Intelligent concrete compactness detection method based on distributed optical fibers

Abstract

The invention discloses a distributed optical fiber-based concrete compactness intelligent detection method. The method comprises the following steps: acquiring a pre-established spatial mapping relationship between a distributed optical fiber mileage coordinate and a concrete member spatial position; the method comprises the following steps: collecting vibration response data in a concrete construction process in real time by using distributed optical fibers laid in a concrete member; based on the space mapping relation, analyzing the vibration response data, and extracting a vibration characteristic quantity reflecting the physical characteristics of the vibration operation; based on the vibration characteristic quantity, the concrete compactness state of the corresponding spatial position is evaluated, and a concrete compactness evaluation result is output. Qualitative vibration monitoring is upgraded into quantitative physical parameter inversion, the problems of evaluation errors and surface layer detection blind areas caused by uncertainty of manual vibration operation are effectively solved, and accurate quality evaluation of the whole life cycle is achieved.

Description

Technical Field

[0001] This invention belongs to the field of non-destructive testing of engineering quality, and in particular, it is a method for intelligent detection of concrete density based on distributed optical fiber. Background Technology

[0002] The density of concrete structures directly determines the mechanical strength and durability of engineering projects. Hidden defects such as honeycomb, pitting, and voids are often the root cause of early structural failure. Therefore, real-time, comprehensive monitoring of the internal density during the golden window of quality control—concrete pouring and hardening—is of great engineering significance for ensuring the safe service of critical infrastructure.

[0003] Current research and applications rely heavily on traditional detection methods, such as hardened core sampling or ultrasonic testing, which are both time-consuming and destructive. In recent years, distributed optical fiber acoustic sensing (DAS) technology has been introduced into this field due to its advantages of long-distance operation and interference resistance. Existing monitoring schemes primarily collect vibration signals through pre-embedded optical fibers, using the signal amplitude, energy, or time-frequency characteristics to determine the range and duration of the vibrator's action. This focuses on monitoring the presence or absence of construction actions, thereby indirectly inferring construction quality.

[0004] However, existing vibration energy monitoring technologies generally suffer from misjudgments due to unclear source-medium coupling and lack a full lifecycle accuracy calibration mechanism. Specifically, the vibration response signal is the result of the combined effect of the excitation source intensity and the medium propagation characteristics. Existing methods cannot eliminate the randomness of the excitation source caused by differences in manual operation (such as the insertion depth of the vibrator, contact state, and power fluctuations), making it impossible to distinguish whether the signal attenuation is caused by the medium's lack of compaction or by weak initial excitation, thus leading to false negatives or false positives. Furthermore, deeply buried optical fibers have low sensitivity to defects such as segregation and bleeding on the concrete surface, and single-stage vibration monitoring has a surface blind zone, lacking the closed-loop capability to use data from subsequent processes (such as smoothing and hardening) for systematic error correction. Summary of the Invention

[0005] The purpose of this invention is to provide an intelligent method for detecting the density of concrete based on distributed optical fibers, so as to solve the above-mentioned problems existing in the prior art.

[0006] The technical solution, a smart method for detecting concrete density based on distributed optical fibers, includes:

[0007] Obtain the pre-established spatial mapping relationship between distributed fiber optic mileage coordinates and the spatial location of concrete components;

[0008] By using distributed optical fibers laid in concrete components, vibration response data during concrete construction can be collected in real time.

[0009] Based on the spatial mapping relationship, the vibration response data is analyzed to extract vibration characteristic quantities that reflect the physical characteristics of the vibration compaction operation;

[0010] Based on vibration characteristic quantities, the concrete compaction state at the corresponding spatial location is evaluated, and the concrete compaction assessment results are output.

[0011] Beneficial effects: This invention upgrades qualitative vibration monitoring to quantitative physical parameter inversion, effectively solving the problems of evaluation errors and surface detection blind spots caused by the uncertainty of manual vibration operation, and realizing accurate quality assessment throughout the entire life cycle. Attached Figure Description

[0012] Figure 1 This is a flowchart illustrating the steps of an intelligent concrete density detection method based on distributed optical fiber, as provided in an embodiment of this application.

[0013] Figure 2 A flowchart illustrating the steps for determining the near-field center location of a vibration event, as provided in an embodiment of this application.

[0014] Figure 3 A flowchart illustrating the steps for calculating the excitation intensity calibration value of a vibration event, as provided in an embodiment of this application.

[0015] Figure 4 A flowchart illustrating the steps for calculating the propagation attenuation parameters of a vibration event, provided in an embodiment of this application.

[0016] Figure 5 This is a schematic diagram of the distribution of distributed optical fibers on the surface of the reinforcing steel layer, as provided in an embodiment of this application.

[0017] Figure 6 This is a schematic diagram of fiber threading inside a column provided in an embodiment of this application. Detailed Implementation

[0018] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. 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 should fall within the scope of protection of the present invention.

[0019] It should be noted that the terms include and have, and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to such process, method, product, or device.

[0020] like Figure 1 As shown, a method for intelligent detection of concrete density based on distributed optical fiber includes the following steps:

[0021] Obtain the pre-established spatial mapping relationship between distributed fiber optic mileage coordinates and the spatial location of concrete components.

[0022] In this embodiment, it is necessary to construct a spatial location model of the distributed optical fiber within the concrete component. The spatial mapping relationship serves as a bridge connecting the one-dimensional optical fiber signal and the three-dimensional solid structure. Specifically, during the rebar binding stage before concrete pouring, the distributed acoustic sensing (DAS) optical fiber is fixed to the rebar cage according to a preset path. The fiber's deployment path can adopt a serpentine, loop, or grid structure to ensure effective coverage of the monitoring area. For example, the horizontal spacing of the optical fibers should ideally be within 2 meters to ensure the sensitivity of vibration signal capture. When establishing the mapping relationship, the three-dimensional coordinates (x, y, z) of the optical fiber at key inflection points, binding points, and characteristic locations are recorded using a total station or real-time dynamic positioning (RTK) device. Simultaneously, the positioning function of the optical fiber demodulator is used to determine the corresponding optical fiber mileage coordinates (s) of these physical points. Based on these discrete corresponding point pairs, a continuous mapping function p is constructed using linear interpolation or spline interpolation algorithms. n =M(s n ); where s n p represents the mileage of the nth sampling point on the optical fiber. n This represents the three-dimensional position of the point in the component coordinate system, and M() is the spatial mapping function. The mapping relationship can be stored as a lookup table or a parameterized formula for subsequent steps.

[0023] Distributed optical fibers laid in concrete components are used to collect vibration response data in real time during concrete construction.

[0024] Specifically, the laid sensing optical fibers are connected to a distributed optical fiber acoustic wave (DAS) demodulation host. During concrete pouring, vibration, and subsequent construction, the DAS host sends probe light pulses into the optical fiber and receives Rayleigh scattered light returning along the fiber. External vibrations cause microscopic strain in the optical fiber, which in turn changes the phase of the scattered light. The DAS host demodulates the phase change to reconstruct the acoustic wave or vibration signal at various locations along the optical fiber. The vibration response data is typically represented as a spatiotemporal matrix V(s,t), where s represents the fiber optic mileage and t represents time.

[0025] In some preferred embodiments, to achieve three-dimensional monitoring of large-volume concrete structures, distributed optical fibers can be deployed in layers along the thickness direction of the component. For example, an upper fiber optic mesh can be deployed approximately 100 mm from the top surface of the component, a middle fiber optic mesh can be deployed at half the thickness of the component, and a lower fiber optic mesh can be deployed approximately 100 mm from the bottom surface of the component. This allows for a more comprehensive capture of the propagation characteristics of vibration waves at different depths.

[0026] Based on spatial mapping relationships, vibration response data are analyzed to extract vibration characteristic quantities that reflect the physical properties of vibration compaction operations.

[0027] In this embodiment, the acquired one-dimensional time-domain vibration signal V(s,t) is mapped to the three-dimensional spatial position of the component using a spatial mapping relationship. Vibration data at each spatial position are analyzed to extract characteristic indicators that characterize the vibration compaction operation. The vibration characteristic quantities reflecting the physical properties of the vibration compaction operation can be specified into different physical quantities depending on the application scenario or accuracy requirements. As a basic implementation, the vibration characteristic quantity can be an intuitive indicator such as the amplitude, energy, and effective value (RMS) of the vibration signal; as an advanced implementation, the characteristic quantity can also be a deeper parameter reflecting the density of the medium, such as the propagation attenuation coefficient, propagation velocity, or frequency domain characteristics of the vibration wave.

[0028] Based on vibration characteristic quantities, the concrete compaction state at the corresponding spatial location is evaluated, and the concrete compaction assessment results are output.

[0029] Specifically, based on the extracted vibration characteristic quantities, a preset evaluation logic is used to determine the compaction state of the concrete. The evaluation results can be a qualitative judgment of the vibration coverage, such as whether it has been vibrated or not, or a quantitative rating of the compaction index, such as excellent, good, average, or poor. The system outputs the evaluation results to on-site construction personnel or supervisors in the form of a graphical interface, reports, or early warning signals to guide subsequent vibration supplementation work or quality acceptance. This achieves closed-loop control from signal acquisition to quality assessment.

[0030] In one possible implementation, the vibration characteristic quantity includes the vibration signal amplitude or energy intensity.

[0031] Specifically, the physical quantity that most directly reflects the intensity of vibration is selected as the characteristic index. For any spatial location p n , corresponding fiber optic mileage s n Within the time window [t, t+Δt], its vibrational characteristics can be obtained by calculating the peak amplitude or short-time energy of the signal. For example, the energy intensity E(t) can be expressed as the integral of the square of the signal amplitude, i.e., E(t) = ∫ t t+Δt (s n (τ))2 dτ; where E(t) is the short-time energy intensity at time t; s(τ) is the vibration signal amplitude at time τ; t is the start time of the current calculation window; Δt is the length of the integration time window; and dτ is the integration variable. It can quickly respond to the opening and movement of the vibrating rod.

[0032] In a further embodiment, assessing the concrete compaction state at the corresponding spatial location includes:

[0033] Based on spatial mapping relationships, the amplitude or energy intensity of vibration signals is projected onto a three-dimensional spatial model of concrete components.

[0034] In this embodiment, the system pre-loads a 3D BIM model or geometric mesh model of the concrete component. Using mapping relationships, the amplitude or energy data calculated at each sampling point along the optical fiber are assigned point-by-point to the corresponding spatial unit or node in the 3D model. For areas where the optical fiber does not directly pass through, Kriging interpolation or inverse distance weighted interpolation algorithms can be used to estimate the vibration intensity of the area based on the values ​​of surrounding optical fiber measuring points, forming vibration field data covering the entire component.

[0035] Analyze the spatial distribution of vibration signal amplitude or energy intensity to determine the coverage status of vibration operation at various parts of the concrete component.

[0036] Specifically, the system determines the execution status of vibration compaction by analyzing the spatial distribution characteristics of the vibration field data. When the vibrator operates, it generates high-intensity vibration waves, the energy center of which corresponds to the insertion position of the vibrator, and the energy decreases with distance. Therefore, the local maxima of energy intensity in space usually indicate the point of action of the vibrator, while the connected regions of the high-energy area represent the coverage area of ​​the vibration compaction operation.

[0037] In an exemplary embodiment, determining the coverage status of vibration operations at various parts of a concrete member includes:

[0038] Obtain the preset effective vibration threshold; compare the vibration signal amplitude or energy intensity corresponding to each spatial location with the effective vibration threshold point by point; when the vibration signal amplitude or energy intensity at a certain spatial location continuously exceeds the effective vibration threshold, it is determined that the vibration operation at that spatial location has been completed.

[0039] In this embodiment, in order to eliminate interference from background noise such as construction machinery movement and formwork vibration, it is necessary to set an effective vibration threshold A. th The effective vibration threshold can be set based on the actual measured background noise level, for example, 3 to 5 times the background noise amplitude. The specific judgment logic is as follows: for any location in the model, monitor its vibration amplitude or energy intensity. If this value exceeds the effective vibration threshold A... thFurthermore, the duration of the over-limit condition exceeded the preset minimum vibration time T. min If the vibration intensity is consistently below the threshold, or the duration of vibration exceeding the limit is insufficient, then the location is determined to have been vibrated or to have met the required standard. Conversely, if the vibration intensity remains below the threshold or the duration of vibration exceeding the limit is insufficient, it is determined to have not been vibrated or to have been under-vibrated. The shortest vibration time T is specified in the table. min The time can be determined based on the concrete slump and the thickness of the component, for example, 5 seconds or 10 seconds.

[0040] In some implementations, numerical simulation results can be used to assist in the determination. For example, before pouring, finite element simulation can be used to simulate the theoretical vibration field distribution when the vibrator is working at different positions. The measured data can be compared with the simulation results. If the measured attenuation trend is consistent with the theory and the amplitude meets the standard, the effectiveness of vibration can be further confirmed.

[0041] In a further embodiment, the method further includes: generating a visual distribution map of vibration energy of concrete components based on spatial mapping relationships and vibration signal amplitude or energy intensity; using different colors or brightness to mark the differences in vibration energy at different spatial locations, and intuitively displaying blind spots or overlapping areas of vibration operation.

[0042] Specifically, to facilitate intuitive understanding by on-site personnel, the system transforms the calculation results into visual graphs. For example, it can generate time-series diagrams of multi-channel energy changes, or cloud maps reflecting the distribution of vibration energy at the current moment or over a cumulative period. In the cloud maps, color coding is used to represent energy levels: for example, red represents strong vibration areas, i.e., the center of the vibration point; yellow represents the effective influence area; and blue represents low vibration areas, i.e., blind zones. Through visualization, construction personnel can see which areas have been properly vibrated (red / yellow covered areas), which areas have missed vibration (blue areas), and whether there is over-vibration (red areas remaining for extended periods). This intuitive feedback mechanism is crucial for on-site quality control. This embodiment is suitable for basic application scenarios with high real-time requirements, limited computing resources, or where only the vibration coverage needs to be determined.

[0043] In a preferred implementation, the raw vibration response data is preprocessed to improve the signal-to-noise ratio. Specifically, the spatiotemporal matrix V(s,t) is de-DC drifted, i.e., V'(s... n , t m ) = V(s n , t m ) - mean(V(s n ,:)), to eliminate low-frequency trend terms; where V'(s n , t m V(s) represents the amplitude of the preprocessed vibration signal; n , t m ) represents the amplitude of the original acquired vibration signal; s nFor the nth fiber spatial sampling point; t m For the m-th time sampling point; mean(V(s) n , :)) indicates that the average value of all time series data for that spatial point is calculated. The data is filtered using a bandpass filter, and the frequency range to be retained is usually set to 20 Hz to 150 Hz, which covers the fundamental frequency of commonly used concrete vibrators (usually 50 Hz or 60 Hz) and its second and third harmonics, while filtering out high-frequency noise and low-frequency environmental interference.

[0044] Optionally, the preprocessing of the raw acquired vibration response data includes an amplitude normalization step, specifically: calculating the vibration signal amplitude statistics for each spatial location throughout the entire acquisition period. For fiber optic mileage location s... n Calculate the root mean square amplitude of its vibration signal, denoted as A. rms (s n The calculation formula is as follows:

[0045] A rms (s n = sqrt((1 / M) * ∑ m=1 M (V'(s n , t m )) 2 );

[0046] Where A rms (s n ) represents the position s n The root mean square magnitude; sqrt represents the square root operation; M is the total number of time sampling points; ∑ m=1 M This represents the summation of m from 1 to M. Calculate the median of the root mean square magnitudes for all spatial locations, denoted as A. med The median is more robust than the mean, effectively resisting the interference of local outliers. The vibration signal at each spatial location is normalized to obtain the normalized vibration signal, denoted as V. norm (s n , t m The calculation formula is as follows:

[0047] V norm (s n , t m ) = V'(s n , t m ) / (A rms (s n ) + ε*);

[0048] Where V norm (s n , t mA represents the normalized amplitude of the vibration signal; rms (s n ) represents the root mean square magnitude at that location; ε* is the numerical stability parameter used to prevent division by zero errors, and its preferred value is 1e. -9 Amplitude normalization can eliminate systematic deviations in amplitude at various locations along the optical fiber caused by factors such as differences in sensor sensitivity, optical power attenuation, and changes in coupling efficiency. Normalization processing ensures that the vibration responses at different locations are comparable during subsequent feature extraction.

[0049] In some implementations, amplitude normalization can also be achieved using time-varying normalization methods. Specifically, the entire acquisition period is divided into several time windows, and the length of each time window is denoted as T. win The preferred value range is 10 to 60 seconds. For each time window, the local root mean square amplitude at each spatial location within that window is calculated and normalized accordingly. Time-varying normalization can adapt to the variation of excitation intensity over time during construction.

[0050] like Figure 2 As shown, according to one aspect of this application, vibration characteristic quantities reflecting the physical characteristics of the vibration operation are extracted, including determining the near-field center location of the vibration event, specifically:

[0051] Spatiotemporal energy analysis was performed on the vibration response data to identify a predetermined number of discrete, time-discontinuous vibration events.

[0052] In this embodiment, the vibration signal exhibits a distinct intermittent characteristic on the time axis. The system calculates short-time energy sequences from the preprocessed data at each sampling point along the optical fiber. Specifically, a sliding time window is set, for example, with a window length of 0.5 seconds and an overlap rate of 50%, and the sum of squares of the signal amplitude within the window is calculated. When the short-time energy at a certain location exceeds a preset event trigger threshold, it is marked as the start of an event; when the energy falls back below the end threshold and remains below it for a certain period of time, it is marked as the end of an event. The continuous data stream is divided into several independent vibration events. i Each event contains its start time index m start and end time index m end .

[0053] For each vibration event, the near-field center position corresponding to the vibration event is determined based on the peak value of the vibration energy distribution along the optical fiber.

[0054] Specifically, within the locked event time window [m start m endWithin the system, the cumulative energy of all sampling points along the optical fiber is scanned. Since the vibrator acts as a point source, the vibration energy is strongest near its point of action. Therefore, the system searches for the spatial energy maximum point, and the index of the optical fiber sampling point corresponding to this maximum point is denoted as n. i0 This is the near-field center location of the vibration event. This location physically approximates the specific spatial point where the vibrator is inserted into the concrete.

[0055] like Figure 3 As shown, in a further embodiment, extracting vibration characteristic quantities reflecting the physical characteristics of the vibration operation also includes calculating the excitation intensity calibration value of the vibration event, specifically:

[0056] Using the near-field center position as a reference, a fiber optic segment within a preset spatial range is selected as the near-field reference segment.

[0057] In this embodiment, to obtain the initial excitation information of the seismic source, it is necessary to delineate a near-field region that is not significantly affected by medium attenuation. The system sets a near-field half-width parameter N0, which is typically determined based on the spatial sampling interval Δs of the optical fiber. For example, N0 is chosen so that the near-field range covers an area of ​​approximately 0.2 to 0.5 meters around the vibrator. The near-field reference segment is defined as the index interval [n i0 -N0, n i0 The fiber segment within [+N0] represents the signal intensity primarily reflecting the input intensity of the seismic source, without being corrected for far-field medium inhomogeneities.

[0058] The vibration energy integral of the vibration event in the near-field reference section is calculated to obtain the excitation intensity calibration value of the vibration event. The excitation intensity calibration value is used to characterize the initial energy intensity input into the concrete medium during the vibration operation.

[0059] Specifically, the excitation intensity calibration value E0 i It is obtained by performing a double integral (discrete summation) on the vibration energy of all sampling points within the near-field reference segment over the duration of the event. The calculation formula is:

[0060] E0 i =Σ m=mi，start mi，end Σ n=ni0-N0 ni0+N0 (V'(s n , t m )) 2 *Δt*Δs;

[0061] Among them, E0 i Let m be the near-field excitation intensity calibration value for the i-th vibration event; i，start This is the index of the start time of the vibration event; m i，end n is the index of the end time of the vibration event. i0N0 is the near-field center position index of the vibration event; N0 is the half-width index number of the near-field reference segment; V'(s) n , t m The amplitude is the pre-processed vibration value, Δt is the time sampling interval, and Δs is the spatial sampling interval. The excitation intensity calibration value quantifies the total energy actually input into the concrete during the i-th vibration operation. E0 is calculated... i This embodiment implements an event-level self-calibration mechanism. Regardless of whether the vibrator is high-power or low-power, regardless of whether the operator inserts it deeply or shallowly, and regardless of whether the vibrator contacts the rebar, E0... i Both can objectively reflect the current state of excitation. In subsequent decay calculations, E0 can be used as the basis. i Normalization is performed as the denominator to eliminate the interference of seismic source uncertainty on the compaction assessment. Furthermore, in some optional implementations, the vibration duration T can also be calculated. d，i As an auxiliary feature, the cumulative duration for which the amplitude of the near-field center signal exceeds the effective threshold is counted to identify whether the vibration time meets the specification requirements, such as the minimum of 20 seconds as specified in the specification.

[0062] This embodiment is a fundamental step in realizing the quantitative inversion of concrete density. It solves the problem of signal source energy differences caused by differences in the power of vibration equipment, different operator techniques, and unstable contact conditions, and provides a normalized benchmark for subsequent medium parameter inversion.

[0063] like Figure 4 As shown, in a further embodiment, extracting vibration characteristic quantities reflecting the physical characteristics of the vibration operation also includes calculating the propagation attenuation parameters of the vibration event, specifically:

[0064] Calculate the far-field response energy of the vibration event at each far-field location along the optical fiber.

[0065] In this embodiment, for each vibration event, in addition to calculating the near-field energy, it is also necessary to calculate other locations along the fiber (i.e., far-field locations s). n The response energy of the far-field response. The far-field response energy E i (s n The calculation method for ) is similar to that for near-field energy, that is, for this position s n Integrating the square of the signal amplitude within the event time window over time: E i (s n )=Σ m=mi，start mi，end (V'(s n , t m )) 2 *Δt.

[0066] The normalized attenuation is calculated based on the ratio of the excitation intensity calibration value to the far-field response energy.

[0067] Specifically, the normalized decay A i (s n This reflects the proportion of vibrational energy lost relative to the intensity of the seismic source. To eliminate the influence of absolute energy dimensions, it is preferable to use the logarithmic ratio form for calculation:

[0068] A i (s n )=ln((E0 i +ε*) / (E i (s n )+ε*));

[0069] Where A i (s n ) represents the i-th event at position s n The normalized decay at point; ln is the natural logarithm function; E0 i This is the excitation intensity calibration value; ε* is a very small positive number, such as 1e. -6 It is used to prevent numerical calculation errors when the denominator is zero or the logarithm argument is zero.

[0070] Based on the spatial mapping relationship, the propagation distance of the far-field position relative to the near-field center position is calculated; the normalized attenuation is standardized using the propagation distance to obtain the propagation attenuation parameters of the vibration event on the corresponding transmission path.

[0071] In this embodiment, the normalization process includes two levels of physical correction: geometric diffusion compensation and inversion of medium absorption attenuation. During geometric diffusion compensation, as the vibration wave propagates in a three-dimensional solid, its energy naturally attenuates inversely with the square of the distance (spherical wave diffusion), regardless of the medium's density. Therefore, preferably, geometric compensation is first applied to the far-field response energy:

[0072] E corrected (s n )=E i (s n )*(d n / d0) 2 ;

[0073] Where E corrected (s n ) represents the energy value after geometric diffusion compensation; d n For the far-field position s n To the near-field center s ni0 The Euclidean distance is d0, where d0 is a preset reference distance, for example, 0.1 meters. When performing the inversion of medium absorption attenuation, the geometrically compensated energy is substituted into the attenuation formula, or the attenuation coefficient is directly calculated using distance normalization. That is, the propagation attenuation parameter is:

[0074] k i (s n )=A i (s n ) / (|s n -s ni0 |+δ);

[0075] Where |s n -s ni0 | for position s n To the near-field center s ni0 The distance along the fiber or the spatial distance; δ is the distance stability parameter, the propagation attenuation parameter k. i This directly reflects the damping characteristics of concrete: the lower the density, the more internal voids, and the stronger the scattering and absorption, k i The larger the value.

[0076] In some alternative implementations, the propagation attenuation parameter can also be calculated using linear regression. This involves selecting a segment of the optical fiber originating from the near-field center and calculating the normalized attenuation A at each point within that segment. i (s n ) and propagation distance d n Perform linear fitting: A i (s n )=k i *d n +c, where c is a constant offset. The fitted slope k i This is the average propagation attenuation coefficient of the event in the current region.

[0077] According to one aspect of this application, the propagation attenuation parameter can also be calculated using the multi-segment median method, specifically: the effective range of the vibration event is divided into several continuous spatial segments. The number of segments is denoted as Q, preferably ranging from 3 to 7. Each segment contains several continuous fiber sampling positions. For each segment q, a local estimate of the attenuation parameter for all positions within that segment is calculated. The local estimate is obtained by calculating the median attenuation rate of adjacent position pairs within that segment, denoted as α. local (q). For all Q segments, the median of the local estimates is calculated as the final estimate of the propagation attenuation parameter. This final estimate is denoted as α. robust The calculation formula is as follows:

[0078] α robust = median(α local (1), α local (2), ..., α local (Q));

[0079] Where α robustThe median is the estimated value of the robust decay parameter; median indicates the median operation; α local (q) represents the estimated local attenuation parameter of the q-th segment. Using the multi-segment median method, even if some segments are affected by local interference leading to abnormal estimations, the median calculation can automatically eliminate these outliers, ensuring the robustness of the final estimation result. This method is suitable for situations where there are local defects or poor coupling along the fiber optic cable.

[0080] In one embodiment of this application, extracting vibration characteristic quantities reflecting the physical properties of the vibration operation further includes calculating the proportion of high-frequency energy in the vibration event, specifically:

[0081] Calculate the propagation speed of the vibration wave generated by the vibration event from the near-field center to various locations along the optical fiber.

[0082] In this embodiment, to obtain the elastic modulus information of the medium, it is necessary to calculate the wave velocity, which can be achieved using cross-correlation analysis or a crest tracking algorithm. Specifically, the signal at the near-field center is selected as the reference signal s. ref (t), selecting the far-field position s n The signal at that location serves as the response signal s resp (t), where t is the time variable in seconds. Calculate the cross-correlation function R(τ) = ∫0^t τ. T s ref (t)*s resp (t+τ)dt; where R(τ) is the value of the cross-correlation function at time delay τ; τ is the time delay variable in seconds; T is the signal acquisition duration in seconds; ∫0 T This represents the definite integral operation from 0 to T; s ref (t) represents the amplitude of the reference signal at time t; s resp (t+τ) represents the amplitude of the response signal at time t+τ. The cross-correlation function reaches its maximum value when the waveforms of the reference signal and the response signal are most similar; the corresponding time delay is the time required for the vibration wave to propagate from the near-field center to the far-field position. Find the time delay τ at which R(τ) reaches its maximum value. peak The calculation formula is: τ peak = argmax τ R(τ); where argmax τ This represents the value of τ that maximizes R(τ). Propagation speed v i (s n The calculation is as follows:

[0083] v i (s n )=|s n -s ni0 | / (|τ peak |+Δt sys);

[0084] Where Δt sys This is the inherent delay of the system, i.e., the time stability parameter, used to prevent division-by-zero errors when the time delay is zero. Its value should be much smaller than the minimum measurable time delay, preferably one-tenth of the time sampling interval Δt. A faster wave velocity means a harder and denser concrete medium.

[0085] In some preferred embodiments, the window length for the cross-correlation analysis is set to cover the time length of at least two to three vibration cycles. If the vibratory rod operates at a frequency of 50 Hz, then a single cycle is 0.02 seconds, and the window length is preferably set to 0.04 to 0.06 seconds. Furthermore, using a 50% window overlap rate strikes a balance between temporal resolution and computational efficiency.

[0086] The vibration response data of the vibration event is transformed in the frequency domain, and the ratio of the preset high-frequency band energy to the full-frequency band energy is calculated to obtain the high-frequency energy proportion.

[0087] Specifically, to capture the Rayleigh scattering effect caused by microscopic gaps, a Fast Fourier Transform (FFT) is performed on the vibration signal to obtain the spectrum S(f). The high-frequency cutoff frequency f is defined. c For example, take twice the fundamental frequency of the vibration, i.e., 100 Hz. Calculate the high-frequency energy E. HF =∫ fc fmax |S i (f)| 2 df, and the total energy E across the entire frequency band total =∫0 fmax |S i (f)| 2 df, high-frequency energy percentage R HF =E HF / E total ;where S i (f) is the spectral function of the vibration signal; f is the frequency variable, f max The Nyquist frequency for signal acquisition; |S i (f)| 2 Represents the power spectral density. In dense concrete, high-frequency components have lower propagation losses, R... HF Higher; however, in concrete with honeycomb-like pitting, high-frequency components are strongly filtered out, resulting in R... HF reduce.

[0088] According to one aspect of this application, the propagation speed can also be calculated using a linear regression method, specifically: multiple far-field locations within the effective range of the vibration event are selected as observation points, forming an observation point set. The position in the observation point set is denoted as s. n1 s n2Until s nK There are a total of K observation points. K represents the number of observation points, preferably ranging from 5 to 20. For each observation point s... nk The time delay relative to the near-field center is calculated using cross-correlation analysis, denoted as τ. k Simultaneously calculate the propagation distance from the observation point to the near-field center, denoted as d. k With the propagation distance d k As the independent variable, with time delay τ k Using the wave as the dependent variable, a linear regression model is established. The linear regression model assumes that the vibration wave propagates at a constant speed, meaning that the propagation distance is linearly related to the time delay. The expression for the regression model is:

[0089] τ k = (1 / v est ) * d k + b;

[0090] Where τ k d represents the time delay at the k-th observation point. k v is the propagation distance at the k-th observation point; est The propagation speed to be estimated; 1 / v est β is the regression slope, representing the propagation time per unit distance; b is the regression intercept, used to absorb systematic time shifts. The regression coefficients are solved using the least squares method. The estimated slope is denoted as β1, and its calculation formula is:

[0091] β1 = (∑ k=1 K (d k - d') * (τ k - τ')) / (∑ k=1 K (d k - d') 2 );

[0092] Where β1 is the slope estimate; d' is the arithmetic mean of all propagation distances; τ' is the arithmetic mean of all time delays; ∑ represents the summation over k from 1 to K. The propagation velocity estimate is calculated based on the slope estimate. The propagation velocity estimate is denoted as v. est Its calculation formula is: v est = 1 / β1. Calculate the coefficient of determination of the regression model, denoted as R. 2 The coefficient of determination is used to evaluate the reasonableness of the linear hypothesis. Its calculation formula is as follows:

[0093] R 2 = 1 - (∑ k=1 K (τ k– τ* k ) 2 ) / (∑ k=1 K (τ k - τ') 2 );

[0094] Where R 2 The coefficient of determination, τ*, ranges from 0 to 1. k Let τ* be the predicted time delay value for the k-th observation point. k =β1·d k +b. When the coefficient of determination R... 2 Below the preset goodness-of-fit threshold θ R When the linear propagation assumption is not met, the velocity estimate for the vibration event is marked as low confidence. Goodness-of-fit threshold θ R The preferred value is 0.85.

[0095] This embodiment utilizes information from multiple observation points for joint estimation, which can effectively suppress single-point noise interference and improve the robustness of velocity estimation; at the same time, it provides a confidence evaluation index through the coefficient of determination.

[0096] In a preferred embodiment, the extraction of vibration characteristic quantities reflecting the physical properties of the vibration operation also includes calculating the propagation speed of the vibration event and the proportion of high-frequency energy.

[0097] In this embodiment, for each spatial location p n The system may capture multiple vibration events at different times, meaning multiple vibrations cover the same point. Therefore, it is preferable to aggregate the physical parameters calculated from all valid events at that location. For example, the median of multiple calculation results can be taken: k avg (p n )=median(k i ), v avg (p n )=median(v i ), R avg (p n )=median(R HF，i Using median aggregation can effectively remove outliers caused by accidental collisions with steel bars or equipment malfunctions.

[0098] Assess the concrete compaction condition at the corresponding spatial location, including:

[0099] Construct a multidimensional feature vector containing propagation attenuation parameters, propagation speed, and high-frequency energy proportion; input the multidimensional feature vector into a preset weighted fusion formula or a pre-trained machine learning regression model to calculate the concrete density index of spatial location.

[0100] Specifically, during the actual detection phase, the system will calculate the aggregated feature vector X=[k in real time. avg v avg R avg The input is fed into the trained model. As a preferred implementation, a non-linear weighted fusion formula is used:

[0101] CI(p n )=1 / (1+exp(-(w1*(v avg / v ref )+w2*(R avg / R ref )-w3*(k avg / k ref )+b)));

[0102] Where CI(p) n ) represents the spatial location p n The concrete density index at the location; exp is the exponential function; w1, w2, w3 are the weighting coefficients of each feature; b is the bias term; v ref R ref k ref This is a reference value for standard dense concrete. The output is limited to the range (0, 1).

[0103] In a preferred implementation, the pre-defined weighted fusion formula or pre-trained machine learning regression model is constructed in the following way:

[0104] Concrete pouring tests were conducted in a pre-defined calibration area, and fiber optic vibration data and true density indices from core sampling or ultrasonic testing were collected simultaneously.

[0105] Specifically, in the first project, a parameter mapping model needs to be established. In the calibration test area, standard vibration operations are performed, and data is collected using a DAS system. After the concrete hardens, core samples are taken at different locations to determine density, or ultrasonic testing is performed to obtain wave velocity, resulting in a series of true value sample pairs {(k, v, R}. HF ) j CI true_j}, where j is the sample number, k is the propagation attenuation parameter, v is the vibration propagation velocity, and R HF For the proportion of high-frequency energy, CI true_j This is the true density index.

[0106] The propagation attenuation parameters, propagation velocity, and high-frequency energy ratio corresponding to the fiber vibration data are extracted as input features. The true density index is used as the target label. The input features are subjected to regression fitting or supervised training to determine the weight coefficients of the weighted fusion formula or the network parameters of the machine learning regression model.

[0107] In this embodiment, based on the collected ground truth samples, the fusion formula is fitted using the least squares method, or a regression model is trained using algorithms such as Support Vector Machine (SVM) or Random Forest. The goal of training is to make the predicted value output by the model as close as possible to the physical ground truth CI. true .

[0108] Among them, the concrete density index is negatively correlated with the propagation attenuation parameter, and positively correlated with the propagation speed and the proportion of high-frequency energy.

[0109] Specifically, this ensures the physical plausibility of the evaluation results: the greater the attenuation (the greater k), the lower the density; the faster the wave velocity (the greater v), the more high-frequency components are retained (R). HF The larger the value, the higher the density. Based on this, this embodiment further provides a preferred grading evaluation standard: Grade A (Excellent): CI ≥ 0.90, indicating sufficient vibration and high density; Grade B (Acceptable): 0.75 ≤ CI < 0.90, indicating that basic requirements are met; Grade C (Pending): 0.60 ≤ CI < 0.75, indicating the risk of local defects, and targeted vibration supplementation is recommended; Grade D (Unacceptable): CI < 0.60, indicating serious voids or insufficient vibration, requiring immediate rework. This embodiment achieves a precise mapping from physical parameters to engineering quality levels.

[0110] According to one aspect of this application, the method further includes calculating the confidence field during the vibration stage, specifically:

[0111] The confidence level of the vibration stage is calculated based on the number of times vibration events cover each spatial location. The confidence level of the vibration stage characterizes the sufficiency of coverage of that spatial location by effective vibration events. The calculation formula is as follows:

[0112] Conf vib (p n ) = N event (s n ) / (N event (s n ) + 1);

[0113] Conf vib (p n ) represents the spatial location p n The confidence level for the vibration stage at N is an open interval ranging from 0 to 1; event (s n ) represents the fiber optic mileage location s n The number of valid vibration events, N, is a non-negative integer. When a location is not covered by any vibration event, N... eventA value of 0 corresponds to a confidence level of 0, indicating that the compaction assessment result at that location is completely unreliable. As more and more vibration events cover a location, the confidence level gradually approaches 1, indicating that the compaction assessment result at that location becomes increasingly reliable. The formula for calculating the confidence level during the vibration stage uses a saturation function to avoid the situation where the confidence level exceeds 1 when the number of events is extremely large.

[0114] In some preferred embodiments, the condition for determining a valid vibration event is: the vibration event occurs at position s. n The peak amplitude of the vibration generated at point A exceeds the preset effective vibration threshold A. th And the vibration duration exceeds the preset minimum vibration time T. min Effective vibration threshold A th The value is three to five times the background noise amplitude; the shortest vibration time T min The value is determined based on the concrete slump and the thickness of the component, with a preferred value range of 5 to 20 seconds.

[0115] According to another aspect of this application, the calculation of the concrete density index also includes a low-confidence skipping mechanism, specifically:

[0116] Before calculating the density index, it is determined whether the confidence level of each input feature meets the minimum requirement. The minimum confidence threshold is denoted as θ. min The preferred value is 0.1. When a certain spatial location p n Confidence level of vibration stage vib (p n Below the minimum confidence threshold θ min If a location is not effectively covered by any vibration event, the density index calculation for that location is skipped, and the density index for that location is marked as invalid. Simultaneously, the output report will indicate insufficient data for that location, suggesting manual review. This low-confidence skipping mechanism ensures that when a location is not effectively covered by any vibration event, there are no vibration propagation characteristics available for analysis, and forcibly calculating the density index will produce meaningless results. This avoids providing users with misleading assessment conclusions.

[0117] According to another aspect of this application, the low-confidence skipping mechanism can also be combined with spatial interpolation, specifically:

[0118] For locations marked as invalid, spatial interpolation can be used to fill in the gaps if valid density index estimates exist in adjacent locations. Spatial interpolation can be implemented using the inverse distance weighting method. The filled density index is denoted as CI. interp (p n The calculation formula is as follows:

[0119] CI interp (p n ) = (∑k∈Nvalid w k * CI(p k )) / (∑ k∈Nvalid w k );

[0120] CI interp (p n ) represents the density index of the interpolation fill; N valid For position p n The effective neighborhood set, including distance p n Several locations that are closest to and have an effective density index; w k The inverse distance weight is equal to the position p. n With position p k The reciprocal of the distance between them; CI(p) k ) represents position p k The effective density index. The confidence level of the positions filled by interpolation should be reduced accordingly. The reduced confidence level is half of the lowest confidence level in the neighborhood.

[0121] Vibration signals during the concrete smoothing or finishing stage are extracted from the vibration response data; the surface response characteristics of the vibration signals during the smoothing or finishing stage at each spatial location are calculated to generate a surface correction amount; the surface correction amount is used to superimpose and correct the concrete density assessment results to obtain a calibrated assessment result; the surface correction amount is used to compensate for the surface density assessment error caused by the limited action depth of the vibrator.

[0122] In this embodiment, the vibratory rod typically acts on the deep layers of concrete, and is not sufficiently sensitive to the density of the protective layer or surface layer, such as the area 0 to 50 mm from the surface. In contrast, the trowel or finisher, as a surface-contact mechanical device, generates vibration waves that primarily propagate along the concrete surface, making it highly sensitive to surface laitance, segregation, or voids. Therefore, the trowel is considered a mobile secondary detection source.

[0123] Specifically, since the frequency of a trowel is typically higher and more stable than that of a vibratory tamper, the system extracts the troweling stage data segment V from the continuously acquired vibration data stream based on the timestamps recorded in the construction log or the spectral characteristics of the vibration signal. f (s, t). Calculate the surface response characteristics H. j (s n (), can be defined as the position s through which the trowel passes. n At that time, the normalized vibrational energy response is:

[0124] H j (s n )=(∫(V f (s n ,t)) 2 dt) / (Esource +ε*);

[0125] Where E source This is the reference seismic source intensity for the trowel. If the surface layer at a certain location is loose or contains a large amount of laitance, the vibration wave energy will be rapidly absorbed, leading to H... j (s n The surface correction amount ΔCI is reduced. surface (p n The correction logic is based on the following empirical rule: if the vibration assessment result CI vib The display is dense, but the smoothed response H j The abnormally low value indicates a defect in the surface layer. The correction formula can be designed as follows:

[0126] ΔCI surface (p n )=-η*(H ref -H j (s n ));

[0127] Where η is the corrected weighting coefficient, for example, taking values ​​from 0.1 to 0.2, H ref This is the reference response value for a standard dense surface layer. The final calibration evaluation result is CI. final (p n )=CI vib (p n )+ ΔCI surface (p n By using negative feedback correction, false positives and misjudgments can be effectively prevented due to overlooking surface defects caused by inadequate deep compaction.

[0128] Alternatively, the formula for calculating the surface response characteristics during the smoothing stage can also be:

[0129] H j (s n ) = (∑ m∈Wj (V f (s n , t m )) 2 *Δt) / (F0 j +ε*);

[0130] Where H j (s n Let ) be the j-th smoothing event at position s. n Surface response characteristics at the location; ∑ represents the summation operation; W j V is the set of time window indices for the j-th smoothing event; f (s n , t m F0 represents the pre-processed vibration signal during the smoothing stage.j Let be the focal intensity calibration value for the j-th smoothing event; ε* is the numerical stability parameter. The corresponding surface calibration formula is:

[0131] CI final (p n ) = CI vib (p n ) -η* (H ref - H median (s n ));

[0132] CI final (p n The result of the calibration evaluation is the compactness index after calibration; CI vib (p n ) represents the compaction index obtained during the vibration stage; η represents the surface layer correction weight coefficient; H ref H is the reference response characteristic value for a standard dense surface layer; median (s n ) represents the position s n The median of the event response characteristics after multiple smoothing processes.

[0133] According to one aspect of this application, extracting vibration signals during the concrete smoothing or finishing stage further includes constructing a set of smoothing events, specifically:

[0134] The raw vibration data from the smoothing stage undergoes the same preprocessing operations as those from the compaction stage, including DC drift removal and bandpass filtering, to obtain preprocessed vibration data for the smoothing stage. Using the same methodological framework as for compaction event detection, spatiotemporal energy analysis is performed on the preprocessed vibration data from the smoothing stage to identify several discrete smoothing events. All identified smoothing events are denoted as the smoothing event set F, and its elements are denoted as f. j That is, F equals all f j The set of events. Where j is the sequence number of the smoothing event. Each smoothing event f j It contains three elements: the event start time index denoted as m. j_start The event end time index is denoted as m. j_end The index of the location of the event's center of action is denoted as n. j0 The location of the event's center can be determined by the peak value of the vibrational energy distribution along the optical fiber during the smoothing phase.

[0135] It should be noted that the signal characteristics of troweling events and vibration events differ as follows: vibration events exhibit point source characteristics, characterized by energy concentration in a localized area and a distinct peak; troweling events exhibit area source characteristics, characterized by energy sliding along the fiber optic direction and relatively uniform amplitude. The system can segment data based on the timestamps recorded in the construction log, or automatically identify them based on signal spectrum characteristics. The operating frequency of the troweling machine is typically higher than that of the vibrator and is more stable.

[0136] According to another aspect of this application, the method further includes calculating an excitation intensity calibrator value for the smoothing event, specifically:

[0137] For each smoothing event f j With its center of action at position n j0 Based on this, a section of optical fiber within a preset spatial range is selected as the near-field reference segment. The near-field reference segment is defined in the same way as the vibration event, i.e., the index interval is n. j0 Reduce N0 to n j0 Add N0. Calculate the vibration energy integral of the smoothing event within the near-field reference segment to obtain the excitation intensity calibration value of the smoothing event, denoted as F0. j The calculation formula is as follows:

[0138] F0 j =∑ m=mj_start mj_end ∑ n=nj0-N0 nj0+N0}(V f '(s n , t m )) 2 *Δt *Δs;

[0139] Among them, F0 j Let m be the excitation intensity calibration value for the j-th smoothing event; ∑ represents the summation operation; m j_start This is the index of the start time of the smoothing event; m j_end n is the index of the end time of the smoothing event. j0 N0 is the index of the center of action of the smoothing event; N0 is the half-width index of the near-field reference segment, and its value is consistent with the vibration stage; V f '(s n , t m Δt represents the vibration signal amplitude after preprocessing during the smoothing stage; Δs represents the spatial sampling interval; and Δt represents the excitation intensity calibration value F0 for the smoothing event. j The total energy actually input to the concrete surface during the j-th troweling operation can be quantified. This calibration value will be used to normalize the far-field response to eliminate differences in excitation intensity caused by factors such as different trowel models, operating speeds, and contact pressures.

[0140] According to another aspect of this application, the method further includes calculating the surface confidence level, specifically:

[0141] Based on the coverage of each spatial location by the smoothing event, the surface confidence score is calculated. The surface confidence score characterizes the sufficiency of the effective smoothing operation covering the spatial location and the reliability of surface feature extraction. The calculation formula is as follows:

[0142] Conf surface (p n ) = N smooth (s n ) / (N smooth (s n ) + 1);

[0143] Conf surface (p n ) represents the spatial location p n The surface confidence level at N is an open interval ranging from 0 to 1; smooth (s n ) represents the fiber optic mileage location s n The number of valid smoothing events is a non-negative integer. When a location is not covered by any smoothing events, the surface confidence is 0, indicating that the surface correction amount at that location is unreliable and should rely on the assessment results of the tamping stage; when a location is covered by multiple smoothing events, the surface confidence approaches 1, indicating that the surface correction amount at that location has high reliability.

[0144] According to another aspect of this application, a weighted fusion method based on confidence level is used for superposition correction using surface correction amounts, specifically:

[0145] Define the weights for the vibration compaction stage and the surface layer stage. The weight for the vibration compaction stage is denoted as w. v (p n Its value is equal to the confidence level of the vibration stage. vib (p n The weight of the surface stage is denoted as w. s (p n Its value is equal to the surface confidence level Conf. surface (p n The calibrated density index, denoted as CI1(p), was calculated using a weighted fusion formula. n The calculation formula is as follows:

[0146] CI1(p n ) = (w v (p n CI vib (p n ) + w s (pn ) * (CI vib (p n ) +ΔCI surface (p n ))) / (w v (p n ) + w s (p n ) + ε*);

[0147] Among them, CI1(p n ) represents the spatial location p n Density index after surface calibration; w v (p n ) represents the weight of the vibration stage; w s (p n ) represents the surface stage weight; CI vib (p n ) represents the compaction index calculated during the vibration stage; ΔCI surface (p n ε* represents the surface correction amount; ε* is the numerical stability parameter, used to prevent the denominator from being zero, and its value is a very small positive number. The vibration assessment results and the surface calibration results are weighted averaged according to their respective confidence levels. When the confidence level of the vibration stage is high and the confidence level of the surface stage is low, the final result mainly depends on the vibration assessment; when the confidence level of the surface stage is high, the surface correction amount has a significant impact on the final result. This makes the stage with higher data quality contribute more to the assessment results.

[0148] According to another aspect of this application, using surface correction amounts for superposition correction can also result in:

[0149] When the surface confidence level Conf surface (p n The surface confidence threshold θ is lower than the preset threshold. s At that time, the calibrated density index CI1(p) is directly set. n () equals the compaction index CI during the vibration stage vib (p n No surface correction is performed to avoid introducing evaluation errors from low-quality smoothed data. The surface confidence threshold θ s The preferred value range is 0.2 to 0.4.

[0150] According to another aspect of this application, the method further includes calculating the fusion confidence score, specifically:

[0151] The fusion confidence score is calculated based on the confidence scores for the vibration stage and the surface layer. The fusion confidence score characterizes the overall reliability of the compaction assessment results after surface layer calibration. Its calculation formula is as follows:

[0152] Conffuse (p n ) = (w v (p n ) + w s (p n )) / (w v (p n ) + w s (p n ) + 1);

[0153] Conf fuse (p n ) represents the spatial location p n Fusion confidence at the location; w v (p n ) represents the weight of the vibration stage; w s (p n The weights are for the surface stage. The fused confidence score will be used as input to the subsequent crack posterior constraints, and will be used to update the confidence score when a crack event occurs. When the fused confidence score is high but a crack event occurs at the corresponding location, the system will reduce the final confidence score at that location and output a re-inspection warning.

[0154] In one embodiment of this application, the method further includes using monitoring data from the hardening phase for posterior constraints, specifically:

[0155] Strain monitoring data during the hardening stage of concrete is acquired using distributed optical fibers; spatiotemporal abrupt change analysis is performed on the strain monitoring data to identify crack events and their locations; the locations of crack events are compared with the calibrated evaluation results for consistency; when a crack event is determined to occur in a high-density spatial location, the confidence level of the calibrated evaluation results corresponding to that spatial location is reduced, and a re-inspection warning is output.

[0156] Specifically, once the concrete enters the hardening stage, the monitoring mode of the distributed optical fiber switches from high-frequency vibration monitoring to, or performs in parallel with, low-frequency static strain monitoring. The system collects strain data ε(s,t) distributed along the optical fiber in real time. When microcracks propagate within the concrete, they cause local strain abrupt changes or stress release. The spatiotemporal gradient of strain ▽ε is calculated, and when the gradient value exceeds a preset crack threshold ε... th When, it is marked as a crack event r k Record the location p where it occurs. k and the time of occurrence. When performing consistency ratios, check position p. k The density assessment results at the location CI final (p kIf a location is rated as Grade A (Excellent) or Grade B (Acceptable), but cracking occurs early in the hardening process, such as within 24 hours of pouring, it indicates that the initial compaction assessment may have been misjudged, for example, due to the influence of reinforcement shielding, or defects caused by non-vibration factors such as material shrinkage. In this case, the system performs a downgrade: forcibly lowering the assessment grade of the location, for example, from Grade A to Grade C, or reducing its confidence index Conf(p). k )=Conf(p k )*0.5, and immediately output a warning for suspected defect re-inspection to the terminal, prompting the inspection personnel to use an ultrasonic or rebound hammer for on-site verification.

[0157] As a supplementary implementation method, it is preferable to combine temperature monitoring data for comprehensive judgment. Distributed optical fibers can measure temperature simultaneously. A large amount of heat of hydration is released during the hardening process of concrete. The system monitors the temperature rise curve T(t) of each part in real time. If the peak temperature of a certain area is lower than the theoretical value, for example, more than 10 degrees Celsius lower than the expected peak temperature, it may indicate that the concrete filling is not compacted or that the aggregate segregates, possibly due to low cement paste content. If the internal temperature difference of a certain area exceeds the specification limit, such as 25 degrees Celsius, it indicates an extremely high risk of temperature cracking. Combining temperature anomaly indicators with crack events can further improve the accuracy of posterior constraints.

[0158] According to one aspect of this application, spatiotemporal abrupt change analysis is performed on strain monitoring data, including calculating strain difference values, specifically:

[0159] Strain monitoring data during the maintenance phase is acquired from a distributed fiber optic sensing system, denoted as ε(s). n , t m Where ε is the strain value, in units of microstrain; s n For fiber optic mileage location; t m For the monitoring time, calculate the strain difference between adjacent spatial locations, denoted as Δε(s). n , t m The calculation formula is as follows:

[0160] Δε(s n , t m ) = ε(s n+1 , t m ) - ε(s n , t m );

[0161] Where Δε(s) n , t m ) represents the position s n With position s n+1 Between time t m The strain difference fraction; ε(s) n+1 , tm ) represents the position s n+1 At time t m The strain value; ε(s) n , t m ) represents the position s n At time t m The strain value is calculated. When cracks exist inside the concrete, the strain on both sides of the crack exhibits a discontinuous distribution, manifested as an abrupt change in the strain difference value. In normal dense concrete, the strain changes continuously along the fiber direction, with a small and stable strain difference value. The cumulative time of calculating the strain difference value is denoted as Δε. cum (s n The calculation formula is as follows:

[0162] Δε cum (s n ) = ∑ m=mcure_start mcure_end |Δε(s n , t m )| *Δt;

[0163] Where Δε cum (s n ) represents the position s n Strain differential accumulation at the point; m cure_start Index for the start time of monitoring during the maintenance phase; m cure_end Index for the end time of monitoring during the maintenance phase; |Δε(s) n , t m )| represents the absolute value of the strain difference; Δt represents the time sampling interval.

[0164] According to one aspect of this application, identifying crack events and their locations specifically includes:

[0165] Set a threshold for crack detection. The strain differential threshold is denoted as θ. Δε It is used to determine whether there is a sudden change in strain in a single measurement; the cumulative threshold is denoted as θ. cum This is used to determine whether a crack development trend exists during long-term monitoring. Strain differential threshold θ Δε The preferred value range is 50 microstrain to 200 microstrain, with the specific value determined based on the concrete strength grade and the spatial resolution of the optical fiber. Cumulative threshold θ cum The optimal value is determined based on the monitoring duration and sampling frequency. Crack detection is performed for each spatial location. When location s n A crack event is determined to exist at a location if any of the following conditions are met: Condition 1 is a transient change condition, that is, there exists a certain time t. m The absolute value of the strain difference exceeds the strain difference threshold, expressed as the existence of t. m Make |Δε(s) n , tm )| Greater than θ Δε Condition two is the cumulative development condition, that is, the cumulative strain difference exceeds the cumulative threshold, expressed as Δε. cum (s n ) greater than θ cum The locations that satisfy any of the above conditions are marked as crack locations, forming a set of crack locations, denoted as S. crack The set of crack locations is used for subsequent confidence penalty calculations.

[0166] According to another aspect of this application, identifying crack events and their locations can also be achieved using a sliding window method for local anomaly detection, specifically as follows:

[0167] Set the half-width of the sliding window to W. s 1 sampling point. For each location s n Calculate the local mean and local standard deviation of the strain difference values ​​within a window centered at that location. The local mean is denoted as μ. local (s n The local standard deviation is denoted as σ. local (s n Calculate the standardized outlier score for this location, denoted as Z. crack (s n The calculation formula is as follows:

[0168] Z crack (s n ) = (|Δε(s n , t m )| - μ local (s n )) / (σ local (s n ) + ε*);

[0169] Z crack (s n ) represents the position s n Standardized anomaly score; μ local (s n ) represents the local mean; σ local (s n ) represents the local standard deviation; ε* is the numerical stability parameter. When the standardized outlier score exceeds the preset outlier threshold θ... Z At that time, a crack is determined to exist at that location. Anomaly threshold θ Z The preferred value range is 2.5 to 3.5, which corresponds to a tail probability of approximately 1% to 0.05% for a normal distribution.

[0170] In one possible implementation, the posterior constraint process is implemented using a confidence penalty mechanism, specifically as follows:

[0171] For the set of crack locations S crack Each position s in n Calculate the confidence penalty factor for that position. The confidence penalty factor is denoted as P. crack (s n The value ranges from 0 to 1. The severity index of the crack is defined as the ratio of the cumulative strain difference to the cumulative threshold, denoted as R. severity (s n The calculation formula is as follows:

[0172] R severity (s n ) =Δε cum (s n ) / θ cum ;

[0173] Where R severity (s n ) represents the position s n The severity index of cracks; Δε cum (s n ) represents the cumulative differential strain at that location; θ cum The cumulative threshold is used. A confidence penalty factor is calculated based on the crack severity index. The penalty factor uses an exponential decay function, and its calculation formula is as follows:

[0174] P crack (s n ) = exp(-λ p * (R severity (s n ) - 1));

[0175] Where P crack (s n ) represents the confidence penalty factor; exp represents the natural exponential function; λ p The penalty attenuation coefficient is used to control the penalty intensity, and its preferred value range is 0.5 to 2.0; R severity (s n Subtracting 1 indicates the portion of the crack severity exceeding a threshold. A penalty is applied to the fusion confidence score of the crack location to obtain the final confidence score with posterior constraints. The final confidence score is denoted as Conf. final (p n The calculation formula is as follows:

[0176] Conf final (p n ) = Conf fuse (p n ) * P crack (s n );

[0177] Conf final (p n ) represents position p n Final confidence level; Conf fuse (p n ) represents the fusion confidence level; P crack (s n ) represents the confidence penalty factor. For locations not in the crack location set, the confidence penalty factor is 1, meaning no penalty is applied, and the final confidence level equals the fused confidence level.

[0178] In a preferred implementation, the confidence penalty mechanism also includes a spatial diffusion penalty, specifically:

[0179] Even if no cracks are directly detected in the adjacent area where the crack is located, the reliability of the density assessment results should still be affected. The penalty diffusion radius is defined as D. spread The preferred value range is 0.5 meters to 2 meters. For non-crack locations within the penalty diffusion radius of the crack location, a decayed confidence penalty is applied. The decay penalty factor is denoted as P. spread (p n The calculation formula is as follows:

[0180] P spread (p n ) = 1 - (1 - P crack (s nearest )) * (1 - d nearest / D spread );

[0181] Where P spread (p n ) represents the diffusion penalty factor; P crack (s nearest ) represents the distance from position p n Penalty factor for the nearest crack location; d nearest For position p n Distance to the nearest crack location; D spread To penalize the diffusion radius. When position p n Distance d to the nearest crack location nearest Greater than or equal to the penalty diffusion radius D spread When the diffusion penalty factor is 1, no diffusion penalty is applied.

[0182] This embodiment utilizes data from the two subsequent stages inherent in concrete construction processes—smoothing and finishing, and static hardening—to perform secondary correction and logical verification on the evaluation results of the vibration stage. This solves the problem that single vibration monitoring is insufficient to accurately assess surface density and to eliminate interference from subsequent environmental factors.

[0183] As an optional implementation, the distributed optical fiber is arranged in a predetermined spatial grid pattern inside the concrete component; in other words, the distributed optical fiber is arranged in a multi-layered spatial grid pattern inside the concrete component.

[0184] The method further includes:

[0185] The arrival time of the same vibration event is extracted on optical fibers at different layers; the arrival time difference of the vibration wave to the optical fibers at different layers is calculated based on the arrival time; using the arrival time difference and the pre-stored spatial coordinates of the optical fibers, a set of spatial distance equations is constructed, and the three-dimensional spatial coordinates of the center of the vibration anomaly region are obtained by solving the equations.

[0186] In this embodiment, hardware deployment is a prerequisite for achieving three-dimensional positioning. The optical fibers are not only distributed in a grid pattern on the horizontal plane, but also arranged in layers vertically. For example, three optical fiber planes are defined: upper plane z=z1, middle plane z=z2, and lower plane z=z3. When a point P inside the concrete... a (x a y a , z a When an abnormal vibration occurs, such as a sudden drop in wave velocity or enhanced reflection due to a cavity, or when a point source is present, the vibration wave will propagate outwards in the form of a spherical wave. The system searches for the response signal of this abnormal event in the three layers of fiber data. Assume the peak response point detected in the upper fiber is P1(x1, y1, z1) with an arrival time of τ1; the response point detected in the middle fiber is P2(x2, y2, z2) with an arrival time of τ2; and the response point detected in the lower fiber is P3(x3, y3, z3) with an arrival time of τ3. Based on the relationship between distance and time, a system of nonlinear equations is constructed. For example, the specific form of the system of equations can be expressed as:

[0187] (x a -x i ) 2 +(y a -y i ) 2 +(z a -z i ) 2 =(v avg *(τ i -t0)) 2 ;

[0188] Where x a y a , z a x represents the three-dimensional coordinates of the center of the anomaly region to be solved; i y i , z iLet v be the coordinates of the measurement point on the i-th layer of optical fiber where an abnormal signal was detected, i = 1, 2, 3, corresponding to measurement points on the three layers of optical fiber respectively; avg τ is the average wave velocity of the concrete medium. i Let be the absolute time when the abnormal signal arrives at the i-th layer measurement point; t0 is the start time of the abnormal event, i.e., the unknown time offset. The solution is obtained by subtracting each pair of equations to eliminate t0. To eliminate the influence of t0, the arrival time difference algorithm is preferred. Subtracting each pair of the above equations constructs the hyperboloid equation system:

[0189] sqrt((x a -x i ) 2 +(y a -y i ) 2 +(z a -z i ) 2 )-sqrt((x a -x j ) 2 +(y a -y j ) 2 +(z a -z j ) 2 )=v avg *(τ i -τ j );

[0190] Where i, j ∈ {1, 2, 3} and i ≠ j. Since the system of equations is nonlinear, a numerical optimization algorithm is used to solve it. For example, the loss function J(x) is defined. a y a , z a = Σ(Theoretical distance difference - Measured distance difference) 2 The coordinates (x, y) that minimize the loss function are searched using the Newton-Raphson iterative method or the Levenberg-Marquardt algorithm. a y a , z a This coordinate is the three-dimensional geometric center of the anomaly region.

[0191] In some preferred embodiments, in addition to locating the center coordinates, a method for estimating the extent of the anomalous region is also provided. The anomalous region is approximated as a three-dimensional ellipsoid. Based on the length ΔL1 of the anomalous segment detected on the upper fiber, the length ΔL2 of the anomalous segment in the middle fiber, and the time delay difference between the responses of the upper and lower fibers, the lengths of the three semi-axes of the ellipsoid are estimated. For example, the horizontal semi-axis length a≈max(ΔL1, ΔL2) / 2, and the vertical semi-axis length c≈|z1-z3|*k scale , where k scale This is a scaling factor set according to the beam spread angle. By outputting the coordinates of the anomaly center and the estimated size of the area, it can provide construction personnel with an accurate three-dimensional defect model, guiding subsequent drilling grouting or excavation repair.

[0192] This embodiment overcomes the limitation that a single optical fiber can only perform one-dimensional mileage positioning. By utilizing the three-dimensional grid constructed in space by the optical fiber, it achieves accurate calculation of the three-dimensional spatial coordinates of the defect area, which is particularly suitable for the internal quality control of large-volume concrete structures such as dams, abutments, and base slabs.

[0193] In one possible embodiment, the intelligent detection method for concrete density based on distributed optical fiber can also be as follows: When light propagates in an optical fiber, due to the microscopic inhomogeneities in the optical properties such as density and refractive index of the fiber material, the incident light will scatter. When the probe light propagates inside the fiber, it may encounter strain, breaks, cracks, and vibrations, which will change the scattered light returning to the light source. By analyzing the image of the scattered light in the fiber, it is possible to monitor sudden situations. That is, when a disturbance occurs in the environment where the fiber is buried, the phase of the light will change accordingly, and this change will be reflected in a specific image. Based on the image, the disturbance and its specific location can be determined. Accordingly, before concrete pouring, the optical fiber is pre-laid inside or on the surface of the reinforcing cage according to the design path. The environment around the fiber (such as vibration, temperature, crack development, etc.) is continuously monitored during the concrete pouring and curing stages. Combining numerical simulation and empirical data, the collected data is deeply analyzed to find abnormal data characteristics, and machine learning is used to achieve intelligent judgment and early warning. Finally, the concrete density at any location where the optical fiber is laid is evaluated.

[0194] In specific implementation, the steel reinforcement layer is applied according to the following... Figure 5 and Figure 6The illustrated scheme involves laying distributed optical fibers, covering all parts including beams, slabs, and columns, with fiber spacing ≤2m. Monitoring begins as soon as concrete pouring starts. By collecting data on the vibration impact on the optical fibers, the vibration status at any location can be determined, providing a preliminary assessment of the concrete's compaction. Further assessment of the concrete's compaction status can be achieved by collecting vibration data from leveling and finishing machinery. Concrete generates significant heat of hydration during hardening; by collecting temperature change data and combining it with numerical simulation results and empirical data, a further evaluation of the concrete's compaction status and crack development can be made. In-depth analysis of the collected data identifies abnormal data characteristics, and the results are output to a program for learning, ultimately achieving intelligent and rapid identification of abnormal states, providing data support for on-site concrete quality control.

[0195] In summary, the intelligent concrete density detection method based on distributed optical fiber includes: establishing a spatial mapping relationship between optical fiber and structural components; collecting vibration response data throughout the construction process; identifying discrete vibration events through spatiotemporal analysis and calculating the near-field self-calibration excitation intensity to normalize the far-field response; inverting the propagation attenuation parameters, propagation velocity, and frequency domain characteristics to eliminate source differences; and calculating the density index. Based on this, surface calibration is performed using vibration data from the smoothing process, and posterior consistency constraints are applied by combining crack monitoring data from the hardening period.

[0196] This invention segments a continuous signal into independent vibration events and calculates the energy integral of each event in the near field of the source in real time as the excitation intensity calibration value for that operation. This calibration value is used to normalize the far-field response signal, mathematically eliminating the uncertainty of the source intensity and accurately extracting the propagation attenuation coefficient and wave velocity parameters that are only related to the density of the concrete medium, thus achieving a leap from qualitative waveform analysis to quantitative parameter measurement. A trowel is used as a secondary moving source to capture high-frequency response characteristics sensitive to surface defects, compensating for and correcting the evaluation results of deep optical fibers. Simultaneously, strain and crack monitoring data during the hardening period are introduced as posterior constraints; once cracks are detected in high-scoring areas, their confidence level is reduced in reverse. A closed-loop time axis of vibration inversion-trowel calibration-crack constraint is constructed, effectively solving the limitations of single-stage monitoring.

[0197] The preferred embodiments of the present invention have been described in detail above. However, the present invention is not limited to the specific details in the above embodiments. Within the scope of the technical concept of the present invention, various equivalent transformations can be made to the technical solutions of the present invention, and these equivalent transformations all fall within the protection scope of the present invention.

Claims

1. A method for intelligent detection of concrete density based on distributed optical fiber, characterized in that, include: Obtain the pre-established spatial mapping relationship between distributed fiber optic mileage coordinates and the spatial location of concrete components; By using distributed optical fibers laid in concrete components, vibration response data during concrete construction can be collected in real time. Based on the spatial mapping relationship, the vibration response data is analyzed to extract vibration characteristic quantities that reflect the physical characteristics of the vibration compaction operation; Based on vibration characteristic quantities, the concrete compaction state at the corresponding spatial location is evaluated, and the concrete compaction assessment results are output.

2. The method according to claim 1, characterized in that, Extract vibration characteristic quantities that reflect the physical properties of the vibration operation, including determining the near-field center location of the vibration event, specifically: Spatiotemporal energy analysis was performed on the vibration response data to identify a predetermined number of discrete, time-discontinuous vibration events. For each vibration event, the near-field center position corresponding to the vibration event is determined based on the peak value of the vibration energy distribution along the optical fiber.

3. The method according to claim 2, characterized in that, Extracting vibration characteristic quantities that reflect the physical properties of the vibration operation also includes calculating the excitation intensity calibration value of the vibration event, specifically: Using the near-field center position as a reference, a fiber optic segment within a preset spatial range is selected as the near-field reference segment; The vibration energy integral of the vibration event within the near-field reference segment is calculated to obtain the excitation intensity calibration value of the vibration event.

4. The method according to claim 3, characterized in that, Extracting vibration characteristic quantities that reflect the physical properties of the vibration operation also includes calculating the propagation attenuation parameters of vibration events, specifically: Calculate the far-field response energy of the vibration event at each far-field location along the optical fiber; Based on the ratio of the excitation intensity calibration value to the far-field response energy, the normalized attenuation is calculated; Based on the spatial mapping relationship, the propagation distance of the far-field position relative to the near-field center position is calculated; The normalized attenuation is standardized using the propagation distance to obtain the propagation attenuation parameters of the vibration event along the corresponding transmission path.

5. The method according to claim 2, characterized in that, Extracting vibration characteristic quantities that reflect the physical properties of vibration compaction operations also includes calculating the proportion of high-frequency energy in vibration events, specifically: Calculate the propagation speed of the vibration wave generated by the vibration event from the near-field center to various locations along the optical fiber; The vibration response data of the vibration event is transformed in the frequency domain, and the ratio of the preset high-frequency band energy to the full-frequency band energy is calculated to obtain the high-frequency energy proportion.

6. The method according to claim 4, characterized in that, Extracting vibration characteristic quantities that reflect the physical properties of vibration compaction operations also includes calculating the propagation speed of vibration events and the proportion of high-frequency energy. Assess the concrete compaction condition at the corresponding spatial location, including: Construct a multidimensional feature vector that includes propagation attenuation parameters, propagation speed, and the proportion of high-frequency energy; The multidimensional feature vector is input into a preset weighted fusion formula or a pre-trained machine learning regression model to calculate the concrete density index of spatial location. Among them, the concrete density index is negatively correlated with the propagation attenuation parameter, and positively correlated with the propagation speed and the proportion of high-frequency energy.

7. The method according to claim 1, characterized in that, This also includes surface calibration of concrete density assessment results using vibration data from the smoothing or finishing process, specifically: Extract vibration signals from the concrete smoothing or finishing stage from vibration response data; Calculate the surface response characteristics of the vibration signal at each spatial location during the smoothing or finishing stage, and generate the surface correction amount; The surface correction amount is used to superimpose and correct the concrete density assessment results to obtain the calibrated assessment results.

8. The method according to claim 7, characterized in that, It also includes using monitoring data from the hardening phase for posterior constraints, specifically: Strain monitoring data during the hardening stage of concrete is acquired using distributed optical fibers; spatiotemporal abrupt change analysis is performed on the strain monitoring data to identify crack events and their locations; The location of the crack event was compared with the calibrated assessment results for consistency. When a crack event is determined to occur in a space with high density, the confidence level of the calibrated assessment result corresponding to that space location is reduced, and a re-inspection warning is issued.

9. The method according to claim 1, characterized in that, The method further includes: Distributed optical fibers are arranged in a predetermined spatial grid pattern within the concrete structure; Extract the arrival time of the same vibration event on optical fibers at different layers; The time difference of arrival of vibration waves to optical fibers at different layers is calculated based on the time of arrival. By utilizing the time difference of arrival and the pre-stored spatial coordinates of the optical fiber, a set of spatial distance equations is constructed, and the three-dimensional spatial coordinates of the center of the vibration anomaly region are obtained by solving them.

10. The method according to claim 6, characterized in that, The pre-defined weighted fusion formula or pre-trained machine learning regression model is constructed in the following way: Concrete pouring tests were conducted in the pre-defined calibration area, and fiber optic vibration data and true density indices from core sampling or ultrasonic testing were collected simultaneously. The propagation attenuation parameters, propagation velocity, and high-frequency energy ratio corresponding to the fiber vibration data are extracted as input features. Using the truth density index as the target label, regression fitting or supervised training is performed on the input features to determine the weight coefficients of the weighted fusion formula or the network parameters of the machine learning regression model.

Images