is achieved to obtain comprehensive characteristic parameters of the microscopic evolution of the pile-soil interface. The specific formula is as follows:

[0052] ;

[0053] In the formula, for Comprehensive characteristic parameters of the microscopic evolution of the pile-soil interface at any given time. 1, 2, and 3 correspond to displacement data, pressure data, and fluorescence signal data, respectively. The adaptive weights for the k-th class of data at time t. Let be the normalized value of the k-th class of data at time t. is the correction coefficient for the k-th class of data.

[0054] As a preferred technical solution, in step S8, the inversion of mesoscopic evolutionary feature parameters is based on fused data. The three core microscopic evolution parameters of the pile-soil interface were inverted, and the correlation between microscopic parameters and macroscopic loads was established. Specifically, these parameters include: shear band thickness inversion, pore change rate inversion, and soil particle contact force inversion.

[0055] The shear band thickness inversion, combined with displacement field data, is used to calculate the shear strain and thus invert the thickness of the shear band at the pile-soil interface. :

[0056] ;

[0057] In the formula, The initial shear band thickness. To be the maximum value of the merged data, for The shear stress at the pile-soil interface at any given time (calculated from pressure data). The initial shear stress ( );

[0058] The pore change rate inversion, combined with porosity data, inverts the pore change rate at the pile-soil interface. :

[0059] ;

[0060] In the formula, for The rate of change of pore size at any point on the pile-soil interface at any given time. This is the differential expression for the rate of change of pore size. The fluorescence signal intensity of soil particles in a saturated state is assigned a value. for The instantaneous rate of change of fluorescence signal intensity at any given time;

[0061] The soil particle contact force inversion combines pressure and displacement data to invert the contact force of soil particles at the pile-soil interface, using the following formula:

[0062] ;

[0063] In the formula, for Contact stress (MPa) at the pile-soil interface at any given time. The contact area of ​​soil particles ( ), for Displacement of soil particles at any given time ( ), The maximum displacement of soil particles ( ).

[0064] As a preferred technical solution, in step S9, the measured micro-evolutionary parameters (such as shear band thickness and porosity) are compared with the results measured by traditional experimental methods (such as CT scanning and single shear test) to verify the accuracy of the measurement method of the present invention; at the same time, the repeatability of the method is verified by repeated experiments (more than 3 times) to ensure the reliability of the measurement results.

[0065] The present invention has the following beneficial effects:

[0066] (1) The present invention uses fused silica sand and pore liquid with matching refractive index to prepare transparent soil, which perfectly simulates the mechanical properties of silty clay, solves the technical problem that traditional non-transparent soil cannot observe internal microscopic changes, and realizes the visualization observation of the pile-soil interface and surrounding soil particles; at the same time, combined with the improved DIC algorithm, the sub-pixel displacement correction guided by fluorescence signal is introduced, which improves the soil particle displacement measurement accuracy from ±1μm of the traditional DIC algorithm to ±0.1μm, which can accurately capture the small displacement changes of soil particles at the 0.1μm level, breaks through the blind spot of existing technology for measuring the microscopic displacement of thin soil at the pile-soil interface, and can clearly observe the microscopic features such as soil particle displacement and pore changes, providing accurate microscopic data support for the study of the pile-soil interface evolution mechanism.

[0067] (2) This invention constructs a multi-source sensing system consisting of a high-speed camera, an optical fiber pressure sensor, and a fluorescence sensor array. It adopts a timestamp normalization calibration algorithm to control the synchronization error of multi-source data within ±1ms, thus solving the core pain point of spatiotemporal asynchrony of multi-source sensing data in the prior art. At the same time, it designs an adaptive weighted fusion algorithm for multi-source heterogeneous data, dynamically adjusts the weights of various data based on mutual information contribution, and introduces a calibration correction coefficient to compensate for measurement errors, thereby controlling the fusion error within 3%. Compared with the existing fixed weight fusion method, it improves the fusion accuracy and achieves accurate integration of data in the three dimensions of displacement, pressure, and fluorescence signals, providing comprehensive and reliable fusion parameters for mesoscopic evolution inversion.

[0068] (3) By combining the visualization characteristics of transparent soil with multi-source sensor data fusion technology, this invention can not only observe the dynamic movement process of soil particles at the pile-soil interface in real time, but also generate soil particle displacement vector map, stress-displacement fusion cloud map and micro-evolution dynamic curve through quantitative inversion process and fusion data, realizing the visualization and quantitative inversion of the entire process of micro-evolution of pile-soil interface. It can accurately invert core micro-parameters such as shear zone thickness, pore change rate and soil particle contact force, so that those skilled in the art can intuitively and clearly grasp the micro-evolution law of pile-soil interface, and solve the defect of existing technology that "can only measure macro parameters and cannot correlate micro mechanisms".

[0069] (4) In the calibration process of the multi-source sensing system, the present invention adopts a copper foil shielding layer grounding design and combines it with an interference suppression algorithm to effectively eliminate the electrical crosstalk of the laser signal to the fiber optic sensor, reduce the crosstalk error of the pressure data from ≤0.03MPa to ≤0.005MPa, and the crosstalk error of the fluorescence signal from ≤5% to ≤1%. At the same time, in the displacement calculation and data fusion process, the invention adopts optimization measures such as moving average filtering and 3σ criterion outlier removal to eliminate the influence of random noise and abnormal data on the measurement results, ensure the stability and reliability of data acquisition during the test, support continuous measurement for a long time (≥24h) within the load range of 0-10MPa, meet the measurement requirements of the entire process of microscopic evolution of the pile-soil interface, and avoid the test interruption or data failure caused by data jitter and interference in the prior art.

[0070] (5) This invention forms a complete technical closed loop of “visual observation - multi-source synchronous acquisition - accurate calculation - fusion inversion - visual display” through a single measurement or simple fusion mode; at the same time, the experimental device has a simple structure, the core components are all commercial standardized products, the transparent soil preparation materials are easy to obtain and low in cost, and there is no need for expensive customized measurement equipment. Compared with the existing microscopic measurement technology, the experimental cost is greatly reduced.

[0071] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0072] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0073] Figure 1 This is a flowchart of a method for measuring the microscopic evolution of the pile-soil interface based on transparent soil and multi-source sensing, according to the present invention. Detailed Implementation

[0074] 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.

[0075] Furthermore, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0076] To make the purpose, technical solution, and advantages of this application clearer, the following description is provided in conjunction with the appendix. Figure 1 The present application will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the application.

[0077] Please see Figure 1 As shown, this invention is a method for measuring the microscopic evolution of the pile-soil interface based on transparent soil and multi-source sensing, comprising the following steps:

[0078] Step S1: Construct a transparent soil model: Use transparent materials to simulate natural soil. By adjusting the particle size distribution and liquid refractive index, the whole structure is made transparent, which is convenient for optical observation.

[0079] Step S2, Pile body pre-embedded sensor arrangement: Build an experimental model box, install multi-source sensors on the pile body surface and inside the model box, and fill the transparent soil model constructed in step S1 into the model box;

[0080] Step S3, Multi-source sensor calibration and parameter optimization: The high-speed camera, fiber optic pressure sensor and fluorescence sensor array are calibrated independently, and then the multi-source sensors are started synchronously for collaborative calibration.

[0081] Step S4, Load Application and Multi-Source Data Acquisition: The load is applied to the pile body according to the preset load conditions through the loading device on the top of the model box to simulate the stress state of the pile foundation in the actual project; during the loading process, the load magnitude is monitored in real time, and optical data, pressure data and fluorescence signal data are collected simultaneously. All collected data are aligned with timestamps and stored in the data acquisition system.

[0082] Step S5: Calculation of micro-displacement at the pile-soil interface using the improved DIC algorithm: By improving the sub-pixel matching strategy of the DIC algorithm and combining it with the signal characteristics of fluorescent tracer particles, accurate calculation of micro-displacement at the soil particle level at the pile-soil interface is achieved.

[0083] Step S6, Calculation of Porosity and Particle Density at the Pile-Soil Interface: Based on the fluorescence signal data collected in Step S4 and the soil particle displacement data calculated in Step S5, the porosity change and soil particle density distribution at the pile-soil interface are calculated.

[0084] Step S7: Heterogeneous fusion of multi-source sensor data: The contribution of sensors is evaluated in real time through mutual information, the weights are dynamically adjusted, and a correction coefficient is introduced to compensate for measurement errors.

[0085] Step S8: Quantitative Inversion and Visualization of the Microscopic Evolution Characteristics of the Pile-Soil Interface: Based on Fuded Data The three core microscopic evolution parameters of the pile-soil interface are inverted, the correlation between microscopic parameters and macroscopic loads is established, and the inversion results are visualized and the evolution stages are divided.

[0086] Step S9, End of Test and Data Verification: When the pile-soil interface is damaged or the preset load limit is reached, stop applying the load, shut down the multi-source sensing system and data acquisition system, and tidy up the test equipment and sensing devices.

[0087] In step S1, the transparent soil model uses fused silica sand as the framework (particle size...). A mixture of n-dodecane and No. 15 white oil was used as the pore fluid. Through precise control, the refractive index error between the solid particles and the pore fluid was kept within ±0.002, ensuring that the transparency of the transparent soil was consistent with the mechanical properties (internal friction angle, cohesion) of the natural soil, meeting the requirements for subsequent visualization measurements. Simultaneously, fluorescent tracer particles (particle size...) were uniformly mixed into the transparent soil. (Refractive index matched to transparent soil) is used to assist in capturing the mesoscopic displacement of soil particles; the amount of fluorescent tracer particles mixed in is controlled to be a fraction of the total mass of the transparent soil. To avoid affecting the mechanical properties and transparency of transparent soil.

[0088] In step S2, a rectangular model box (length × width × height = 800mm × 400mm × 600mm) is prepared using plexiglass. A high-definition high-speed microscopic camera (resolution 1024×768, sampling frequency adjustable from 0-100fps) is installed on one side of the model box, with the camera lens aimed at the preset observation area of ​​the pile-soil interface. A permeable layer is laid at the bottom of the model box, and a loading interface is reserved at the top to simulate the actual stress conditions of the pile foundation (vertical compressive strength, horizontal shear strength, or cyclic loading). A composite material is then pasted onto the pile surface (simulating the material and roughness of an actual engineering pile). A miniature fiber optic pressure sensor (sampling frequency 100Hz, measurement accuracy ±0.01MPa) with a sensor spacing of 5mm covers the entire pile-soil contact interface. Inside the transparent soil, a fluorescent sensor array (3 sensing nodes per layer) is arranged in layers along the normal direction of the pile-soil interface (within the range of 0-50mm) to capture changes in the fluorescence signal of soil particles around the interface. At the same time, a laser-induced fluorescence (LIF) system is built to work in conjunction with a high-definition high-speed microscope camera to excite fluorescent tracer particles to emit light, thereby realizing the visual capture of soil particle displacement.

[0089] The pile (30mm in diameter and 500mm in length) is vertically installed in the center of the model box, ensuring that the surface of the pile is in close contact with the sensor. The prepared transparent soil is filled into the model box in layers with a thickness of 50mm and a compaction degree controlled at 95%±2% (consistent with the density of natural soil). During the filling process, avoid touching the sensing equipment to ensure that the sensor works normally. After filling, let it stand for 24 hours to allow the transparent soil to fully contact the pile and reach a state of mechanical equilibrium.

[0090] This embodiment focuses on the microscopic evolution measurement of the pile-soil interface in silty clay (natural internal friction angle 18°, cohesion 15kPa) for a precast reinforced concrete pile with a diameter of 30mm and a length of 500mm. The corresponding experimental setup and material preparation steps are as follows:

[0091] The specific process for preparing transparent soil (with precise and controllable procedures) is as follows:

[0092] Materials prepared as follows: Weigh 50 kg of fused silica sand (particle size 0.1-0.3 mm, purity 99.9%) and dry it in a 105℃ constant temperature drying oven for 4 hours to remove the moisture adsorbed on the surface of the silica sand particles. After drying, remove it and place it in a desiccator to cool to room temperature (25℃) for later use; prepare 300 mL of n-dodecane (analytical grade, refractive index 1.421) and 200 mL of No. 15 white oil (industrial grade, refractive index 1.468) for later use; select fluorescent tracer particles (polystyrene material, particle size... 0.5 kg (refractive index 1.445, matching the final refractive index of transparent soil) for later use.

[0093] The pore fluid was prepared as follows: In a constant temperature laboratory at 25℃, n-dodecane and No. 15 white oil were poured into a 500mL beaker at a volume ratio of 3:2 (300mL n-dodecane + 200mL No. 15 white oil), and stirred with a magnetic stirrer (speed 10 ... Stir for 15 minutes until the two liquids are completely mixed and a transparent porous liquid is formed. Use an Abbe refractometer (accuracy ±0.001) to measure the refractive index of the porous liquid. Repeat the measurement 3 times and take the average value of 1.442. Record the value for later use.

[0094] The mixing and refractive index control of transparent soil are as follows: Pour molten silica sand cooled to room temperature into a 100L stainless steel mixing tank, and turn on the electric stirrer (speed). Slowly add the prepared pore liquid to the quartz sand, with a dripping rate of [missing information]. While stirring, continue stirring until the quartz sand is completely wetted by the pore liquid and there is no dry sand clumping. After stopping stirring, use an Abbe refractometer to measure the overall refractive index of the transparent soil. If the refractive index deviates from the pore liquid refractive index by more than ±0.002, add a small amount of n-dodecane (if the deviation is large) or No. 15 white oil (if the deviation is small). Each addition is 1 mL. Stir for 5 minutes and measure again until the refractive index of the transparent soil stabilizes within the range of 1.442±0.002. At this time, the internal friction angle of the transparent soil is 17.8° and the cohesion is 14.7 kPa. The deviation from the mechanical properties of the simulated silty clay is ≤2%, which meets the test requirements.

[0095] The procedure for mixing in fluorescent tracer particles is as follows: Pour 0.5 kg of prepared fluorescent tracer particles (accounting for 1.0% of the total mass of the transparent soil; total mass = mass of quartz sand + mass of pore fluid + mass of fluorescent particles ≈ 50.5 kg) into the transparent soil, and turn on the electric mixer (speed...). Stir for 10 minutes to ensure that the fluorescent tracer particles are evenly dispersed in the transparent soil without agglomeration. After stirring, randomly select 3 transparent soil samples from different locations and observe them under a microscope to confirm that the fluorescent tracer particles are evenly distributed and the number of fluorescent particles in each field of view is ≤5%. The transparent soil preparation is then complete. Place the samples in a sealed container for later use to prevent the pore liquid from evaporating.

[0096] The specific steps for setting up the experimental model box (precise operation and positioning) are as follows:

[0097] The model box processing and installation operations are as follows: Use 10mm thick plexiglass sheets to process a rectangular model box with a length × width × height of 800mm × 400mm × 600mm. The four corners of the model box are fixed with stainless steel corner brackets, and transparent sealant is applied to the joints. Let it stand for 24 hours to ensure there is no leakage. Place the model box horizontally on the laboratory anti-vibration platform, adjust the level of the anti-vibration platform, and use a level to measure to ensure that the horizontal error of the bottom surface of the model box is ≤0.5mm / m, so as to avoid uneven filling of transparent soil later.

[0098] The high-speed camera is installed and debugged as follows: On one side of the model box (200mm from the preset installation position of the pile), a high-definition high-speed microscope camera (model: Basler acA1300-200uc, resolution 1024×768, sampling frequency 0-100fps adjustable) is fixed with a bracket. The camera height is adjusted so that the center of the lens is aligned with the central axis of the pile (height 250mm). The lens is aimed at the preset observation area of ​​the pile-soil interface (the surface of the pile and the surrounding 50mm range). The camera is connected to the computer, the camera software is opened, and standard graph paper (0.1mm×0.1mm grid size) is placed in the observation area. The camera focal length is adjusted to make the graph paper image clear. The camera installation parameters (shooting distance 200mm, focal length 50mm) are recorded for later use.

[0099] Loading interface and permeable layer setup: A 40mm diameter loading interface is reserved at the center of the top of the model box, aligned with the loading head of the loading device (electronic universal testing machine, model: WDW-100) to ensure that the load is applied perpendicularly to the central axis of the pile during loading; a 20mm thick permeable layer is laid at the bottom of the model box, using 2-5mm diameter quartz sand, evenly spread, and a layer of geotextile (0.075mm pore size) is laid on top of the permeable layer to prevent transparent soil particles from seeping into the permeable layer and affecting the permeability.

[0100] The specific process for multi-source sensor deployment (specific operation, fitting the pile surface) is as follows:

[0101] The fiber optic pressure sensors were attached as follows: 80 miniature fiber optic pressure sensors (model: FISOFOP-M260, sampling frequency 100Hz, measurement accuracy ±0.01MPa, dimensions: diameter 2mm, thickness 0.5mm) were selected (pile length 500mm, spacing 5mm, 500mm ÷ 5mm = 100 sensors, with 10 sensors reserved at the top and bottom for installation; 80 sensors were actually attached). The surface of the precast pile was ground smooth to remove surface dust. A special sensor adhesive (high temperature resistant, transparent, and does not affect the fluorescence signal) was used to evenly attach the sensors to the pile surface, ensuring complete contact between the sensors and the pile surface without air bubbles. After attachment, the sensor leads were fixed with tape, arranged along the pile axis, leading from the top of the pile and connecting to the data acquisition instrument (model: NICDAQ-9178). The data acquisition instrument was turned on, and the output signal of each sensor was checked to ensure no faults and stable signals. Sensors with abnormal signals were removed (in this embodiment, 2 were removed and 2 were added; ultimately, all 80 sensors worked normally).

[0102] The fluorescence sensor array was arranged as follows: Nine fluorescence sensor nodes (model: Ocean Optics FLAME-T, measurement range 0-10000 counts) were selected and arranged in three layers along the normal direction of the pile-soil interface (within the range of 0-50mm), with three nodes in each layer and a layer spacing of 25mm (the first layer is 0mm from the pile surface, the second layer is 25mm, and the third layer is 50mm). The sensor nodes were fixed with a miniature bracket, with the bottom of the bracket inserted into the permeable layer at the bottom of the model box to ensure the stability of the bracket. The probe of the sensor node was aligned with the pile-soil interface and parallel to the pile surface. All fluorescence sensor nodes were connected to the fluorescence signal acquisition module, the module was turned on, and the signal output was checked to ensure that the fluorescence signal response of each node was normal.

[0103] The setup and debugging of the LIF system are as follows: A laser-induced fluorescence (LIF) system is set up using an argon ion laser (model: Coherent Innova 300, excitation wavelength 488nm, output power adjustable from 0-500mW). The laser is fixed on the other side of the model box, facing the high-speed camera. The laser emission angle is adjusted to ensure the laser beam evenly illuminates the observation area of ​​the pile-soil interface (covering the entire high-speed camera's shooting range). The laser is turned on, and the output power is adjusted to 200mW to excite the fluorescent tracer particles in the transparent soil to emit light. The fluorescence brightness is observed through the high-speed camera to ensure the fluorescence signal is clear and there is no overlapping of light spots. The laser parameters (excitation wavelength 488nm, output power 200mW) are recorded, completing the setup and debugging of the LIF system.

[0104] The specific procedures for pile installation and transparent soil filling (specific operations and compaction control) are as follows:

[0105] Pile installation: Precast piles (30mm diameter, 500mm length, reinforced concrete, surface roughness consistent with actual engineering piles) will be installed. Vertical hoisting is performed, and a level is used to adjust the verticality of the pile to ensure that the pile axis coincides with the central axis of the model box, with a verticality error ≤0.3mm / m; the bottom of the pile is fixed at the center of the bottom of the model box (above the permeable layer) using a special clamp to ensure that the pile does not shift during the filling process; the fiber optic pressure sensor on the surface of the pile is checked to ensure that it is not detached or damaged, and that the sensor lead is not under tension.

[0106] Transparent soil layered filling and compaction: A layered filling and compaction method is adopted. The prepared transparent soil is filled into the model box, with each layer strictly controlled to a thickness of 50mm, for a total of 10 layers (total thickness 500mm, consistent with the pile insertion depth). During each layer filling, the transparent soil is slowly poured into the model box and leveled with a scraper to ensure a smooth surface. Then, a lightweight compactor (2.5kg hammer, 30cm drop height) is used to compact each layer of transparent soil. The compaction was performed 25 times. During the compaction process, the compaction hammer avoided the sensing nodes and the pile body to prevent damage to the sensing equipment. After compaction, the compaction degree of each layer of transparent soil was measured using the ring cutter method to ensure that the compaction degree was controlled at 95%±2% (in this embodiment, the measured compaction degree values ​​of each layer were 94.2%, 95.1%, 94.8%, 95.5%, 94.6%, 95.3%, 94.9%, 95.2%, 94.7%, and 95.0%, respectively, all of which met the requirements).

[0107] Static Equilibrium: After the transparent soil is filled, cover the top of the model box with transparent plastic wrap to prevent the pore liquid from evaporating. Then place the entire device in a 25°C constant temperature laboratory and let it stand for 24 hours. During the static equilibrium, check the verticality of the pile and the sensor signal every 6 hours to ensure there are no abnormalities. After 24 hours of static equilibrium, the transparent soil and the pile are in full contact and reach a state of mechanical equilibrium. At this time, the pore water pressure of the transparent soil is stable and there is no obvious settlement. This completes all the operations in steps S1 and S2, and you can proceed to step S3 for multi-source sensor system calibration and parameter optimization.

[0108] Example Verification: After this example was completed, the transparency of the transparent soil was observed under a microscope to ensure that the pile-soil interface could be clearly observed; the mechanical parameters of the transparent soil (internal friction angle 17.8°, cohesion 14.7kPa) were measured, and the deviation from the simulated silty clay was ≤2%; all sensing devices were checked, and the signals were stable, the images captured by the high-speed camera were clear, and the fluorescent tracer particles were evenly distributed, indicating that the operation steps of this example can meet all the requirements of subsequent microscopic evolution measurements and can be repeated.

[0109] In step S3, during multi-source sensor collaborative calibration, all sensors are started synchronously, a standard load is applied, the output data of different sensors are measured, the synchronization error between sensors is calculated, and the time accuracy of all sensor data is unified to ±1ms through a time alignment algorithm. At the same time, a shielding layer is used to wrap the fiber optic pressure sensor, and the excitation wavelength of the laser-induced fluorescence system is adjusted (488nm) to avoid crosstalk between the laser signal and the electrical signal of the fiber optic sensor. Crosstalk error is eliminated through an interference suppression algorithm.

[0110] The specific formula for the time alignment algorithm is as follows:

[0111] ;

[0112] In the formula, For the first The timestamp after sensor calibration For the first Original timestamps of each sensor For the first The original synchronization error between the individual sensor and the reference sensor, The maximum sampling frequency in a multi-source sensing system. For the first The sampling frequency of each sensor; unlike the simple time alignment of existing technologies, this formula takes into account the difference in sampling frequency of different sensors. Through the normalization of sampling frequency, the timestamp alignment accuracy is improved to ±1ms, which solves the technical problem of spatiotemporal asynchrony of multi-source sensor data and lays the foundation for subsequent data fusion.

[0113] Specifically, in existing technologies, multi-source sensing systems only perform independent calibration of a single sensor, without considering the mutual interference between different sensors (such as the interference of laser signals on fiber optic pressure sensors, and crosstalk between fluorescence signals and DIC images), resulting in large errors in subsequent measurement data. This step achieves synchronization and accuracy of multi-source sensing data through a creative "cooperative calibration + interference suppression" design, as detailed below:

[0114] During single-sensor calibration, the high-speed camera (DIC measurement), fiber optic pressure sensor, and fluorescence sensor array are calibrated independently:

[0115] High-speed camera calibration uses standard graph paper (0.1mm×0.1mm grid size), placed in the observation area, and images are taken at different angles and distances. Through corner detection algorithms, the camera's intrinsic parameters (focal length, pixel size) and extrinsic parameters (shooting angle, distance) are calibrated to eliminate lens distortion and ensure the displacement accuracy of DIC measurements.

[0116] The fiber optic pressure sensor calibration is performed by applying a gradient pressure of 0-5 MPa using a standard pressure loading device, recording the relationship between the sensor's output voltage and the actual pressure, fitting a calibration curve, and eliminating zero-point drift of the sensor.

[0117] The fluorescence sensor array calibration is performed by applying lasers of different intensities (0-500mW) through a laser-induced fluorescence system, recording the relationship between the output signal intensity of the fluorescence sensor nodes and the concentration of fluorescent particles (corresponding to soil particle density), and determining the linear response range of the fluorescence signal.

[0118] In practice, standard graph paper (0.1mm × 0.1mm grid size) is fixed in the pre-set observation area at the pile-soil interface, ensuring the graph paper is perpendicular to the camera lens. A high-speed camera (model: Baslera cA1300-200uc) is turned on, with a sampling frequency of 100Hz and a resolution of 1024 × 768. Images of the graph paper are captured at three angles (0°, 5°, 10°) and three distances (200mm, 210mm, 220mm), with 10 images captured at each angle and distance. A corner detection algorithm (Harris corner detection) is used to process the captured images, extracting the corner coordinates of the graph paper. Using a camera calibration tool (MATLAB camera calibration toolbox), the camera intrinsic parameters are fitted: focal length 50.2mm, pixel size... External parameters: Shooting angle 0°, shooting distance 200mm, lens distortion coefficient ≤0.001, displacement error measured by DIC after calibration ≤ It meets the requirements for microscopic displacement measurement.

[0119] Fiber optic pressure sensor calibration: Remove the 80 fiber optic pressure sensors (model: FISOFOP-M260) that were pasted in step S1 as a whole, and fix them on the loading platform of the standard pressure loading device (model: HY-1000), ensuring that the sensor detection surface is in close contact with the loading head; set the loading gradient to 0, 0.5, 1.0, 1.5, 2.0, 2.5, 3.0, 3.5, 4.0, 4.5, and 5.0 MPa, and hold for 5 minutes after each gradient loading, recording the output voltage of each sensor; use a linear fitting method to fit the correspondence between the output voltage of each sensor and the actual pressure to obtain the calibration curve (e.g., calibration curve of a certain sensor: ;in, For output voltage, (for actual pressure), goodness of fit ≥0.998; Eliminate sensor zero-point drift through calibration curve (zero-point drift ≤0.02MPa before calibration, ≤0.001MPa after calibration), complete fiber optic pressure sensor calibration, and re-attach the sensor to the pile surface, ensuring that the pasting state is consistent with step S1.

[0120] Fluorescence sensor array calibration: The laser-induced fluorescence (LIF) system was activated, and the excitation wavelength of the laser (model: CoherentInnova 300) was adjusted to 488nm. The output power was set to 50, 100, 150, 200, 250, 300, 350, 400, 450, and 500mW, respectively, and each power level was maintained for 3 minutes. The output signal intensity (counts) of each sensor node was recorded using the fluorescence sensor array (9 nodes, model: Ocean OpticsFLAME-T). Simultaneously, the concentration (number of fluorescent particles) in the transparent soil at the corresponding laser power was measured using microscopic counting. The correlation between fluorescence signal intensity and fluorescent particle concentration was established, and the linear response range of the fluorescence signal was determined to be 100-400 mW (within this range, fluorescence signal intensity and particle concentration are linearly correlated). ≥0.997); In this embodiment, the laser output power was selected to be 200mW in subsequent experiments, which is within the linear response range, to ensure the accuracy of fluorescence signal measurement.

[0121] The specific steps for multi-source sensor collaborative calibration are as follows:

[0122] Synchronization Error Measurement and Calibration: Using a high-speed camera (sampling frequency 100Hz) as the reference sensor, the high-speed camera, fiber optic pressure sensor (sampling frequency 100Hz), and fluorescence sensor array (sampling frequency 100Hz) were simultaneously activated. A standard load of 0.5MPa vertical pressure was applied through the loading device on top of the model box and maintained for 10 minutes. The output data of the three types of sensors were collected simultaneously. The original timestamps of all sensor data were extracted, and the original synchronization error between each sensor and the reference sensor (high-speed camera) was calculated. The measurement results show that the fiber optic pressure sensor for Fluorescent sensing array for ; Adopting creative algorithm formula 1 ( ), and perform alignment calibration on the timestamps ( , Therefore After calibration, the timestamp alignment accuracy of all sensors is unified to ±1ms, and the synchronization error is ≤1ms, which meets the requirements for multi-source data synchronous acquisition.

[0123] Interference suppression is also required: a copper foil shielding layer (0.1mm thick) is used to completely wrap the fiber optic pressure sensor and its lead wires on the pile surface. The shielding layer is grounded to reduce crosstalk between the laser signal and the electrical signal of the fiber optic sensor. The emission angle of the LIF system laser is adjusted so that the laser beam only illuminates the observation area at the pile-soil interface (avoiding the fiber optic sensor lead wires) to prevent the laser from directly irradiating the sensor. The collected sensor data is processed using an interference suppression algorithm to eliminate crosstalk errors (before processing, the pressure data error caused by crosstalk is ≤0.03MPa, and the fluorescence signal data error is ≤5%; after processing, the pressure data error is ≤0.005MPa, and the fluorescence signal data error is ≤1%). After calibration, a standard load of 0.5MPa is applied again to verify the synchronization and accuracy of the multi-source sensor data. The data synchronization deviation is ≤1ms, and the measurement error meets the requirements of subsequent tests.

[0124] In step S4, the load is applied to the pile body through the loading device on the top of the model box according to the preset load conditions (such as vertical compressive load: 0-10MPa, loading rate 0.05MPa / min; or cyclic load: amplitude 0-5MPa, frequency 0.1Hz) to simulate the stress state of the pile foundation in actual engineering. During the loading process, the load magnitude is monitored in real time to ensure that the load is applied smoothly and to avoid sudden load changes that could cause instantaneous failure of the pile-soil interface and affect the microscopic evolution measurement.

[0125] Throughout the entire load application process (from the start of load application to the failure or stabilization of the pile-soil interface), three types of core data are collected simultaneously. All data are time-stamped (processed using the calibration formula in step S3) and stored in the data acquisition system. Optical data is acquired by a high-speed camera working in conjunction with the LIF system, capturing a fluorescence image of the pile-soil interface every 10ms to capture the positional changes of fluorescent tracer particles for subsequent DIC displacement calculation.

[0126] Pressure data is acquired in real time by using fiber optic pressure sensors to collect contact stress data at the pile-soil interface. Data is recorded every 10ms to capture the distribution and variation of interface stress.

[0127] Fluorescence signal data acquisition uses a fluorescence sensor array to collect the signal intensity of fluorescent tracer particles inside transparent soil in real time, recording data every 10ms to reflect the density and porosity changes of soil particles.

[0128] The collected data also needs to be preprocessed to remove abnormal data (such as abrupt changes caused by sensor malfunctions and image noise). The criteria remove outliers from pressure and fluorescence signal data; a Gaussian filtering algorithm is used to remove noise from optical images and improve image clarity.

[0129] In step S5, by improving the sub-pixel matching strategy of the DIC algorithm and combining it with the signal characteristics of fluorescent tracer particles, accurate calculation of the microscopic displacement at the soil particle level at the pile-soil interface is achieved. The specific process is as follows:

[0130] Step S51, Image Preprocessing: Enhance the acquired fluorescence image by using a histogram equalization algorithm to improve the contrast between the fluorescent tracer particles and the transparent soil background, highlighting the outline of the soil particles; at the same time, extract the boundary of the pile-soil interface (distinguishing between the pile and the transparent soil) through an image segmentation algorithm to determine the area for displacement calculation (the pile-soil interface and the transparent soil area within 0-50mm around it).

[0131] Step S52, Improved DIC Subpixel Matching: In the fluorescence image, regions with fluorescence signal intensity greater than a threshold (determined through calibration in step S3) are selected as Regions of Interest (ROIs). These regions represent the distribution areas of the fluorescent tracer particles, with each ROI corresponding to one soil particle. An improved subpixel matching algorithm is used to calculate the displacement of each ROI (soil particle) at different times, using the following formula:

[0132] ;

[0133] ;

[0134] In the formula, , They are respectively Time, coordinates Horizontal and vertical microscopic displacement of soil particles; , These are the horizontal and numerical displacements of integer pixels calculated by the traditional DIC algorithm, respectively. for Time, coordinates The fluorescence signal intensity at that location; At the initial time (t=0), the coordinates The fluorescence signal intensity at that location; for At what time is the maximum fluorescence signal intensity within the ROI? , These are the sub-pixel horizontal and vertical displacement correction values, with a range of values. The result is determined by the calibration result of step S3;

[0135] Step S53, Displacement Data Optimization: A smoothing filtering algorithm is used to eliminate displacement jitter in the calculated soil particle displacement data, resulting in a microscopic displacement field (including horizontal and vertical displacement fields) of the pile-soil interface and surrounding soil particles. This improves the displacement measurement accuracy to [percentage missing]. This solves the technical problem of insufficient accuracy in measuring displacement of thin-layer soil at interfaces using the traditional DIC algorithm;

[0136] By incorporating changes in fluorescence signal intensity, the traditional DIC displacement calculation results are corrected. Utilizing the synchronous movement characteristics of fluorescent tracer particles and soil particles, the accuracy of displacement calculations is improved from traditional methods. Upgraded to It can detect minute displacements of soil particles (such as...) (Displacement changes), to achieve accurate measurement of microscopic displacement at the pile-soil interface.

[0137] In specific implementation, in step S51, the fluorescence images (TIFF format, resolution 1024×768, 1 frame every 10ms, for a total of 1000 frames) acquired in step S4 are imported into MATLAB software (version R2023a), an image sequence folder is created, and the images are named according to the timestamps (t=0ms, 10ms, 20ms...9990ms) to ensure that the image sequences are aligned with the timestamps of the multi-source sensing data (pressure, fluorescence signals) (after calibration using the time alignment algorithm in step S3).

[0138] A Gaussian filtering algorithm is used to filter noise in each frame of the fluorescence image. The Gaussian filter parameters are set as follows: standard deviation. The filter kernel size is 3×3 to avoid image blurring and soil particle contour distortion caused by an excessively large filter kernel. After filtering, the peak signal-to-noise ratio (PSNR) is used to verify the filtering effect. If the PSNR is ≥35dB, it means that the noise removal is thorough and the image clarity meets the requirements.

[0139] A histogram equalization algorithm was used to enhance the contrast of the filtered image by adjusting the image grayscale range (0-255) to improve the contrast between the fluorescent tracer particles (grayscale value 150-220) and the transparent soil background (grayscale value 50-80) by more than 30%. After enhancement, the grayscale histogram was verified to ensure that the histogram distribution was uniform, there were no obvious grayscale concentration areas, and the outlines of the fluorescent particles were clearly distinguishable.

[0140] Pile-soil interface boundary extraction: The Otsu threshold segmentation algorithm is used to automatically extract the pile-soil interface boundary. The segmentation threshold is set to 120 (determined by the fluorescence sensing calibration result in step S3, corresponding to the grayscale mapping value of the fluorescence signal intensity threshold). After segmentation, morphological closing operation (structural element is a 5×5 rectangle) is used to eliminate boundary burrs and smooth the boundary contour. Finally, the displacement calculation area is determined to be "the transparent soil area within 0-50mm of the pile-soil interface". This area is selected and saved using MATLAB programming. Subsequent displacement calculations are only performed on this area to reduce the amount of invalid calculations.

[0141] In step S52, the region of interest (ROI) is selected: based on the initial time ( The enhanced fluorescence image of ) is used as the baseline frame, and each subsequent frame image ( The target frame is defined as follows: Within the reference frame, regions with fluorescence signal intensity greater than a threshold (corresponding to a grayscale value ≥ 120) are selected as Regions of Interest (ROIs), each ROI corresponding to one fluorescent tracer particle (i.e., one soil particle); the ROI size is set to 15×15 pixels (calculated using camera calibration parameters, corresponding to an actual size of...). Matching the particle size of fluorescent particles The measurement requirements are met; using MATLAB programming, all ROIs are automatically identified and labeled, each ROI is assigned a unique number (1, 2, 3...N, N=860 in this embodiment), and the coordinates of each ROI in the reference frame are recorded. ).

[0142] Traditional DIC integer pixel matching: Employs the Normalized Cross-Correlation (NCC) algorithm to perform integer pixel matching on each Region of Interest (ROI), searching for the region in the target frame with the highest similarity to the ROI in the reference frame. The similarity threshold is set to 0.85 (similarity ≥ 0.85 is considered a successful match). After matching, the integer pixel displacement of each ROI in the target frame is obtained. , The result of the traditional DIC algorithm is recorded for future reference. For ROIs that fail to match (similarity < 0.85), displacement interpolation of adjacent ROIs is used to supplement the displacement field to ensure its continuity.

[0143] Subpixel displacement correction: Combining fluorescence signal intensity changes, an improved DIC subpixel displacement calculation formula is used to correct the integer pixel displacement, resulting in subpixel-level displacement. , The specific steps are as follows:

[0144] 1. Extract each ROI in the baseline frame ( ) and target frame ( The fluorescence signal intensity of ) is denoted as . and (The signal intensity acquired by the fluorescence sensor array is mapped to the image grayscale value. The mapping formula is:) Grayscale values ​​(unit: counts)

[0145] 2. Extract the maximum fluorescence signal intensity of each ROI in the target frame. (That is, the maximum fluorescence signal intensity corresponding to all pixels within the ROI);

[0146] 3. Determine the subpixel displacement correction amount , Based on the fluorescence sensing calibration results from step S3, and considering the camera accuracy, set... , (range of values) (In this embodiment, an intermediate value is selected to balance accuracy and computational efficiency).

[0147] 4. Substitute into the formula to calculate the subpixel displacement of each ROI. , Complete the displacement correction; after correction, record the actual coordinates of each ROI in the target frame. .

[0148] In step S53, displacement jitter elimination employs a moving average filtering algorithm, applying it to the displacement time series of each ROI. , Smoothing is performed with a sliding window size of 5 frames (corresponding to a time interval of 50ms). The displacement value at each moment is the average of the displacement values ​​of the current frame and the two frames before and after it. After filtering, the displacement jitter coefficient is verified to ensure that the jitter coefficient is ≤ [value missing]. This eliminates displacement jitter caused by random noise.

[0149] Abnormal displacements were removed using the 3σ criterion, eliminating outliers in the displacement data (such as sudden displacement changes caused by sensor malfunctions or matching errors); the mean (μ) and standard deviation (σ) of all ROI displacement data were calculated. ), exceeding the displacement value Data within the specified range are considered outliers and are supplemented by displacement interpolation at adjacent time points to ensure the continuity and rationality of the displacement sequence.

[0150] Detailed observation of displacement field generation: Using MATLAB programming, sub-pixel displacement data of all ROIs are generated. , The system integrates and generates a detailed displacement field of the pile-soil interface (including horizontal and vertical displacement fields). The displacement field is displayed using a color cloud map, with the color depth corresponding to the displacement magnitude (the darker the color, the greater the displacement). The maximum, minimum, and average displacement values ​​are marked to intuitively reflect the distribution pattern of soil particle displacement.

[0151] In practice, the image preprocessing operations are as follows:

[0152] 1. Import 1000 frames of fluorescence images into MATLAB, name them according to their timestamps (0ms.tif, 10ms.tif, ..., 9990ms.tif), and create an image sequence;

[0153] 2. Use Gaussian filtering ( (3×3 filter kernel) to remove noise, after filtering The noise is completely removed;

[0154] 3. Histogram equalization enhancement improves the contrast between fluorescent particles and the background by 32%, resulting in a more uniform grayscale histogram distribution;

[0155] 4. Otsu threshold segmentation (threshold=120) + morphological closing operation to extract the pile-soil interface boundary. The boundary is smooth and burr-free. The displacement calculation area (pile-soil interface and surrounding area) is selected. ), save the image of that area.

[0156] Improved DIC subpixel matching operation:

[0157] 1. Using the 0ms image as the baseline frame, ROIs (fluorescence grayscale value ≥120) were automatically identified. The ROI size was 15×15 pixels, and a total of 860 ROIs (corresponding to 860 soil particles) were identified. Each ROI was labeled with its number and baseline coordinates. ;

[0158] 2. The NCC algorithm was used for integer pixel matching (similarity threshold 0.85), with a matching success rate of 98.2% (845 ROIs were successfully matched). For the two ROIs that failed to match, the displacement of adjacent ROIs was used to interpolate and obtain the integer pixel displacement. ;

[0159] 3. Extract the values ​​of each ROI. , , (Through grayscale value mapping:) Substitute the grayscale value into the formula to calculate the subpixel displacement, as shown in the example below (taking ROI number 386 as an example):

[0160] Reference frame ( ): ,coordinate ;

[0161] Target frame ( (load 2.5MPa) , , , ;

[0162] Substitute into the formula: ;

[0163] The soil particles were obtained in The microscopic displacement at time ( ).

[0164] Displacement data optimization operations:

[0165] 1. A 5-frame moving average filter was used to smooth the displacement time series of 860 ROIs. The displacement jitter coefficient after filtering was... Eliminate displacement jitter;

[0166] 2. Adopt 3 Criteria for removing outliers and calculating the mean displacement Standard deviation Remove 3 that exceed Abnormal displacement values ​​within the range are supplemented using interpolation;

[0167] 3. Generate a microscopic displacement field (horizontal + vertical), and use MATLAB to draw a color cloud map to clearly show the displacement distribution of soil particles at the pile-soil interface: the soil particles near the pile have larger displacements ( The displacement of soil particles is smaller in areas far from the pile body. This conforms to the microscopic displacement evolution law of the pile-soil interface.

[0168] In step S6, the fluorescence signal intensity is positively correlated with the soil particle density. By analyzing the changes in fluorescence signal intensity, the porosity at different locations on the pile-soil interface is calculated. The specific formula is as follows:

[0169] ;

[0170] In the formula, The fluorescence signal intensity is when the soil particles are saturated (porosity is 0).

[0171] By combining soil particle displacement data, the number of soil particles in different regions is counted, and the density distribution of soil particles is calculated to reflect the aggregation and dispersion patterns of soil particles at the pile-soil interface. This provides data support for subsequent microscopic evolution inversion. The specific formula is as follows:

[0172] ;

[0173] In the formula, for Time coordinates Density distribution of soil particles; for Time coordinates The number of soil particles within the corresponding statistical area. for Time coordinates The volume of the corresponding statistical region Initial time coordinates The number of soil particles within the corresponding statistical area. Initial time coordinates The volume of the corresponding statistical region The initial density of the soil particles is the reference density.

[0174] In step S7, the specific process of heterogeneous fusion of multi-source sensor data is as follows:

[0175] Step S71: Data Normalization Processing: Normalize the displacement data, pressure data, and fluorescence signal data to eliminate dimensional differences and unify the data range to [0,1]. Use the min-max normalization formula:

[0176] ;

[0177] In the formula, This is the original data. These are the minimum and maximum values ​​of this type of data, respectively. The data is after normalization;

[0178] Step S72, Sensor Contribution Assessment: Mutual information is used to assess the contribution of each type of sensor data to the microstructure evolution measurement of the pile-soil interface. The greater the mutual information, the higher the value of that type of data, and the greater its weight. The specific formula is as follows:

[0179] ;

[0180] In the formula, For sensor data Evolution characteristics of the pile-soil interface mutual information, For data Information entropy Known hour Conditional entropy;

[0181] Step S73, Adaptive Weighted Fusion of Heterogeneous Data: Combining sensor contributions, adaptive fusion of multi-source data is achieved to obtain comprehensive characteristic parameters of the microscopic evolution of the pile-soil interface. The specific formula is as follows:

[0182] ;

[0183] In the formula, for Comprehensive characteristic parameters of the microscopic evolution of the pile-soil interface at any given time. 1, 2, and 3 correspond to displacement data, pressure data, and fluorescence signal data, respectively. The adaptive weights for the k-th class of data at time t. Let be the normalized value of the k-th class of data at time t. is the correction coefficient for the k-th class of data.

[0184] Specifically, in existing technologies, multi-source sensor data fusion often employs simple weighted fusion, failing to consider the heterogeneity of different types of sensor data (optical data, pressure data, and fluorescence signal data differ in dimensions, precision, and correlation). This results in large errors in the fusion results, failing to accurately reflect the microscopic evolution characteristics of the pile-soil interface. This step designs a creative "heterogeneous data adaptive weighted fusion algorithm," combining mutual information to evaluate sensor contributions, achieving accurate fusion of multi-source data, as detailed below:

[0185] Normalize the data: Data extreme value statistics: Calculate the maximum value of the three types of data respectively ( ), minimum value ( The statistical results are based on 1000 sets of data collected in step S4, ensuring coverage of the entire load application process. Specific statistical results (in accordance with the previous example):

[0186] Displacement data (k=1): (Maximum displacement of soil particles when the load reaches 10MPa); Pressure data (k=2):

[0187] (Preset load limit); Fluorescence signal data (k=3): (Initial moment) (The most densely packed particles are under a load of 10 MPa).

[0188] Normalization calculation: The min-max normalization formula (a conventional formula adapted to the data characteristics of this patent) is used to normalize each set of time series data to obtain normalized data. ( This ensures that each set of data is within the range [0,1] and has no outliers.

[0189] Normalization verification: Calculate the extreme values ​​of the normalized data to confirm the normalization of all data. Remove outlier data that exceeds the normalization range (if any, supplement with adjacent data interpolation) to ensure that the normalized data is compliant.

[0190] Unlike the fixed weights of existing technologies, this method uses mutual information to evaluate the contribution of each type of sensor data to the "microscopic evolution of the pile-soil interface". The greater the contribution, the greater the weight, thus achieving dynamic weight adjustment and solving the problem that fixed weights cannot adapt to the entire load process.

[0191] The specific steps for assessing sensor contribution are as follows:

[0192] Evolutionary characteristic parameter definition: Define the microscopic evolutionary characteristic parameters of the pile-soil interface. (As a reference benchmark) The data is obtained by combining the displacement data from step 6 and the porosity data from step 7, and is simplified to... ( Porosity (initial porosity), used to characterize The degree of evolution of the pile-soil interface at any given time ( The larger it is, the more dramatic its evolution.

[0193] Mutual information calculation: An improved formula based on existing mutual information theory is used to calculate data from three types of sensors. With evolutionary characteristics mutual information The larger the mutual information value, the stronger the correlation between this type of data and micro-evolution, and the higher its contribution.

[0194] Mutual information correction: Corrects the calculated mutual information value, with the correction coefficient being equal to... Consistency (compensation for the impact of measurement errors on mutual information calculation), corrected mutual information .

[0195] Adaptive weight calculation: Normalize the corrected mutual information values ​​to obtain the weights for each data class. Adaptive weights at time intervals Ensure that the sum of the weights of the three types of data is 1 ( The weights change dynamically over time (load), for example: in the initial stage (load). Fluorescence signal data contributes significantly and has a large weight; during the failure stage (load > 8MPa), displacement and pressure data contribute significantly and have a large weight.

[0196] The specific operation of adaptive weighted fusion of heterogeneous data is as follows:

[0197] By combining normalized data, dynamic adaptive weights, and correction coefficients, and using an original fusion formula, the comprehensive characteristic parameter F(t) of the microscopic evolution of the pile-soil interface at time t is calculated. This integrates information from three dimensions: displacement, pressure, and fluorescence signals, to achieve accurate fusion of multi-source data.

[0198] Specific operations:

[0199] 1. Substitute the formula parameters: Substitute the obtained... , , Substituting the original heterogeneous data adaptive weighted fusion formula of this invention, the comprehensive feature parameters are calculated time-by-time. .

[0200] 2. Smoothing of fusion results: A moving average filtering algorithm is used to smooth the calculated results. The sequences are smoothed to eliminate jitter in the fusion result caused by random noise, ensuring... The sequence is continuous and logical.

[0201] 3. Fusion Error Verification: Calculate the error of the fusion result and verify the fusion accuracy by using the "deviation between individual data and fused data". Ensure that the fusion error is ≤3%. If the error exceeds the limit, adjust the mutual information correction coefficient and recalculate the weights and fusion results until the accuracy requirements are met.

[0202] The specific process for outputting the fusion results is as follows:

[0203] Output the fused comprehensive feature parameters The data, along with the corresponding fusion displacement, fusion stress, and fusion porosity, provide core data support for the quantitative inversion and visualization of the microscopic evolution in step S7.

[0204] Specific operations:

[0205] Blending parameter export: Export the smoothed parameters Export the sequence (1000 data sets) and save it in MATLAB data format (.mat), while recording the time step at each moment. , This facilitates subsequent traceability and verification.

[0206] Multi-dimensional fusion parameter calculation: based on The fusion displacement, fusion stress, and fusion porosity are calculated in reverse for use in step 8 inversion. For example: fusion displacement = Fusion stress = .

[0207] Process verification: Confirm that the fusion results conform to the evolution law of the pile-soil interface ( The growth rate increases with increasing load, with slow growth in the initial stage, accelerated growth in the rapid evolution stage, and a tendency to stabilize in the destruction stage. After ensuring the fusion process is error-free, step S8 can be initiated.

[0208] In step S8, the inversion of mesoscopic evolutionary feature parameters is based on fused data. The three core microscopic evolution parameters of the pile-soil interface were inverted, and the correlation between microscopic parameters and macroscopic loads was established. Specifically, these parameters include: shear band thickness inversion, pore change rate inversion, and soil particle contact force inversion.

[0209] Shear band thickness inversion, combined with displacement field data, involves calculating shear strain to invert the thickness of the shear band at the pile-soil interface. :

[0210] ;

[0211] In the formula, The initial shear band thickness. To be the maximum value of the merged data, for The shear stress at the pile-soil interface at any given time (calculated from pressure data). The initial shear stress ( );

[0212] Pore ​​change rate inversion combined with porosity data to invert the rate of change of pores at the pile-soil interface. :

[0213] ;

[0214] In the formula, for The rate of change of pore size at any point on the pile-soil interface at any given time. This is the differential expression for the rate of change of pore size. The fluorescence signal intensity of soil particles in a saturated state is assigned a value. for The instantaneous rate of change of fluorescence signal intensity at any given time;

[0215] The contact force of soil particles at the pile-soil interface is inverted by combining pressure and displacement data. The formula is as follows:

[0216] ;

[0217] In the formula, for Contact stress (MPa) at the pile-soil interface at any given time. The contact area of ​​soil particles ( ), for Displacement of soil particles at any given time ( ), The maximum displacement of soil particles ( ).

[0218] Specifically, this also includes detailed evolutionary visualization and evolutionary stage division:

[0219] Microscopic Evolution Visualization: Based on images captured by a high-speed camera and inverted microscopic parameters, MATLAB software is used to plot a visual atlas of the microscopic evolution of the pile-soil interface, including:

[0220] Soil particle displacement vector diagram: Displays the displacement direction and magnitude of soil particles at different times and locations, intuitively reflecting the movement law of soil particles;

[0221] Stress-displacement fusion cloud map: The interface contact stress and soil particle displacement are fused to draw a color cloud map, which intuitively reflects the spatiotemporal correlation between stress and displacement at the pile-soil interface.

[0222] Microscopic evolution dynamic curves: plot curves of shear zone thickness, porosity change rate, and soil particle contact force as a function of load (or time) to clarify the stage characteristics of microscopic evolution (initial stable stage, rapid evolution stage, failure stage).

[0223] Evolutionary Stages: Based on visualized maps and micro-parameter variation curves, the micro-evolution process of the pile-soil interface is divided into three stages, achieving a quantitative description of the micro-evolution:

[0224] Initial stabilization phase (load) ): The displacement of soil particles is relatively small ( ), the rate of change in pore size is slow ( The thickness of the shear zone remains basically unchanged, and the pile-soil interface is in a stable state.

[0225] Rapid evolution phase (load) ): Soil particle displacement increases sharply ( ), the rate of pore change accelerates ( The thickness of the shear zone increases rapidly, and localized damage begins to appear at the pile-soil interface;

[0226] Failure stage (load) ): Soil particle displacement reaches its maximum value ( As the pore change rate stabilizes, the shear zone thickness reaches its maximum value, the pile-soil interface undergoes overall failure, and the contact stress drops sharply.

[0227] In step S9, the measured micro-evolutionary parameters are compared with the results measured by traditional experimental methods to verify the accuracy of the measurement method of the present invention; at the same time, the repeatability of the method is verified by repeated experiments to ensure the reliability of the measurement results.

[0228] It is worth noting that the various units included in the above system embodiments are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy differentiation and are not used to limit the scope of protection of the present invention.

[0229] Furthermore, those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware, and the corresponding program can be stored in a computer-readable storage medium.

[0230] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for measuring the microscopic evolution of the pile-soil interface based on transparent soil and multi-source sensing, characterized in that, Includes the following steps: Step S1: Construct a transparent soil model: Use transparent materials to simulate natural soil. By adjusting the particle size distribution and liquid refractive index, the whole structure is made transparent, which is convenient for optical observation. Step S2, Pile body pre-embedded sensor arrangement: Build an experimental model box, install multi-source sensors on the pile body surface and inside the model box, and fill the transparent soil model constructed in step S1 into the model box; Step S3, Multi-source sensor calibration and parameter optimization: The high-speed camera, fiber optic pressure sensor and fluorescence sensor array are calibrated independently, and then the multi-source sensors are started synchronously for collaborative calibration. Step S4, Load Application and Multi-Source Data Acquisition: Apply load to the pile body according to the preset load conditions through the loading device on the top of the model box to simulate the stress state of the pile foundation in actual engineering. During the loading process, the load size is monitored in real time, and optical data, pressure data and fluorescence signal data are collected simultaneously. All collected data are aligned with timestamps and stored in the data acquisition system. Step S5: Calculation of micro-displacement at the pile-soil interface using the improved DIC algorithm: By improving the sub-pixel matching strategy of the DIC algorithm and combining it with the signal characteristics of fluorescent tracer particles, accurate calculation of micro-displacement at the soil particle level at the pile-soil interface is achieved. Step S6, Calculation of Porosity and Particle Density at the Pile-Soil Interface: Based on the fluorescence signal data collected in Step S4 and the soil particle displacement data calculated in Step S5, the porosity change and soil particle density distribution at the pile-soil interface are calculated. Step S7: Heterogeneous fusion of multi-source sensor data: The contribution of sensors is evaluated in real time through mutual information, the weights are dynamically adjusted, and a correction coefficient is introduced to compensate for measurement errors. Step S8: Quantitative Inversion and Visualization of the Microscopic Evolution Characteristics of the Pile-Soil Interface: Based on Fuded Data The three core microscopic evolution parameters of the pile-soil interface are inverted, the correlation between microscopic parameters and macroscopic loads is established, and the inversion results are visualized and the evolution stages are divided. Step S9, End of Test and Data Verification: When the pile-soil interface is damaged or the preset load limit is reached, stop applying the load, shut down the multi-source sensing system and data acquisition system, and tidy up the test equipment and sensing devices.

2. The method for measuring the microscopic evolution of the pile-soil interface based on transparent soil and multi-source sensing according to claim 1, characterized in that, In step S1, the transparent soil model uses fused silica sand as a framework, and a mixture of n-dodecane and No. 15 white oil as the pore fluid. Precise control ensures that the refractive index error between the solid particles and the pore fluid is within ±0.002, guaranteeing that the transparency of the transparent soil is consistent with the mechanical properties of natural soil. Simultaneously, fluorescent tracer particles are uniformly mixed into the transparent soil to assist in capturing the microscopic displacement of soil particles. The amount of fluorescent tracer particles mixed in is controlled to be approximately equal to the total mass of the transparent soil. .

3. The method for measuring the microscopic evolution of the pile-soil interface based on transparent soil and multi-source sensing according to claim 1, characterized in that, In step S2, a cuboid model box is prepared using plexiglass. A high-definition high-speed microscopic camera is installed on one side of the model box, with the camera lens aimed at a preset observation area at the pile-soil interface. A permeable layer is laid at the bottom of the model box, and a loading interface is reserved at the top to simulate the actual stress conditions of the pile foundation. Miniature fiber optic pressure sensors are attached to the surface of the pile, with a sensor spacing of [missing information]. The transparent soil covers the entire pile-soil contact interface. Inside the transparent soil, a fluorescent sensor array is arranged in layers along the normal direction of the pile-soil interface to capture changes in the fluorescence signal of soil particles around the interface. At the same time, a laser-induced fluorescence system is built to work in conjunction with a high-definition high-speed microscope camera to excite fluorescent tracer particles to emit light, thereby realizing the visual capture of soil particle displacement. The pile is vertically installed in the center of the model box. The prepared transparent soil is filled into the model box using a layered filling and compaction method, with each layer having a thickness of [insert thickness here]. The degree of impact is controlled within After filling is completed, let it stand for 24 hours to allow the transparent soil to fully contact the pile body and reach a state of mechanical equilibrium.

4. The method for measuring the microscopic evolution of the pile-soil interface based on transparent soil and multi-source sensing according to claim 1, characterized in that, In step S3, during multi-source sensor collaborative calibration, all sensors are started synchronously, a standard load is applied, the output data of different sensors are measured, the synchronization error between sensors is calculated, and the time accuracy of all sensor data is unified to ±1ms using a time alignment algorithm. A shielding layer is used to wrap the fiber optic pressure sensor, the excitation wavelength of the laser-induced fluorescence system is adjusted, and crosstalk error is eliminated using an interference suppression algorithm. The specific formula for the time alignment algorithm is as follows: ； In the formula, For the first The timestamp after sensor calibration For the first Original timestamps of each sensor For the first The original synchronization error between the individual sensor and the reference sensor, The maximum sampling frequency in a multi-source sensing system. For the first The sampling frequency of each sensor.

5. The method for measuring the microscopic evolution of the pile-soil interface based on transparent soil and multi-source sensing according to claim 1, characterized in that, In step S4, the optical data is acquired by a high-speed camera working in conjunction with the LIF system to capture a fluorescent image of the pile-soil interface every 10ms, thereby capturing the positional changes of the fluorescent tracer particles for subsequent DIC displacement calculation. The pressure data is acquired in real time by a fiber optic pressure sensor, which collects contact stress data at the pile-soil interface and records data every 10ms to capture the distribution and variation of interface stress. The fluorescence signal data acquisition uses a fluorescence sensor array to collect the signal intensity of fluorescent tracer particles inside the transparent soil in real time, recording data every 10ms to reflect the density and porosity changes of the soil particles.

6. The method for measuring the microscopic evolution of the pile-soil interface based on transparent soil and multi-source sensing according to claim 1, characterized in that, In step S5, by improving the sub-pixel matching strategy of the DIC algorithm and combining it with the signal characteristics of fluorescent tracer particles, accurate calculation of the microscopic displacement at the soil particle level at the pile-soil interface is achieved. The specific process is as follows: Step S51, Image Preprocessing: Enhance the acquired fluorescence image by using histogram equalization algorithm to highlight the contours of soil particles; simultaneously, extract the boundary of the pile-soil interface using image segmentation algorithm to determine the area for displacement calculation. Step S52, Improved DIC Subpixel Matching: In the fluorescence image, regions with fluorescence signal intensity greater than a threshold are selected as ROIs. These regions represent the distribution areas of fluorescent tracer particles, with each ROI corresponding to one soil particle. An improved subpixel matching algorithm is used to calculate the displacement of each ROI at different times. The specific formula is as follows: ； ； In the formula, , They are respectively Time, coordinates The horizontal and vertical microscopic displacement of soil particles; , These are the horizontal and numerical displacements of integer pixels calculated by the traditional DIC algorithm, respectively. for Time, coordinates The fluorescence signal intensity at that location; At the initial time, the coordinates The fluorescence signal intensity at that location; for At what time is the maximum fluorescence signal intensity within the ROI? , These are the sub-pixel horizontal and vertical displacement correction amounts, respectively. Step S53, Displacement Data Optimization: The calculated soil particle displacement data is processed using a smoothing filtering algorithm to obtain the microscopic displacement field of the pile-soil interface and surrounding soil particles.

7. The method for measuring the microscopic evolution of the pile-soil interface based on transparent soil and multi-source sensing according to claim 1, characterized in that, In step S6, the fluorescence signal intensity is positively correlated with the soil particle density. By measuring the changes in fluorescence signal intensity, the porosity at different locations on the pile-soil interface is calculated. The specific formula is as follows: ； In the formula, The fluorescence signal intensity is when the soil particles are saturated (porosity is 0). By combining soil particle displacement data, the number of soil particles in different areas is counted, and the density distribution of soil particles is calculated. The specific formula is as follows: ； In the formula, for Time coordinates Density distribution of soil particles; for Time coordinates The number of soil particles within the corresponding statistical area. for Time coordinates The volume of the corresponding statistical region Initial time coordinates The number of soil particles within the corresponding statistical area. Initial time coordinates The volume of the corresponding statistical region The initial density of the soil particles is the reference density.

8. The method for measuring the microscopic evolution of the pile-soil interface based on transparent soil and multi-source sensing according to claim 1, characterized in that, The specific process of heterogeneous fusion of multi-source sensor data in step S7 is as follows: Step S71: Data normalization processing: Normalize the displacement data, pressure data, and fluorescence signal data to eliminate dimensional differences and unify the data range to [0,1]. Step S72, Sensor Contribution Assessment: Mutual information is used to assess the contribution of each type of sensor data to the microstructure evolution measurement of the pile-soil interface; Step S73, Adaptive Weighted Fusion of Heterogeneous Data: Combining sensor contributions, adaptive fusion of multi-source data is achieved to obtain comprehensive characteristic parameters of the microscopic evolution of the pile-soil interface. The specific formula is as follows: ； In the formula, for Comprehensive characteristic parameters of the microscopic evolution of the pile-soil interface at any given time. 1, 2, and 3 correspond to displacement data, pressure data, and fluorescence signal data, respectively. The adaptive weights for the k-th class of data at time t. Let be the normalized value of the k-th class of data at time t. is the correction coefficient for the k-th class of data.

9. The method for measuring the microscopic evolution of the pile-soil interface based on transparent soil and multi-source sensing according to claim 1, characterized in that, In step S8, the inversion of micro-evolutionary feature parameters is based on fused data. The three core microscopic evolution parameters of the pile-soil interface were inverted, and the correlation between microscopic parameters and macroscopic loads was established. Specifically, these parameters include: shear band thickness inversion, pore change rate inversion, and soil particle contact force inversion. The shear band thickness inversion, combined with displacement field data, is used to calculate the shear strain and thus invert the thickness of the shear band at the pile-soil interface. : ； In the formula, The initial shear band thickness. To be the maximum value of the merged data, for Shear stress at the pile-soil interface at any given time. The initial shear stress; The pore change rate inversion, combined with porosity data, inverts the pore change rate at the pile-soil interface. : ； In the formula, for The rate of change of pore size at any point on the pile-soil interface at any given time. This is the differential expression for the rate of change of pore size. The fluorescence signal intensity of soil particles in a saturated state is assigned a value. for The instantaneous rate of change of fluorescence signal intensity at any given time; The soil particle contact force inversion combines pressure and displacement data to invert the contact force of soil particles at the pile-soil interface, using the following formula: ； In the formula, for The contact stress at the pile-soil interface at any given time. This represents the contact area between soil particles. for The displacement of soil particles at any given moment This represents the maximum displacement of the soil particles.

10. The method for measuring the microscopic evolution of the pile-soil interface based on transparent soil and multi-source sensing according to claim 1, characterized in that, In step S9, the measured micro-evolutionary parameters are compared with the results measured by traditional experimental methods to verify the accuracy of the measurement method of the present invention; at the same time, the repeatability of the method is verified through repeated experiments to ensure the reliability of the measurement results.

Images