Fabricated pavement panel bottom void identification method

Abstract

The invention relates to an assembly type pavement slab bottom void identification method comprising the following steps: obtaining vibration signals of each measuring point of a reference slab and a to-be-identified pavement slab, and respectively calculating the cross-power spectral density of the whole slab; comparing the distribution of the whole-board cross-power spectral densities of the to-be-identified paving board and the reference board in the frequency domain, and judging whether the to-be-identified paving board is void or not; when the judgment result is that void exists, the to-be-identified pavement plate and the reference plate are partitioned, one region corresponds to one sensing unit, and the corresponding relation between each measuring point and each region in the plate is established; respectively calculating regional cross-power spectral densities of different regions in the to-be-identified paving plate and the reference plate; and calculating the difference between the regional cross-power spectral densities of the to-be-identified paving plate and the corresponding region of the reference plate, carrying out integration to obtain a void region identification index, and judging the void position of the to-be-identified paving plate according to the void region identification index. Compared with the prior art, the method has the advantages of high recognition precision, low cost and the like.

Description

Technical Field

[0001] The present invention relates to the field of intelligent monitoring of pavements, and particularly to a method for identifying the debonding of the bottom of prefabricated pavement slabs. Background Art

[0002] Cement concrete pavements have become the main pavement structure form of airports due to their long service life, strong bearing capacity and other advantages. However, with the increase of service life, the joint fillers of pavement slabs fail or are missing, and the base materials are washed away under the action of surface water scouring and pumping suction, and the problem of debonding often appears at the pavement joints. Debonding at the bottom of the slab is a common damage form of concrete pavements, which has a significant impact on the structural performance of concrete pavements. In practical engineering applications, it is necessary to accurately identify the debonding at the bottom of the slab, control the structural damage of the pavement, and ensure the service performance and service life of the pavement.

[0003] In the field of pavement debonding monitoring and identification, the most commonly used is the debonding determination method based on deflection, which is mainly divided into two types: the falling weight deflectometer (HWD / FWD) test and the Benkelman beam test. The falling weight deflectometer (HWD / FWD) is a widely used debonding determination method based on deflection at present. Its detection principle is that under the action of the same impact load, the deflection of the debonding area will increase significantly. This method has the problems of poor timeliness, difficulty in determining the debonding range, degree and spatial distribution, low measurement efficiency, and only being able to measure the debonding information of a single point each time.

[0004] In view of the problems existing in the deflection detection method, in recent years, many new debonding monitoring technologies have emerged. Common technologies include ground penetrating radar (GPR), ultrasonic array, infrared imaging, etc. Such technologies can measure the debonding range and degree more precisely, but most of them are point and line measurements. Each measurement requires the installation of a certain number of sensors, the array networking is complex, and high-performance detection equipment needs to be equipped, resulting in low engineering economy. For example, CN110487910A relates to a method for detecting and positioning the debonding of the face slab of a concrete face rockfill dam based on vibration sensing technology. First, it is judged whether there is debonding under the face slab; then, the debonding position under the face slab is determined; when there is a debonding part on the face slab, by comparing the differences in the vibration signals of the measuring points in each detection area, the debonding position is judged, and the final detection result is visually displayed in an image. In this regard, the existing technologies have the following defects:

[0005] 1. Limitation of measurement method: The existing technologies rely on sensors. In actual use, a certain number of sensors need to be installed, and the installation cost is high, which is difficult to meet the detection requirements of long-distance and large-scale cement concrete pavements.

[0006] 2. Limitation of measurement accuracy: The existing technologies mainly judge the debonding position by the amplitude or acceleration change amount in the time domain, considering a single dimension, resulting in insufficient final identification accuracy.

[0007] 3. Poor adaptability: The existing zoning methods are relatively fixed and difficult to adapt to panels of different sizes and shapes. Summary of the Invention

[0008] The purpose of the present invention is to provide a method for identifying the bottom void of prefabricated floor panels, realizing economic and accurate identification of the void of prefabricated floor panels.

[0009] The purpose of the present invention can be achieved by the following technical solutions:

[0010] A method for identifying the bottom void of prefabricated floor panels includes the following steps:

[0011] S1. Obtain the vibration signals of each measurement point of the reference panel and the panel to be identified, and calculate the cross-power spectral density of the whole panel respectively;

[0012] S2. Compare the distribution of the cross-power spectral density of the whole panel of the panel to be identified and the reference panel in the frequency domain to judge whether the panel to be identified is void;

[0013] S3. When the judgment result is that there is a void, partition the panel to be identified and the reference panel, with one area corresponding to one sensing unit, and establish the corresponding relationship between each measurement point and each area in the panel;

[0014] S4. Calculate the cross-power spectral density of different areas within the panel of the panel to be identified and the reference panel respectively;

[0015] S5. Calculate the difference between the cross-power spectral density of the corresponding areas of the panel to be identified and the reference panel, and perform integration to obtain the void area identification index, and judge the void position of the panel to be identified according to the void area identification index.

[0016] The vibration signal is collected by forming a sensing unit by winding an optical fiber, and multiple measurement points are distributed in each sensing unit.

[0017] The calculation method of the cross-power spectral density of the whole panel is as follows:

[0018] Obtain the vibration signal x i (t), where i represents the measurement point number, calculate the cross-power spectrum based on the vibration signals of all measurement points within the same floor panel, and store the data in the form of (measurement point ID, panel ID). The calculation method of the cross-power spectrum is as follows:

[0019]

[0020] In the formula: P xy(f) represents the cross-correlation function of signals x(n) and y(n) in the frequency domain, which is used to analyze the resonance characteristics of signals at different frequencies. x(n) and y(n) represent the vibration signals of two different measurement points, N is the number of samples, y*(n + f) represents the complex conjugate of signal y(n + f), f is the frequency variable, representing different frequency points in the frequency domain, and the normalization factor is used to ensure P xy represents the average power or energy; the calculated cross-power spectrum P xy (f) is a complex value, and its real and imaginary parts respectively reflect the amplitude and phase information of the signal at different frequencies;

[0021] Take the magnitude of the cross-power spectrum P xy (f) to obtain the amplitude spectrum of the cross-power spectrum |P xy (f)|:

[0022]

[0023] Where: and are respectively the real and imaginary parts of the complex number P xy (f);

[0024] Normalize the amplitude spectrum of the cross-power spectrum to obtain the cross-power spectral density:

[0025]

[0026] In the formula, ∑ f |P xy (f)| is the sum of the magnitudes at all frequency points;

[0027] Calculate the mean value of the cross-power spectral density of different measurement points to obtain the cross-power spectral density P of the entire plate f :

[0028]

[0029] In the formula, M represents the number of measurement points inside the plate, is the cross-power spectral density of the m-th measurement point at frequency f.

[0030] The specific method for establishing the correspondence between each measurement point and each area inside the plate is as follows:

[0031] Compare the vibration responses between each measurement point, that is, the time-domain curves of the vibration signals, calculate the similarity of the vibration responses of each measurement point using the dynamic time warping algorithm, and estimate the mapping relationship between the actual measurement points and the sensing units using the Monte Carlo simulation method.

[0032] The dynamic time warping algorithm calculates the similarity between vibration signals by optimizing the warping path. The similarity is characterized by the DTW distance. The smaller the DTW distance, the higher the signal similarity. Among them, the DTW distance of the mapping combination X is calculated by the following formula:

[0033]

[0034] In the formula, N0 is the number of measurement points of the sensing unit, L0 is the length of the optical fiber contained in the sensing unit, S g is the spacing of each measurement point, DTW ij is the DTW distance between the i-th measurement point and the j-th measurement point in the mapping combination X. I and J are the measured vibration responses of the i-th measurement point and the j-th measurement point respectively. w k is the k-th element in the warping path composed of I and J, and K is the number of elements in the warping path.

[0035] The Monte Carlo simulation method is adopted to calculate the DTW distance based on the measured vibration data of the measurement points, and the corresponding relationship between the measurement points and the sensing unit is estimated according to the DTW distance, so as to realize the accurate positioning of the measurement points in the distributed sensing array.

[0036] The Monte Carlo simulation method performs the following steps:

[0037] Step 1) Input the measured vibration signal data set Data, the number of measurement points N0 of the sensing unit, the number of vibration sensing units M0, and preset the stopping parameter C, where C < 1;

[0038] Step 2) Repeatedly draw data from the measured vibration signal data set Data without replacement according to the Monte Carlo simulation;

[0039] Step 3) Denote the vibration signal data set drawn for the i-th time as Data i , and after preprocessing the vibration signal data set Data i , it is divided into multiple mapping combinations with a window length of N0 and a step length of 1, and its DTW distance vector is calculated;

[0040] Step 4) Search for the peak value of the DTW distance vector based on the minimum value peak-seeking algorithm, and extract the mapping combination with a peak-to-peak spacing of N0 until the sum of the number of extracted mapping combinations and the number of converged mapping combinations reaches M0, and obtain a set of mapping combination results corresponding to M0 sensing units;

[0041] Step 5) Repeat steps 3)-4) for a preset number of times, update the cumulative frequency distribution curve of the unconverged mapping combinations corresponding to each sensing unit based on multiple sets of mapping combination results, extract the maximum likelihood estimation result of the cumulative frequency distribution curve with the maximum value peak-seeking algorithm, and output the maximum probability mapping combination of the sensing unit;

[0042] Step 6) Repeat Step 5). When the same maximum probability mapping combination is output for a continuously preset second number of times, it is considered that the mapping combination has converged;

[0043] Step 7) When the number of newly converged mapping combinations reaches C*M0, use all the converged mapping combinations as conditional parameters and return to Step 3) to exclude the data corresponding to the vibration signals in the converged mapping combinations as the vibration signal dataset Data i , repeat the above loop until the total number of converged mapping combinations reaches M0;

[0044] Step 8) Output the converged mapping combination X.

[0045] The calculation method of the regional cross-power spectral density is as follows:

[0046] Calculate the cross-power spectrum of the vibration signals of each measurement point in the same region, take the amplitude of the cross-power spectrum to obtain the amplitude spectrum of the cross-power spectrum; normalize the amplitude spectrum of the cross-power spectrum to obtain the cross-power spectral density; calculate the mean value of the cross-power spectral densities of different measurement points in the same region to obtain the regional cross-power spectral density, and store it as three-dimensional data of (measurement point ID, region ID, plate ID).

[0047] S5 includes the following steps:

[0048] S51, calculate the difference between the regional cross-power spectral densities of the area of the to-be-identified paving panel and the reference panel:

[0049] ΔP fij (f) = P fij,l (f) - P fij,h (f)

[0050] In the formula, ΔP fij (f) represents the difference in regional cross-power spectral density at frequency f, and ΔP fij,l (f) and ΔP fij,h (f) respectively represent the regional cross-power spectral densities of the corresponding area of the to-be-identified paving panel and the reference area of the reference panel;

[0051] S52, integrate the difference in the regional cross-power spectral density:

[0052]

[0053] In the formula, DI void represents the identification index of the delamination area, and a and b respectively represent the lower and upper frequency limits;

[0054] S53. Determine the void location of the pavement slab to be identified according to the void area identification index: Use the area ID as the abscissa and the identification index of the corresponding void area of the area ID as the ordinate to draw a two-dimensional image, identify the inflection points of the two-dimensional image, and the area corresponding to the identified inflection points is the location of the void area of the pavement slab to be identified.

[0055] In S1, before calculating the cross-power spectral density of the whole slab, preprocess the vibration signal. The preprocessing method used is one or a combination of band-pass filtering, high-pass filtering, and low-pass filtering.

[0056] Compared with the prior art, the present invention has the following beneficial effects:

[0057] 1. Low cost: The present invention uses distributed optical fibers, which can collect vibration information at each point along the pavement slab, thereby realizing distributed measurement, having a higher measurement range and resolution, and a low installation cost, and can meet the detection requirements of long-distance and large-scale cement concrete pavements.

[0058] 2. High precision: By comparing the distribution differences of the cross-power spectral density of the whole slab in the frequency domain, the present invention can more sensitively capture the changes in vibration characteristics caused by voids, and then more accurately judge the existence of voids; by calculating the cross-power spectral density of the area and performing difference integration to obtain the void area identification index, the present invention can more accurately analyze the changes in vibration signals and improve the measurement accuracy.

[0059] 3. Good adaptability: When partitioning in this application, a clear corresponding relationship between each measurement point and each area within the slab can be accurately established. This flexible partitioning strategy can better adapt to pavement slabs with different structures and improve the accuracy of void location judgment. Description of the Drawings

[0060] Figure 1 is the flowchart of the method of the present invention;

[0061] Figure 2 is the layout scheme of full-scale precast concrete pavement slab specimens in an embodiment;

[0062] Figure 3 is the distribution diagram of distributed vibration optical fiber units in an embodiment;

[0063] Figure 4 is the comparison diagram of cross-power spectral density when P1&P2 are excited at the slab edge in an embodiment;

[0064] Figure 5 is the comparison diagram of cross-power spectral density when P1&P2 are excited in the middle of the slab in an embodiment;

[0065] Figure 6The cross-power spectral density comparison diagram of P1&P2 when excited at the plate corner in an embodiment;

[0066] Figure 7 The vibration response result diagram of data at two measurement points within the same mapping combination in an embodiment;

[0067] Figure 8 The DTW index result of the mapping combination under single-load excitation in an embodiment;

[0068] Figure 9 The debonding area identification result diagram in an embodiment. Detailed implementation manners

[0069] The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. This embodiment is implemented on the premise of the technical solution of the present invention, and detailed implementation manners and specific operation processes are given, but the protection scope of the present invention is not limited to the following embodiments.

[0070] Embodiment 1

[0071] This embodiment provides a method for identifying debonding at the bottom of precast pavement slabs, as Figure 1 shown, including the following steps:

[0072] S1. Obtain the vibration signals of each measurement point of the reference slab and the slab to be identified, and calculate the cross-power spectral density of the entire slab respectively.

[0073] In this embodiment, the vibration signals are collected by forming a sensing unit by winding optical fibers. A plurality of measurement points are distributed in each sensing unit. By forming a sensing unit by winding optical fibers, the number of measurement points per unit area can be increased, and the vibration measured by the optical fiber can be corresponded to the actual planar position.

[0074] Before calculating the cross-power spectral density of the entire slab, preprocess the vibration signals. The preprocessing method used is one or a combination of band-pass filtering, high-pass filtering, and low-pass filtering.

[0075] The calculation process of the cross-power spectral density of the entire slab includes the following steps:

[0076] S11. Obtain the vibration signal x i (t) of each measurement point, where i represents the measurement point number. Calculate the cross-power spectrum based on the vibration signals of all measurement points within the same pavement slab, and store the data in the form of (measurement point ID, slab ID). The calculation method of the cross-power spectrum is:

[0077]

[0078] In the formula: P xy(f) represents the cross-correlation function of signals x(n) and y(n) in the frequency domain, which is used to analyze the resonance characteristics of signals at different frequencies. x(n) and y(n) represent the vibration signals at two different measurement points, N is the number of samples, y*(n + f) represents the complex conjugate of signal y(n + f), f is the frequency variable, representing different frequency points in the frequency domain, and the normalization factor is used to ensure P xy represents the average power or energy; the calculated cross-power spectrum P xy (f) is a complex value, and its real and imaginary parts respectively reflect the amplitude and phase information of the signal at different frequencies.

[0079] S12, take the magnitude of the cross-power spectrum P xy (f) to obtain the amplitude spectrum of the cross-power spectrum |P xy (f)|:

[0080]

[0081] In the formula: and are respectively the real and imaginary parts of the complex number P xy (f).

[0082] S13, normalize the amplitude spectrum of the cross-power spectrum so that the sum is 1 to obtain the cross-power spectral density (CPSD):

[0083]

[0084] In the formula, ∑ f |P xy (f)| is the sum of the magnitudes at all frequency points.

[0085] S14, calculate the mean value of the cross-power spectral density of different measurement points within the whole plate to obtain the cross-power spectral density P f :

[0086]

[0087] In the formula, M represents the number of measurement points within the plate, is the cross-power spectral density at the m-th measurement point at frequency f. The average normalized value P f reflects the average energy distribution characteristics of the signal at frequency f by taking the mean value of the normalization results of all measurement points within the paving plate.

[0088] S2, compare the distribution of the cross-power spectral density of the whole plate of the paving plate to be identified and the reference plate in the frequency domain to determine whether the paving plate to be identified is void.

[0089] S3. When the judgment result is that there is voiding, partition the pavement slab to be recognized and the reference slab. One area corresponds to one sensing unit, and establish the corresponding relationship between each measuring point and each area within the slab.

[0090] For the convenience of analysis, the concrete pavement slab is often partitioned when sensors are buried in it, and buried densely in fixed areas. When analyzing the foundation conditions of each area within the slab, it is necessary to establish a mapping method between the optical fiber measuring points and the positions of different areas within the slab, and locate the position of the measuring point within the slab area according to the time-domain curve of the sensor. The vibration data of each measuring point are grouped according to the location where they are located, and cross-power spectral density analysis is carried out, that is, calculate the cross-power spectral density (CPSD) for different measuring points in the same area within the slab.

[0091] The method for establishing the corresponding relationship between each measuring point and each area within the slab is as follows:

[0092] Compare the vibration responses between each measuring point, that is, the time-domain curve of the vibration signal. The measuring points in the same vibration sensing unit can be approximately regarded as the same position in space. Therefore, the vibration responses of these measuring points are highly similar. The dynamic time warping (DTW) algorithm is used to calculate the similarity of the vibration responses of each measuring point, and the Monte Carlo simulation method is used to estimate the mapping relationship between the actual measuring points and the sensing units.

[0093] The dynamic time warping algorithm calculates the similarity between vibration signals by optimizing the warping path. The smaller the DTW distance, the higher the signal similarity. Among them, the DTW distance of the mapping combination X is calculated by the following formula:

[0094]

[0095] In the formula, N0 is the number of measuring points of the sensing unit, L0 is the length of the optical fiber contained in the sensing unit, S g is the distance between each measuring point, DTW ij is the DTW distance between the i-th measuring point and the j-th measuring point in the mapping combination X. I and J are the measured vibration responses of the i-th measuring point and the j-th measuring point respectively, and w k is the k-th element in the warping path composed of I and J, and K is the number of elements in the warping path.

[0096] The Monte Carlo simulation method is used to estimate the corresponding relationship between the measuring points and each area within the slab with the measured data, and finally realize the accurate positioning of the measuring points in the distributed sensing array. To enhance the robustness of the method and reduce the influence of the uncertainty of the load distribution on the mapping of the measuring points, the Monte Carlo simulation is used to obtain the cumulative frequency distribution curve and gradually approximate the actual mapping combination result. To prevent the result from not converging due to data randomness, loop iteration is added and the stopping parameter C is set in the Monte Carlo simulation process, and the converged mapping combination is used as the conditional parameter for the subsequent loop. Specifically, the Monte Carlo simulation method performs the following steps:

[0097] Step 1) Input the measured vibration signal dataset Data, the number of measurement points N0 of the sensing unit, the number of vibration sensing units M0, and preset a cut-off parameter C, where C < 1;

[0098] Step 2) Repeatedly draw data from the measured vibration signal dataset Data without replacement according to Monte Carlo simulation;

[0099] Step 3) Denote the vibration signal dataset drawn in the i-th draw as Data i , and after the vibration signal dataset Data i is preprocessed, it is divided into multiple mapping combinations with a window length of N0 and a step length of 1, and its DTW distance vector is calculated;

[0100] Step 4) Search for the peak of the DTW distance vector based on the minimum value peak-seeking algorithm, extract the mapping combination with a peak-to-peak distance of N0 until the sum of the number of the extracted mapping combinations and the number of the already converged mapping combinations reaches M0, and obtain a set of mapping combination results corresponding to M0 sensing units;

[0101] Step 5) Repeat Steps 3)-4) 50 times, update the cumulative frequency distribution curve of the unconverged mapping combinations corresponding to each sensing unit based on multiple sets of mapping combination results, extract the maximum likelihood estimation result of the cumulative frequency distribution curve with the maximum value peak-seeking algorithm, and output the maximum probability mapping combination of this sensing unit;

[0102] Step 6) Repeat Step 5), and when the same maximum probability mapping combination is output continuously for 5 times, it is considered that this mapping combination has converged;

[0103] Step 7) When the number of newly converged mapping combinations reaches C*M0, use all the converged mapping combinations as conditional parameters, return to Step 3), and use the data of the vibration signals corresponding to the removed converged mapping combinations as the vibration signal dataset Data i , and repeat the above loop until the total number of converged mapping combinations reaches M0;

[0104] Step 8) Output the converged mapping combination X.

[0105] S4. Calculate the regional cross-power spectral density of different regions within the panel to be identified and the reference panel respectively.

[0106] The calculation method of the regional cross-power spectral density can refer to the calculation method of the cross-power spectral density of the whole panel mentioned above, except that the measured point data used is only the measured points within the corresponding region. The specific steps are as follows:

[0107] Calculate the cross-power spectrum for the vibration signals of each measurement point in the same area, take the amplitude of the cross-power spectrum to obtain the amplitude spectrum of the cross-power spectrum; normalize the amplitude spectrum of the cross-power spectrum to obtain the cross-power spectral density; calculate the mean value of the cross-power spectral density of different measurement points in the same area to obtain the regional cross-power spectral density, and store it as three-dimensional data of (measurement point ID, area ID, board ID).

[0108] S5. Calculate the difference between the regional cross-power spectral densities of the corresponding areas of the pavement slab to be identified and the reference slab, and perform integration to obtain the delamination area identification index, and judge the delamination position of the pavement slab to be identified according to the delamination area identification index.

[0109] Specifically, S5 includes the following steps:

[0110] S51. Calculate the difference between the regional cross-power spectral densities of the corresponding areas of the pavement slab to be identified and the reference slab:

[0111] ΔP fij (f) = P fij,l (f) - P fij,h (f)

[0112] In the formula, ΔP fij (f) represents the difference in regional cross-power spectral density at frequency f, and ΔP fij,l (f) and ΔP fij,h (f) respectively represent the regional cross-power spectral densities of the corresponding area of the pavement slab to be identified and the reference area of the reference slab. By calculating the difference between the two, the energy distribution difference between the delamination area and the reference area at frequency f can be quantified, providing a basis for subsequent delamination area identification.

[0113] S52. Integrate the difference in regional cross-power spectral density:

[0114]

[0115] In the formula, DI void represents the identification index of the delamination area, which is obtained by integrating the cross-spectral density difference ΔP fij (f) within the frequency range [a, b], where a and b represent the lower and upper frequency limits respectively.

[0116] S53. Judge the delamination position of the pavement slab to be identified according to the delamination area identification index: draw a two-dimensional image with the area ID as the abscissa and the identification index of the delamination area corresponding to the area ID as the ordinate, identify the inflection point of the two-dimensional image, and the area corresponding to the identified inflection point is the position where the delamination area of the pavement slab to be identified is located.

[0117] Embodiment 2

[0118] This embodiment details the method of Example 1 with a full-scale precast pavement slab void identification experiment.

[0119] The experiment is based on a 4m×4m full-scale precast concrete pavement (PCP) slab specimen, which is cast using C50 cement concrete. According to the JTG3420-2020 specification, the compressive strength of the casting material is measured as 55MPa, the flexural strength is 5MPa, and the flexural tensile elastic modulus is 34GPa. After the precast specimen is cast and cured for 28 days, it is transported to an indoor full-scale test site. The slab is placed on a 3cm thick compacted and leveled sand layer, and below the sand layer is a 20cm thick cement stabilized macadam base. The two specimens are connected by Conda and named P1 and P2 respectively. Set P1 slab as the control group, and ensure the healthy state of the slab bottom throughout the experiment, that is, the reference slab. P2 slab is the experimental group. During the experiment, different void positions are simulated by lifting the slab and removing the sand from below the designated area, and the edge void and corner void forms of the slab are set as Figure 2 shown.

[0120] The falling weight deflectometer (FWD) is used as the excitation load in the test, and the pavement slab is excited at 6 points, which are distributed at the edge, middle, and corner of the two slabs respectively. The specific information is as Figure 2 shown. A DAS fiber optic demodulator with a minimum spatial resolution of 4m is used. Considering the construction convenience, the number of winding turns of each unit is selected as 4 turns, the diameter is 0.4m, and the length of the optical fiber contained in each unit is 5m.

[0121] During the subsequent modal analysis, the distance between adjacent distributed vibration optical fiber units is set as 0.1m to ensure a spatial resolution of 0.5m. Considering the modal analysis requirements of this structure, a total of 36 distributed vibration optical fiber units are arranged in a 6-row and 6-column layout. The distance between rows and columns is 0.1m, as Figure 3 shown.

[0122] The vibration signals measured by the fiber optic demodulator are output, and there are a total of 144 measurement points. The cross-power spectral density of the entire slab is calculated for each of the P1 and P2 slabs, and the CPSD diagrams of the edges, middle, and corners of the P1 and P2 slabs are compared. The results are as Figure 4 , 5 , and 6 shown. It is found that there are obvious differences between the CPSD diagrams of the healthy slab and the void slab: the CPSD of the void slab is concentrated in the high-frequency distribution, while that of the reference slab is concentrated in the low-frequency.

[0123] The concrete panel is partitioned to establish the corresponding relationship between each measurement point and each area within the slab. The vibration responses of the measurement points in the same vibration sensing unit are relatively similar. Figure 7Shows the vibration response of continuous measurement points after data preprocessing. The signal phases and amplitudes from measurement points 151 to 153 are relatively similar, but different from those of measurement points 150 and 154, indicating that measurement points 151 to 153 belong to the same sensing unit, while measurement points 150 and 154 are located in adjacent units or the connection section between units.

[0124] The dynamic time warping (DTW) is used to evaluate the similarity of the signals of the optical fiber measurement points of P1 and P2. The Monte Carlo simulation method is adopted to estimate the corresponding relationship between the measurement points and each area in the plate with the measured data. In this embodiment, the DWT of different mapping combinations of the measurement points of 330 m long part of the optical fiber buried in the plate and the sensing units under a certain working condition is selected, as Figure 8 shown. The smaller the DTW, the higher the similarity of the vibration response curves within the mapping combination, and the closer the mapping combination is to the actual layout.

[0125] Figure 8 The DTW index of the mapping combination in the dashed box fluctuates significantly. The distance between the extreme points identified by the peak seeking algorithm is approximately equal to the array unit length L0, indicating that the mapping result at this place is significant. This may be because this area is close to the load wheel path and is significantly affected by the load excitation. Set the cut-off parameter C to 0.2, input the corresponding vibration data, and obtain the frequency distribution of the mapping combinations converged by the Monte Carlo simulation. The coordinates of the sensing units are sequentially assigned according to the optical fiber series path to realize the spatial reconstruction from the measurement points to the sensing units and then to the plane position. Finally, the corresponding relationship between the measurement point labels and the partition numbers of P1 and P2 plates and the coordinates of the areas in the plate is selected as shown in Table 1.

[0126] Table 1 Corresponding relationship between some measurement points of P2 and partitions

[0127] Measurement point ID Area ID Cir_x Cir_y Measurement point ID Area ID Cir_x Cir_y 228 1 1 1 253 7 2 6 229 1 1 1 254 7 2 6 230 1 1 1 255 8 2 5 232 2 1 2 256 8 2 5 233 2 1 2 257 8 2 5 234 2 1 2 258 8 2 5 236 3 1 3 259 9 2 4 237 3 1 3 260 9 2 4 238 3 1 3 261 9 2 4 240 4 1 4 262 9 2 4 241 4 1 4 263 10 2 3 242 4 1 4 264 10 2 3 244 5 1 5 265 10 2 3 245 5 1 5 266 10 2 3 246 5 1 5 267 11 2 2 248 6 1 6 268 11 2 2 249 6 1 6 269 11 2 2 250 6 1 6 270 11 2 2 251 7 2 6 271 12 2 1 252 7 2 6 272 12 2 1

[0128] Calculate the partition cross-power spectral density (CPSD). For P1 and P2 plates respectively, calculate the cross-power spectral density P xy (f) between different measurement points within the same optical fiber unit, and then calculate the normalized power spectral density P f of each measurement point at each amplitude value. Finally, calculate the average value of the normalized power spectral density of each area in the plate.

[0129] Calculate the difference of the cross-power spectral density of the areas and integrate it for void location. There is a void in P2, and P1 is used as the reference plate. Subtract the partition CPSD of P2 and P1 correspondingly, and integrate the obtained cross-spectrum difference. The maximum peak value of the integration is the identified void area, as Figure 9 shown.

[0130] The preferred specific embodiments of the present invention have been described in detail above. It should be understood that those of ordinary skill in the art can make many modifications and variations based on the concept of the present invention without creative efforts. Therefore, all technical solutions that can be obtained by those skilled in the art in this technical field based on the concept of the present invention through logical analysis, reasoning, or limited experiments on the basis of the prior art should fall within the protection scope determined by the claims.

Claims

1. An identification method for the void under the prefabricated paving slab, characterized in that It includes the following steps: S1. Obtain the vibration signals of each measuring point on the reference board and the paving board to be identified, and calculate the overall board cross-power spectral density respectively; S2. Compare the distribution of the overall board cross-power spectral density of the paving board to be identified and the reference board in the frequency domain, and judge whether the paving board to be identified is delaminated; S3. When the judgment result is that there is delamination, partition the paving board to be identified and the reference board, with one area corresponding to one sensing unit, and establish the corresponding relationship between each measuring point and each area in the board; S4. Calculate the regional cross-power spectral density of different areas within the boards of the paving board to be identified and the reference board respectively; S5. Calculate the difference between the regional cross-power spectral density of the corresponding areas of the paving board to be identified and the reference board, and perform integration to obtain the delamination area identification index, and judge the delamination position of the paving board to be identified according to the delamination area identification index.

2. The method for identifying the bottom void of an assembled paving slab according to claim 1, characterized in that The vibration signals are collected by forming a sensing unit by winding optical fibers, and multiple measuring points are distributed in each sensing unit.

3. The method for identifying the bottom void of an assembled paving slab according to claim 1, characterized in that The calculation method of the overall board cross-power spectral density is as follows: Obtain the vibration signal x of each measurement point i (t), where i represents the measurement point number. Calculate the cross-power spectrum based on the vibration signals of all measurement points within the same paving slab, and store the obtained data in the form of (measurement point ID, slab ID). The calculation method of the cross-power spectrum is as follows: where: P xy (f) represents the cross-correlation function of signals x(n) and y(n) in the frequency domain, which is used to analyze the resonance characteristics of signals at different frequencies. x(n) and y(n) represent the vibration signals of two different measuring points. N is the number of samples. y*(n + f) represents the complex conjugate of signal y(n + f). f is the frequency variable, representing different frequency points in the frequency domain. The normalization factor is used to ensure that P xy represents the average power or energy; the calculated cross-power spectrum P xy (f) is a complex value, and its real and imaginary parts respectively reflect the amplitude and phase information of the signal at different frequencies; Take the magnitude of the cross-power spectrum P xy (f) to obtain the amplitude spectrum of the cross-power spectrum |P xy (f)|: wherein: and are the real part and the imaginary part of the complex number P xy (f), respectively; Normalize the amplitude spectrum of the cross-power spectrum to obtain the cross-power spectral density: Wherein, ∑ f |P xy (f)| is the sum of the amplitudes at all frequency points; Calculate the mean value of the cross-power spectral density at different measurement points to obtain the cross-power spectral density P of the entire plate f : Where M represents the number of measurement points inside the plate, is the cross-power spectral density of the m-th measurement point at frequency f.

4. A method for identifying the bottom void of an assembled paving slab according to claim 1, characterized in that, The specific method for establishing the corresponding relationship between each measuring point and each area in the board is as follows: Compare the vibration responses between each measuring point, that is, the time-domain curve of the vibration signal, calculate the similarity of the vibration responses of each measuring point by using the dynamic time warping algorithm, and use the Monte Carlo simulation method to estimate the mapping relationship between the actual measuring point and the sensing unit.

5. A method for identifying the bottom void of an assembled paving slab according to claim 4, characterized in that The dynamic time warping algorithm calculates the similarity between vibration signals by optimizing the warping path. The similarity is characterized by the DTW distance. The smaller the DTW distance, the higher the signal similarity. Among them, the DTW distance of the mapping combination X is calculated by the following formula: Wherein, N0 is the number of measurement points of the sensing unit, L0 is the length of the optical fiber contained in the sensing unit, S g is the spacing of each measurement point, DTW ij is the DTW distance between the ith measurement point and the jth measurement point in the mapping combination X, I and J are the measured vibration responses of the ith measurement point and the jth measurement point respectively, w k is the kth element in the warping path composed of I and J, and K is the number of elements in the warping path.

6. The method for identifying the bottom void of an assembled paving slab according to claim 4, characterized in that Adopt the Monte Carlo simulation method, calculate the DTW distance based on the measured vibration data of the measuring points, and estimate the corresponding relationship between the measuring points and the sensing unit according to the DTW distance to realize the accurate positioning of the measuring points in the distributed sensing array.

7. A method for identifying the bottom void of an assembled paving slab according to claim 4, characterized in that The Monte Carlo simulation method performs the following steps: Step 1) Input the measured vibration signal data set Data, the number of measuring points N0 of the sensing unit, the number of vibration sensing units M0, and preset the stopping parameter C, where C < 1; Step 2) Repeatedly draw data from the measured vibration signal data set Data without replacement according to the Monte Carlo simulation; Step 3) Denote the vibration signal dataset extracted at the i-th time as Data i , the vibration signal dataset Data i After preprocessing, it is divided into multiple mapping combinations with a window length of N0 and a step size of 1, and its DTW distance vector is calculated; Step 4) Search for the peak value of the DTW distance vector based on the minimum value peak search algorithm, extract the mapping combination with the peak-to-peak spacing N0 until the sum of the number of the extracted mapping combinations and the converged mapping combinations reaches M0, and obtain a set of mapping combination results corresponding to M0 sensing units; Step 5) Repeatedly execute Step 3)-Step 4) for a preset first number of times, update the cumulative frequency distribution curve of the unconverged mapping combinations corresponding to each sensing unit based on multiple sets of mapping combination results, and extract the maximum likelihood estimation result of the cumulative frequency distribution curve by the maximum value peak search algorithm, and output the maximum probability mapping combination of the sensing unit; Step 6) Repeatedly execute Step 5). When the same maximum probability mapping combination is output for a preset second number of times continuously, it is considered that the mapping combination has converged; Step 7) When the number of newly added converged mapping combinations reaches C*M0, use all the converged mapping combinations as conditional parameters and return to Step 3) to exclude the data corresponding to the vibration signals in the converged mapping combinations as the vibration signal dataset Data i , repeat the above loop until the total number of converged mapping combinations reaches M0; Step 8) Output the converged mapping combination X.

8. The method for identifying the bottom void of an assembled paving slab according to claim 1, characterized in that, The calculation method of the regional cross-power spectral density is as follows: Calculate the cross-power spectrum for the vibration signals of each measurement point in the same area, take the amplitude of the cross-power spectrum to obtain the amplitude spectrum of the cross-power spectrum; normalize the amplitude spectrum of the cross-power spectrum to obtain the cross-power spectral density; calculate the mean value of the cross-power spectral density of different measurement points in the same area to obtain the regional cross-power spectral density, and store it as three-dimensional data of (measurement point ID, area ID, plate ID).

9. The method for identifying the bottom void of an assembled paving slab according to claim 1, characterized in that The S5 includes the following steps: S51, calculate the difference between the regional cross-power spectral density of the paving slab to be identified and the corresponding area of the reference slab: ΔP fij P(f) = P fij,l P(f) - P fij,h P(f) where, ΔP fij (f) represents the difference in the cross-power spectral density of the region at frequency f, ΔP fij,l (f) and ΔP fij,h (f) respectively represent the cross-power spectral density of the corresponding region of the paving slab to be identified and the reference region of the reference slab; S52, integrate the difference of the regional cross-power spectral density: where DI void represents the identification index of the void area, and a and b represent the lower and upper frequency limits respectively; S53, judge the void position of the paving slab to be identified according to the void area identification index: draw a two-dimensional image with the area ID as the abscissa and the identification index of the void area corresponding to the area ID as the ordinate, identify the inflection point of the two-dimensional image, and the area corresponding to the identified inflection point is the position where the void area of the paving slab to be identified is located.

10. A method for identifying the bottom void of an assembled paving slab according to claim 1, characterized in that, In the S1, before calculating the cross-power spectral density of the whole slab, preprocess the vibration signal, and the preprocessing method used is one or a combination of band-pass filtering, high-pass filtering, and low-pass filtering.

Images