Concrete full life cycle damage monitoring method and system

Abstract

The invention belongs to the field of concrete monitoring, and discloses a concrete full life cycle damage monitoring method and system, and the method comprises the steps: building a multi-damage classification recognition model through fusing the characteristics of multi-dimensional stress waves: signal energy, sound time, dominant frequency offset and waveform distortion rate; and on the basis of a support vector machine algorithm, intelligent discrimination of typical damages such as microcracks, voids and grouting defects is realized. By adopting the method, three technical effects are realized in a breakthrough manner: firstly, the synchronous identification capability of multi-type damage is remarkably improved through feature fusion and intelligent classification; 2, quantitative evaluation precision is ensured based on a differentiation algorithm of a damage physical mechanism; and 3, the three-dimensional dynamic display technology intuitively presents the damage space distribution characteristics, and provides a multi-dimensional decision basis for the full life cycle monitoring of the complex concrete structure.

Description

Technical Field

[0001] This invention belongs to the field of concrete monitoring technology, specifically relating to a method and system for monitoring damage throughout the entire life cycle of concrete. Background Technology

[0002] Concrete structures, with their high strength, durability, and cost-effectiveness, are widely used in major infrastructure projects such as buildings, bridges, and tunnels. However, during long-term service, concrete structures are prone to various types of damage due to factors such as load, environmental erosion, and material aging, including microcracks, voids, and incomplete grouting. If these damages are not detected and assessed accurately and in a timely manner, they may lead to structural performance deterioration or even safety accidents. Therefore, concrete damage detection technology is crucial for ensuring the safe and stable operation of infrastructure, and the research and optimization of related technologies has become a research hotspot in the field of civil engineering. Among them, piezoelectric ceramic sensors, due to their outstanding advantages such as rapid response, low cost, and ease of implantation into concrete, are increasingly widely used in concrete damage detection, providing effective technical support for structural health monitoring.

[0003] Despite the significant application potential of piezoelectric ceramic sensors in concrete damage detection, existing detection technologies based on these sensors still face several unresolved technical bottlenecks. Regarding sensor layout, current solutions often employ fixed single-point or linear array arrangements. This layout is ill-suited to the irregular damage distribution characteristics of complex concrete structures, resulting in significant differences in sensitivity to damage at different directions and scales, often failing to achieve comprehensive and balanced damage coverage. In terms of signal stability, piezoelectric signals are susceptible to interference from complex environmental factors. Dynamic changes in temperature and humidity within the concrete, and random disturbances in the structural stress field, can all cause piezoelectric signal drift, directly reducing the accuracy and reliability of damage detection. Regarding long-term service performance, sensors inevitably experience performance degradation during long-term use, and current technologies lack effective active compensation mechanisms to correct this performance degradation in real time, severely impacting the continuity and effectiveness of monitoring data. More importantly, in terms of damage identification range, existing technologies mostly focus on detecting single types of damage, failing to meet the practical engineering needs for simultaneous identification and accurate quantitative assessment of multiple types of damage, such as microcracks, voids, and incomplete grouting.

[0004] It is evident that existing concrete damage detection technologies are significantly inadequate in identifying multiple damages, making it difficult to meet the actual needs of accurate monitoring of damage in complex concrete structures. Summary of the Invention

[0005] This invention provides a method and system for monitoring damage throughout the entire life cycle of concrete. This method can effectively solve the problem that existing concrete damage detection technologies are significantly insufficient in terms of multi-damage identification capabilities, thereby meeting the actual needs for accurate monitoring of damage in complex concrete structures.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: A method for monitoring the damage of concrete throughout its entire life cycle, comprising: Acquire the stress wave signal of the concrete to be monitored; wherein the stress wave signal includes signal energy, first wave acoustic time, dominant frequency offset and waveform distortion rate; The stress wave signal is input into a pre-trained multi-damage classification and identification model, and the damage identification result is output. The base model of the multi-damage classification and identification model adopts a support vector machine classification model. The damage identification result includes no damage, microcracks, surface voids, and incomplete grouting. Differential algorithms are used to obtain the corresponding damage parameters based on the damage identification results. Specifically: the crack width is obtained by inversion based on the width-attenuation rate correlation curve that has been experimentally calibrated; the void area is calculated based on the interpolation algorithm, and the surface void depth is obtained by inversion through the correlation between the time difference, the difference in burial depth of the acquisition point, and the propagation speed, combined with the propagation speed of the stress wave in the concrete surface; the signal energy value of the grouting area is compared with the standard energy value when there is no damage, and the damage factor is calculated using the root mean square relative error method. Then, the quantitative evaluation is carried out by combining the correspondence between the experimentally calibrated damage factor and the grouting density to obtain the grouting density. The damage parameters and their corresponding location coordinates are associated with the three-dimensional model of the concrete structure to generate a three-dimensional image of the damaged area. The 3D images are labeled with different colors to output damage monitoring results.

[0007] Furthermore, in acquiring the stress wave signal of the concrete to be monitored, the specific method for acquiring the stress wave signal is as follows: The signal energy is obtained by accumulating the coefficient energy after wavelet packet decomposition. The first wave of sound time is determined and obtained by the signal rising edge threshold method; The main frequency offset is obtained by using the power spectral density of the signal obtained by fast Fourier transform, comparing the difference between the reference main frequency and the real-time main frequency under the undamaged state, and quantifying the frequency characteristic changes caused by damage. The waveform distortion rate is obtained by calculating the waveform similarity between the real-time signal and the reference signal.

[0008] Furthermore, in the process of inputting the stress wave signal into a pre-trained multi-damage classification and identification model and outputting damage identification results, the training samples of the multi-damage classification and identification model are derived from historical monitoring data under different damage scenarios, including signal features of three types of damage: microcracks, surface voids, and incomplete grouting, as well as the no-damage state. The input layer of the multi-damage classification and identification model consists of standardized values ​​of the signal features of microcracks, surface voids, incomplete grouting, and the no-damage state. The hidden layer uses a radial basis kernel function, and the output layer corresponds to the four types of identification results: no damage, microcracks, surface voids, and incomplete grouting.

[0009] Furthermore, The process of inverting the crack width based on a pre-calibrated width-attenuation rate correlation curve to obtain the crack width includes: The stress wave propagation time difference corresponding to three adjacent core sensing units in the crack area is obtained, and the distance from the crack center to each sensing unit is calculated by combining the propagation speed of the stress wave in the concrete. The coordinates of the crack location were determined using the triangulation method. The crack width is calculated using the signal amplitude attenuation rate of the two core sensing units, where the amplitude attenuation rate is the ratio of the signal amplitude under damaged conditions to the amplitude under undamaged conditions. The crack width is obtained by inverting the width-attenuation rate correlation curve, which has been pre-calibrated through experiments. The void area calculated based on the interpolation algorithm, combined with the propagation velocity of stress waves on the concrete surface, is used to inversely determine the surface void depth through the correlation between time difference, depth difference of the sampling point, and propagation velocity. This includes: Select edge sensing unit signals around the de-energized area, calculate the ratio of the signal energy of each edge sensing unit to the energy when there is no damage, and set the area with the energy ratio lower than the first preset threshold as the suspected de-energized area; An energy distribution cloud map of the vacancy region is generated using an interpolation algorithm. The region corresponding to the lowest energy value in the energy distribution cloud map is the vacancy center. The coverage area of ​​edge sensing units with a statistical energy ratio lower than the second preset threshold is estimated by combining the interpolation results to obtain the void area; Based on the first embedment depth h1 of the core sensing unit and the second embedment depth h2 of the edge sensing unit, and the time difference of the same stress wave signal, combined with the propagation speed v of the stress wave on the concrete surface... s '; The detachment depth is obtained by inverting the first burial depth, the second burial depth, the time difference, and the propagation velocity. The specific formula is as follows: Δt=|h1-h2| / v s '; The signal energy value of the grouting area is compared with the standard energy value when there is no damage. The damage factor is calculated using the root mean square relative error method. Then, the quantitative evaluation is performed by combining the correspondence between the damage factor calibrated by the experiment and the grouting density to obtain the grouting density. The specific formula is as follows:

[0010] In the formula, n is the number of samples. For E i The average value of E, where δ is the damage factor; i The signal energy value collected by the core sensing unit in the grouting area.

[0011] Furthermore, after acquiring the stress wave signal of the concrete to be monitored, the method further includes: The stress wave signal is corrected using a multi-field coupling compensation method, and the specific formula is as follows: ; In the formula, U corr To correct the amplitude of the piezoelectric signal, U raw The original signal amplitude collected by the sensing unit; For temperature difference; Humidity difference; Stress difference; , , These are the corresponding weighting coefficients.

[0012] Furthermore, after acquiring the stress wave signal of the concrete to be monitored, the method further includes: A dual approach of wavelet packet thresholding denoising and adaptive noise cancellation is employed to mitigate interference in stress wave signals. The specific steps of the wavelet packet thresholding denoising process are as follows: The stress wave signal is decomposed using wavelet packet decomposition to obtain multiple decomposed frequency bands; the specific formula is as follows: ; Where T is the wavelet packet tree; S i The stress wave signal acquired during the i-th sampling period; N is the number of decomposition layers, ranging from 3 to 5; wname is the wavelet function type. Calculate the wavelet coefficients of each decomposed frequency band, set an adaptive threshold according to the signal energy distribution, set the noise coefficients below the threshold to zero, retain the effective coefficients above the threshold and reconstruct the signal, and finally obtain the denoised signal energy through the wavelet packet energy calculation function. The adaptive noise cancellation process is as follows: The collected pure interference signal is used to generate a cancellation signal with the same amplitude but opposite phase as the interference signal through an adaptive filtering algorithm; The cancellation signal is superimposed on the acquired stress wave signal until the filtering algorithm converges, and the stress wave signal after adaptive noise cancellation is output.

[0013] A concrete life-cycle damage monitoring system, comprising: The signal acquisition module is used to acquire the stress wave signal of the concrete to be monitored; wherein, the stress wave signal includes signal energy, first wave acoustic time, dominant frequency offset and waveform distortion rate; The damage identification module is used to input stress wave signals into a pre-trained multi-damage classification identification model and output damage identification results. The base model of the multi-damage classification identification model adopts a support vector machine classification model. The damage identification results include no damage, microcracks, surface voids, and incomplete grouting. The parameter solving module is used to obtain the corresponding damage parameters based on the damage identification results using differentiated algorithms. Specifically: the crack width is obtained by inversion based on the width-attenuation rate correlation curve that has been experimentally calibrated; the void area is calculated based on the interpolation algorithm, and the surface void depth is obtained by inversion through the correlation between the time difference, the burial depth difference of the acquisition point, and the propagation speed, combined with the propagation speed of the stress wave in the concrete surface; the signal energy value of the grouting area is compared with the standard energy value when there is no damage, and the damage factor is calculated using the root mean square relative error method. Then, the quantitative evaluation is performed by combining the correspondence between the experimentally calibrated damage factor and the grouting density to obtain the grouting density. The image generation module is used to associate damage parameters and corresponding location coordinates with the three-dimensional model of the concrete structure to generate a three-dimensional image of the damaged area. The results output module is used to label the 3D images with different colors to output the damage monitoring results.

[0014] Furthermore, the parameter solving module includes: A crack width calculation unit, used to perform the inversion based on a pre-calibrated width-attenuation rate correlation curve to obtain the crack width, includes: The stress wave propagation time difference corresponding to three adjacent core sensing units in the crack area is obtained, and the distance from the crack center to each sensing unit is calculated by combining the propagation speed of the stress wave in the concrete. The coordinates of the crack location were determined using the triangulation method. The crack width is calculated using the signal amplitude attenuation rate of the two core sensing units, where the amplitude attenuation rate is the ratio of the signal amplitude under damaged conditions to the amplitude under undamaged conditions. The crack width is obtained by inverting the width-attenuation rate correlation curve, which has been pre-calibrated through experiments. The voiding parameter calculation unit is used to execute the calculation of the voiding area based on the interpolation algorithm, and, combined with the propagation velocity of stress waves on the concrete surface, to inversely obtain the surface voiding depth through the correlation between time difference, depth difference of the acquisition point, and propagation velocity, including: Select edge sensing unit signals around the de-energized area, calculate the ratio of the signal energy of each edge sensing unit to the energy when there is no damage, and set the area with the energy ratio lower than the first preset threshold as the suspected de-energized area; An energy distribution cloud map of the vacancy region is generated using an interpolation algorithm. The region corresponding to the lowest energy value in the energy distribution cloud map is the vacancy center. The coverage area of ​​edge sensing units with a statistical energy ratio lower than the second preset threshold is estimated by combining the interpolation results to obtain the void area; Based on the first embedment depth h1 of the core sensing unit and the second embedment depth h2 of the edge sensing unit, and the time difference of the same stress wave signal, combined with the propagation speed v of the stress wave on the concrete surface... s '; The detachment depth is obtained by inverting the first burial depth, the second burial depth, the time difference, and the propagation velocity. The specific formula is as follows: Δt=|h1-h2| / v s '; The grout density calculation unit is used to compare the signal energy value of the grouting area with the standard energy value when there is no damage, calculate the damage factor using the root mean square relative error method, and then perform a quantitative evaluation by combining the correspondence between the damage factor calibrated by the experiment and the grout density to obtain the grout density. The specific formula is as follows:

[0015] In the formula, n is the number of samples. For E i The average value of E, where δ is the damage factor; i The signal energy value collected by the core sensing unit in the grouting area.

[0016] Furthermore, the monitoring system also includes an adaptive signal conditioning and anti-interference module, used to correct the stress wave signal using a multi-field coupling compensation method and to perform anti-interference processing on the stress wave signal using a dual method of wavelet packet threshold denoising and adaptive noise cancellation; wherein: The stress wave signal is corrected using a multi-field coupling compensation method, and the specific formula is as follows: ; In the formula, U corr To correct the amplitude of the piezoelectric signal, U raw The original signal amplitude collected by the sensing unit; For temperature difference; Humidity difference; Stress difference; , , These are the corresponding weight coefficients; The specific steps of the wavelet packet thresholding denoising process are as follows: The stress wave signal is decomposed using wavelet packet decomposition to obtain multiple decomposed frequency bands; the specific formula is as follows: ; Where T is the wavelet packet tree; S i The stress wave signal acquired during the i-th sampling period; N is the number of decomposition layers, ranging from 3 to 5; wname is the wavelet function type. Calculate the wavelet coefficients of each decomposed frequency band, set an adaptive threshold according to the signal energy distribution, set the noise coefficients below the threshold to zero, retain the effective coefficients above the threshold and reconstruct the signal, and finally obtain the denoised signal energy through the wavelet packet energy calculation function. The adaptive noise cancellation process is as follows: The collected pure interference signal is used to generate a cancellation signal with the same amplitude but opposite phase as the interference signal through an adaptive filtering algorithm; The cancellation signal is superimposed on the acquired stress wave signal until the filtering algorithm converges, and the stress wave signal after adaptive noise cancellation is output.

[0017] Furthermore, the signal acquisition module includes a core sensing layer and an edge sensing layer; The core sensing layer includes several spherical piezoelectric ceramic array units, which are embedded in key stress areas of the concrete structure, including beam ends, column joints, and the periphery of prestressed ducts. The edge sensing layer is constructed using a flexible piezoelectric ceramic film, which is either attached to the surface of the concrete structure or embedded in the concrete protective layer in non-critical areas. The core sensing layer and the edge sensing layer achieve data interaction through wireless ad hoc networking technology. The network adopts a star-mesh hybrid topology. The sensing units of the core sensing layer act as master nodes, and the sensing units of the edge sensing layer act as slave nodes. The communication frequency between the master and slave nodes is adaptively adjusted according to the strength of the damaged signal. Each sensing unit is equipped with an independent identification code, which includes the unit type, deployment location coordinates, and installation time, so that the system can identify the location and function of each sensing unit through the identification code.

[0018] Compared with the prior art, the present invention has the following beneficial effects: This invention provides a method for monitoring damage throughout the entire life cycle of concrete. It constructs a multi-damage classification and identification model by integrating multi-dimensional stress wave characteristics: signal energy, acoustic time, dominant frequency shift, and waveform distortion rate. Based on a support vector machine algorithm, it achieves intelligent identification of typical damages such as microcracks, voids, and grouting defects. This method designs differentiated parameter inversion algorithms for different damage physical mechanisms. Specifically, it inverts the width using the energy attenuation characteristics of cracks, calculates the depth based on the propagation time difference of stress waves at the void interface, and assesses grout density through energy loss rate. Finally, it correlates quantitative damage parameters with spatial coordinates to generate a three-dimensional visualization map. This method achieves three major breakthroughs: first, it significantly improves the simultaneous identification capability of multiple damage types through feature fusion and intelligent classification; second, the differentiated algorithm based on the physical mechanism of damage ensures the accuracy of quantitative assessment; and third, the three-dimensional dynamic display technology intuitively presents the spatial distribution characteristics of damage, providing multi-dimensional decision-making basis for the life cycle monitoring of complex concrete structures.

[0019] Preferably, in this invention, the accuracy of signal characterization is significantly improved by optimizing the stress wave feature extraction method: wavelet packet energy accumulation enhances the feature capture capability of non-stationary signals, the rising edge thresholding method improves the anti-interference capability of acoustic time measurement, the dominant frequency offset quantization technology accurately reflects changes in material properties, and waveform distortion rate calculation effectively identifies structural anomalies. This multi-dimensional feature collaborative extraction mechanism lays a highly reliable data foundation for damage classification, fundamentally ensuring the accuracy of subsequent analysis.

[0020] Preferably, in this invention, a support vector machine model is trained based on historical damage scene data, and nonlinear features are processed through a radial basis function kernel to achieve robust classification of four damage states. The standardized input layer eliminates dimensional differences, and the kernel function hidden layer enhances the separability of the feature space, giving the model strong generalization ability. This design significantly improves the accuracy of identifying heterogeneous damage such as microcracks, voids, and grouting defects, and is particularly suitable for identifying complex and variable damage patterns in engineering sites.

[0021] Preferably, this invention employs specialized quantification algorithms for three types of damage characteristics: crack width inversion combined with triangulation and the physical laws of energy attenuation; void depth area calculation integrating energy cloud map interpolation and propagation time difference principles; and grout density assessment using energy statistical damage factor mapping. This differentiated algorithm system fully leverages the advantages of the physical response characteristics of various types of damage, achieving accurate spatial positioning and quantification of concealed damage, and effectively solving the problem of separating composite damage parameters using traditional methods.

[0022] Preferably, this invention introduces a temperature, humidity, and stress multi-field coupling compensation mechanism, which dynamically corrects signal drift caused by environmental interference through a physical model. This method can offset the superimposed effects of temperature changes, humidity penetration, and stress disturbances on piezoelectric signals in real time, significantly improving the stability and reliability of long-term monitoring data and providing continuous and effective raw data support for full life cycle assessment.

[0023] Preferably, this invention employs a dual anti-interference strategy of wavelet packet threshold denoising and adaptive noise cancellation: the former separates the effective signal from high-frequency noise through a frequency band energy threshold, while the latter uses the self-cancellation principle of interference signals to suppress environmental noise. This composite processing mechanism can maintain the integrity of signal features even under strong interference environments, fundamentally improving the extraction quality of damage-sensitive features and ensuring the accuracy of subsequent analysis.

[0024] This invention also provides a concrete full-lifecycle damage monitoring system, which achieves full-lifecycle monitoring of concrete damage through the collaborative work of five modules: signal acquisition, damage identification, parameter solving, image generation, and result output. The signal acquisition module captures multi-dimensional stress wave characteristics as the basis for analysis; the damage identification module uses a support vector machine model to intelligently classify heterogeneous damage such as microcracks, voids, and grouting defects; the parameter solving module designs differentiated inversion algorithms for different damage physical mechanisms (crack energy attenuation, void propagation time difference, and grouting energy loss) to achieve accurate quantification; and the image generation module fuses damage parameters with spatial coordinates to construct a three-dimensional model. Using this system, the following effects can be achieved: the modular architecture systematically solves the problem of simultaneous identification and quantitative assessment of multiple types of damage; the dedicated algorithm based on damage mechanisms significantly improves the positioning accuracy of hidden defects; and dynamic three-dimensional visualization technology enables an intuitive presentation of damage evolution, providing full-process intelligent decision support for the health diagnosis of complex concrete structures.

[0025] Preferably, in this invention, the crack width unit combines triangulation and energy attenuation principles to achieve accurate inversion of hidden cracks; the void parameter unit integrates energy cloud map interpolation and stress wave time difference principles to simultaneously obtain the void area and depth; and the grout density unit assesses internal defects based on energy statistical mapping relationships. This design fully leverages the physical response characteristics of various types of damage, achieving independent and accurate quantification of heterogeneous damage parameters, effectively solving the technical bottleneck of traditional methods in separating and assessing composite damage, and significantly improving the spatial positioning reliability of hidden defects.

[0026] Preferably, this invention includes a new adaptive signal conditioning and anti-interference module: a multi-field coupling compensation mechanism dynamically corrects signal drift caused by temperature, humidity, and stress disturbances through a physical model; wavelet packet threshold denoising and adaptive noise cancellation techniques respectively remove environmental interference from the frequency and time domains. The synergistic effect of these two technologies significantly enhances signal stability, completely resolving the impact of environmental noise and sensor performance degradation on data quality during long-term monitoring, and providing a continuous and reliable data foundation for full lifecycle assessment.

[0027] Preferably, this invention employs a core-edge layered sensing architecture: a spherical piezoelectric array is implanted in the critical stress area to capture deep damage signals, while a flexible thin film covers the surface and non-critical areas to achieve wide-area monitoring; wireless self-organizing network technology dynamically optimizes network transmission efficiency through a star-mesh hybrid topology and adaptive communication mechanism. This layout overcomes the spatial limitations of fixed arrays, achieving full coverage of complex structures and sensitive multi-scale damage detection, fundamentally solving the problem of uneven response of traditional sensor layouts to irregular damage. Attached Figure Description

[0028] Figure 1 A flowchart illustrating a method for monitoring the damage throughout the entire life cycle of concrete, as provided in an embodiment of the present invention; Figure 2 This is a structural schematic diagram of a concrete life-cycle damage monitoring system provided in an embodiment of the present invention. Detailed Implementation

[0029] To further understand the content of this invention, the invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments are merely illustrative and not limiting of the invention.

[0030] This embodiment provides a method for monitoring the damage throughout the entire life cycle of concrete. Specifically, it offers an intelligent piezoelectric ceramic monitoring technology with adaptive sensing capabilities, strong anti-interference properties, and the ability to simultaneously monitor multiple damages. This method overcomes the limitations of traditional technologies and can meet the practical needs of full life cycle health monitoring of concrete structures. This method is applicable to the full life cycle health monitoring of various concrete components, such as building structures, bridge engineering, and tunnel linings, and can achieve precise capture of the entire process from early microcrack initiation to macroscopic damage propagation.

[0031] like Figure 1 As shown in the figure, this embodiment provides a method for monitoring the damage of concrete throughout its entire life cycle. The specific steps are as follows: Acquire the stress wave signal of the concrete to be monitored; wherein the stress wave signal includes signal energy, first wave acoustic time, dominant frequency offset and waveform distortion rate; The stress wave signal is input into a pre-trained multi-damage classification and identification model, and the damage identification result is output. The base model of the multi-damage classification and identification model adopts a support vector machine classification model. The damage identification result includes no damage, microcracks, surface voids, and incomplete grouting. Differential algorithms are used to obtain the corresponding damage parameters based on the damage identification results. Specifically: the crack width is obtained by inversion based on the width-attenuation rate correlation curve that has been experimentally calibrated; the void area is calculated based on the interpolation algorithm, and the surface void depth is obtained by inversion through the correlation between the time difference, the difference in burial depth of the acquisition point, and the propagation speed, combined with the propagation speed of the stress wave in the concrete surface; the signal energy value of the grouting area is compared with the standard energy value when there is no damage, and the damage factor is calculated using the root mean square relative error method. Then, the quantitative evaluation is carried out by combining the correspondence between the experimentally calibrated damage factor and the grouting density to obtain the grouting density. The damage parameters and their corresponding location coordinates are associated with the three-dimensional model of the concrete structure to generate a three-dimensional image of the damaged area. The 3D images are labeled with different colors to output damage monitoring results.

[0032] The following is a further explanation of the methods for monitoring damage throughout the entire life cycle of concrete: The concrete life-cycle damage monitoring method provided in this embodiment relies on a multi-damage quantification identification and localization algorithm. This algorithm uses stress wave signals collected by an intelligent adaptive piezoelectric ceramic network. Through four core steps—feature parameter extraction, multi-damage classification and identification, damage parameter quantification calculation, and three-dimensional localization imaging—it achieves accurate identification and parameter characterization of three typical types of damage in concrete structures: internal microcracks, surface voids, and incomplete grouting. The algorithm runs in the data analysis module of the network master node and can be linked in real time with the correction signal output by the adaptive signal conditioning system to ensure the accuracy and timeliness of damage identification.

[0033] In this embodiment, the feature parameter extraction stage selects four key parameters as the basic features for damage identification for the stress wave signal after signal conditioning and anti-interference processing: signal energy, first wave acoustic time, dominant frequency offset, and waveform distortion rate. Signal energy is obtained by accumulating the coefficient energy after wavelet packet decomposition, using the same wavelet packet energy calculation logic as the edge sensing layer's gap detection to ensure the consistency of parameter calculation. The first wave acoustic time is determined by the signal rising edge threshold method, setting the starting time as when the signal amplitude reaches 10% of its peak value, and recording the time difference between this time and the signal transmission time to accurately capture the time characteristics of stress wave propagation. The dominant frequency offset is obtained by obtaining the signal's power spectral density through fast Fourier transform, comparing the difference between the reference dominant frequency and the real-time dominant frequency under undamaged conditions to quantify the frequency characteristic changes caused by damage. The waveform distortion rate is achieved by calculating the waveform similarity between the real-time signal and the reference signal, using a dynamic time warping algorithm to calculate the distance between them; the larger the distance, the more severe the waveform distortion. This parameter can effectively reflect the interference effect of the grouting non-compacted area on the waveform.

[0034] The multi-damage classification and identification system employs a support vector machine (SVM) algorithm to construct a classification model. The model training samples are derived from historical monitoring data of the core and edge sensing units under different damage scenarios, covering microcracks (width 0.01mm-0.5mm, length 1mm-50mm) and surface voids (area 0.1m²). 2 -1m 2 The algorithm identifies signal features for three types of damage: 2mm-50mm depth, incomplete grouting (60%-98% density), and no damage. The model input layer consists of standardized values ​​of these four feature parameters. The hidden layer uses a radial basis function kernel to map the feature space. The output layer corresponds to the four identification results (no damage, microcracks, surface voids, and incomplete grouting). Cross-validation is used to optimize model parameters, ensuring a stable classification accuracy above 95%. During identification, the algorithm receives feature parameters from each sensing unit in real time, processes them, inputs them into the model, and outputs damage type determination results. If the damage is determined to be microcracks or incomplete grouting, the algorithm automatically triggers the dense monitoring mode of the core sensing units; if it is determined to be surface voids, it triggers the local scanning mode of the edge sensing units, achieving dynamic matching between damage type and monitoring strategy.

[0035] Damage parameter quantification calculations employ differentiated algorithms for different damage types. Microcrack quantification is based on the signal transmission characteristics of the core sensing units. Three adjacent core sensing units (denoted as A, B, and C) within the crack region are selected, and the stress wave propagation time differences Δt1 and Δt2 from A to B and from A to C are obtained. This is combined with the stress wave propagation velocity v in the concrete. s(Using a reference signal calibrated under undamaged conditions, typically 3000m / s-4000m / s), the distance from the crack center to each sensing unit is calculated, and the crack position coordinates are determined using triangulation. The crack width is calculated using the signal amplitude attenuation rate from A to B. The amplitude attenuation rate is the ratio of the signal amplitude under damaged conditions to the amplitude under undamaged conditions. Combined with the width-attenuation rate correlation curve calibrated in advance through experiments, the crack width is inverted. A temperature compensation coefficient is introduced during the calculation process to correct the propagation speed deviation, ensuring that the width measurement accuracy is ≤0.01mm and the length measurement error is ≤5%.

[0036] Surface void quantification relies on continuous monitoring data from the edge sensing layer. Signals from edge sensing units surrounding the void area are selected, and the energy ratio of each unit's signal to that of the undamaged area is calculated. Areas with an energy ratio below 80% are designated as suspected void areas. An energy distribution cloud map of the void area is generated using an interpolation algorithm, and the area corresponding to the lowest energy value in the cloud map is the void center. The void area is estimated by statistically analyzing the coverage area of ​​edge sensing units with energy ratios below 80% and combining this with the interpolation results. The void depth is calculated using the signal time difference between the core sensing unit and the edge sensing unit, specifically by using the time difference Δt between the core sensing unit (depth h1) and the edge sensing unit (depth h2) receiving the same stress wave signal, combined with the stress wave propagation speed v on the concrete surface. s ', using the formula Δt=|h1-h2| / v s 'Invert the depth of the void, with a depth measurement error of ≤3%.'

[0037] The quantification of grout non-compaction was performed by calculating the damage factor using the root mean square relative error method, selecting the signal energy value E collected by the core sensing unit in the grouting area. i The damage factor δ is calculated by comparing it with the standard energy value E0 under no-damage conditions. The calculation logic is based on the quantitative approach for monitoring the compactness of prestressed concrete grouting, and the formula is as follows:

[0038] Where n is the number of samples, For E i The average value of δ is used to determine the grout density. The smaller the δ value, the higher the grout density. The relationship between δ and density is calibrated by test (e.g., δ=0.1 corresponds to 95% density, δ=0.5 corresponds to 75% density), so as to achieve a quantitative assessment of grout density with an assessment accuracy of ≤2%.

[0039] In this embodiment, damage localization imaging utilizes MATLAB software's 3D visualization tool. Damage parameters (location coordinates, crack width / void area / density) of each sensing unit are associated with the 3D model of the structure. A volume rendering algorithm generates a 3D image of the damaged area. Different damage types in the image are identified by differentiated colors (red for microcracks, blue for surface voids, and yellow for inadequate grouting). The degree of damage is distinguished by color intensity (the darker the color, the more severe the damage). The imaging results are updated in real time and uploaded to a remote terminal, allowing users to view damage details through zooming and rotation. Simultaneously, key parameters of the damaged area (such as maximum crack width, void area, and minimum density) are automatically labeled, providing intuitive data support for structural maintenance.

[0040] Explainable, the positioning algorithm also has an outlier removal function. When the signal parameters of a certain sensing unit exceed the normal range (such as the amplitude suddenly increasing by more than 10 times or the time difference exceeding the reasonable range), it is judged as abnormal data. The algorithm automatically uses the parameters of adjacent sensing units to complete the data through interpolation, thus avoiding positioning deviations caused by the failure of a single unit. At the same time, the algorithm updates the stress wave propagation speed and characteristic parameter benchmark values ​​periodically (every 24 hours) and corrects the calibration curve in combination with changes in ambient temperature and humidity to ensure the stability of long-term quantitative calculations.

[0041] For example, the above signal conditioning and anti-interference processing specifically includes: The stress wave signal is corrected using a multi-field coupling compensation method, and the specific formula is as follows: ; In the formula, U corr To correct the amplitude of the piezoelectric signal, U raw The original signal amplitude collected by the sensing unit; For temperature difference; Humidity difference; Stress difference; , , These are the corresponding weighting coefficients.

[0042] Furthermore, a dual approach of wavelet packet threshold denoising and adaptive noise cancellation is employed to perform anti-interference processing on the stress wave signal; the specific steps of the wavelet packet threshold denoising process are as follows: The stress wave signal is decomposed using wavelet packet decomposition to obtain multiple decomposed frequency bands; the specific formula is as follows: ; Where T is the wavelet packet tree; S i The stress wave signal acquired during the i-th sampling period; N is the number of decomposition layers, ranging from 3 to 5; wname is the wavelet function type. Calculate the wavelet coefficients of each decomposed frequency band, set an adaptive threshold according to the signal energy distribution, set the noise coefficients below the threshold to zero, retain the effective coefficients above the threshold and reconstruct the signal, and finally obtain the denoised signal energy through the wavelet packet energy calculation function. The adaptive noise cancellation process is as follows: The collected pure interference signal is used to generate a cancellation signal with the same amplitude but opposite phase as the interference signal through an adaptive filtering algorithm; The cancellation signal is superimposed on the acquired stress wave signal until the filtering algorithm converges, and the stress wave signal after adaptive noise cancellation is output.

[0043] like Figure 2 As shown, this embodiment also provides a concrete life-cycle damage monitoring system, including: a signal acquisition module for acquiring stress wave signals of the concrete to be monitored; wherein the stress wave signals include signal energy, first wave acoustic time, dominant frequency offset, and waveform distortion rate; a damage identification module for inputting the stress wave signals into a pre-trained multi-damage classification identification model and outputting damage identification results; the base model of the multi-damage classification identification model adopts a support vector machine classification model; the damage identification results include no damage, microcracks, surface voids, and incomplete grouting; and a parameter solving module for obtaining the corresponding damage parameters based on the damage identification results using a differentiated algorithm, wherein: based on a pre-calibrated width-attenuation rate correlation curve... The crack width is obtained by inversion of the line; the void area is calculated based on the interpolation algorithm, and the surface void depth is obtained by inversion through the correlation between the time difference, the difference in burial depth of the acquisition point, and the propagation speed, combined with the propagation speed of the stress wave in the concrete surface; the signal energy value of the grouting area is compared with the standard energy value when there is no damage, and the damage factor is calculated using the root mean square relative error method. Then, the quantitative evaluation is carried out by combining the correspondence between the damage factor calibrated by the experiment and the grouting density to obtain the grouting density; the image generation module is used to associate the damage parameters and the corresponding position coordinates with the three-dimensional model of the concrete structure to generate a three-dimensional image of the damaged area; the result output module is used to mark the three-dimensional image with different colors to output the damage monitoring results.

[0044] The signal acquisition module includes a core sensing layer and an edge sensing layer, which provides an intelligent adaptive piezoelectric ceramic network structure. The intelligent adaptive piezoelectric ceramic network adopts a "core-edge" two-layer distributed layout to adapt to the monitoring needs of different areas of the concrete structure. At the same time, through modular design and wireless self-organizing network technology, it can flexibly expand the sensing range and dynamically adjust the monitoring accuracy.

[0045] The core sensing layer consists of several spherical piezoelectric ceramic array units, primarily implanted in critical stress areas of the concrete structure, such as beam ends, column joints, and areas around prestressed ducts—locations prone to damage. Each core sensing unit has a diameter of 20mm-50mm and uniformly encapsulates 6-8 spherical piezoelectric ceramic elements. These elements are connected in a star configuration to a multi-hole connector at the center of the unit via shielded wires. The connector integrates a signal conditioning chip and a micro-storage module, capable of storing in real-time basic data such as the unit's factory-set sensitivity calibration parameters, temperature-voltage response curves, and humidity compensation coefficients. It also records performance degradation data during service, providing a basis for subsequent calibration. The core sensing unit is encased in a high-strength, low-shrinkage concrete matrix, fabricated using 3D printing technology. After molding, an epoxy resin waterproof layer is sprayed onto the surface, with a thickness controlled between 0.5mm and 1mm. This ensures both the interfacial bonding strength between the unit and the concrete structure and resists internal moisture erosion. The spacing of the core sensing units is set according to the structural size and monitoring accuracy requirements. The spacing of conventional components is 0.5m-1m, while the spacing of large components such as bridge box girders and tunnel linings can be extended to 1m-2m to ensure that there are no blind spots in the monitoring of key areas.

[0046] The edge sensing layer is constructed using a flexible piezoelectric ceramic film, with a thickness of 0.1mm-0.3mm and a width of 50mm-100mm. It can be cut to any shape according to the surface morphology of the concrete structure and applied to the surface or embedded in the concrete protective layer in non-critical areas. A signal acquisition node is placed every 100mm-200mm along the length of the flexible piezoelectric ceramic film. Each node contains a miniature piezoelectric sensor and a wireless transmission chip. The nodes are connected in series by flexible wires to form a continuous monitoring band. The edge sensing layer needs to cover more than 90% of the structural surface area. For curved structures such as bridge piers and cylindrical components, full coverage can be achieved by splicing multiple film segments, with conductive adhesive used to seal the joints to ensure continuous signal transmission. The edge sensing layer is mainly used to detect micro-cracks and surface voids on the structural surface. Although its monitoring sensitivity is lower than that of the core sensing layer, it can achieve large-area rapid scanning, complementing the core sensing layer.

[0047] The core sensing layer and the edge sensing layer interact via a wireless ad hoc network. The network employs a star-mesh hybrid topology, with the core sensing unit acting as the master node and the signal acquisition nodes of the edge sensing layer acting as slave nodes. The communication frequency between the master and slave nodes can adaptively adjust according to the strength of the damage signal. When there is no damage in the monitored area, the master node maintains a communication frequency of 1Hz-5Hz, while slave nodes intermittently upload data at a frequency of 1Hz-2Hz, reducing system power consumption. When a slave node acquires a stress wave signal suspected of causing damage, it automatically increases its communication frequency to 10Hz-20Hz and sends a warning signal to surrounding master nodes. Upon receiving the warning, the master nodes simultaneously increase their own communication frequency to 20Hz-50Hz, initiating a dense monitoring mode for surrounding slave nodes to achieve precise location of the damaged area. Network nodes have built-in power monitoring modules. When a node's remaining power is below 20%, the system automatically reduces the node's communication and data acquisition frequencies, prioritizing monitoring of the core area, and simultaneously sends a power warning to remote terminals, prompting maintenance personnel to replace the battery.

[0048] Each sensing unit is equipped with a unique identification code, which includes information such as unit type (core / edge), deployment location coordinates (X / Y / Z 3D coordinates), and installation time. The system can quickly identify the location and function of each unit through this code. When a unit malfunctions, the system can accurately mark the location of the faulty unit and automatically activate nearby backup units. The deployment distance between backup units and the faulty unit does not exceed 0.5m, ensuring the integrity of the monitoring network. Furthermore, the core sensing unit also has an active wake-up function, which can send commands via a remote terminal to wake up units in low-power mode for specialized testing, such as temporary monitoring after structural impact, further enhancing the system's flexibility.

[0049] For example, this monitoring system also introduces an adaptive signal conditioning and anti-interference module. The adaptive signal conditioning and anti-interference module is integrated into the signal processing module of the intelligent adaptive piezoelectric ceramic network. One end is connected to the signal output end of the core sensing unit and the edge sensing unit through a shielded wire, and the other end communicates with the data analysis module of the network master node. It can dynamically adjust the processing parameters according to the signal characteristics collected by different sensing units, while suppressing invalid signals generated by environmental interference and structural vibration, and ensuring the accurate extraction of damage characteristic signals.

[0050] The signal conditioning module employs a combination architecture of a variable gain amplifier and a programmable bandpass filter. The gain adjustment range of the variable gain amplifier is set to 1-1000 times, and its gain value is automatically calibrated by the feedback control circuit based on the amplitude of the input signal. When the amplitude of the stress wave signal acquired by the core sensing unit is lower than 0.1mV, the amplifier automatically increases the gain to 500-1000 times to ensure that weak damage signals are effectively captured. When the signal amplitude is higher than 100mV, the gain decreases to 1-10 times to avoid signal saturation distortion. Due to the wide monitoring range and relatively significant signal attenuation of the edge sensing unit, the amplifier gain is maintained at 100-300 times by default, and adjustments are only made when the signal amplitude fluctuates extremely. The center frequency of the programmable bandpass filter can be continuously switched within the range of 20kHz-2MHz. The switching logic is based on the type of sensing unit and the monitoring scenario: for microcracks inside concrete monitored by the core sensing unit, the center frequency of the filter is locked at 500kHz-1MHz, which corresponds to the high-frequency stress waves generated by the microcracks; for surface voids monitored by the edge sensing unit, the center frequency is adjusted to 20kHz-200kHz to adapt to the low-frequency characteristics of stress waves in the void area; at the same time, the passband width of the filter changes synchronously with the center frequency, with the high-frequency passband width set at 50kHz-100kHz and the low-frequency passband set at 10kHz-50kHz to ensure effective filtering of interference signals outside the frequency band.

[0051] The multi-field coupling compensation module works in conjunction with the temperature, humidity, and stress acquisition components built into the core sensing unit to acquire environmental parameters and structural stress states of the monitored area in real time. A preset compensation algorithm corrects the drift of the piezoelectric signal caused by environmental factors. During the compensation process, the real-time temperature T, humidity H, and stress σ acquired by the core sensing unit are first extracted and compared with parameters under a standard calibration environment. This yields the temperature difference ΔT, humidity difference ΔH, and stress difference Δσ. Then, the pre-stored compensation coefficients in the storage module are called, and the signal correction coefficients are calculated through linear superposition according to the following formula:

[0052] Among them, U corr To correct the amplitude of the piezoelectric signal, U raw The formula, representing the original signal amplitude collected by the sensing unit, quantifies the impact of environmental parameter deviations on the signal to achieve accurate correction of the original signal. The compensation logic updates the compensation parameters every 100ms to ensure that the monitoring data remains stable under scenarios such as concrete hydration heat release, sudden changes in environmental temperature and humidity, and short-term loads on the structure.

[0053] The anti-interference processing employs a dual technique of wavelet packet threshold denoising and adaptive noise cancellation. In the wavelet packet threshold denoising process, the acquired raw signal is first decomposed into 3-5 levels of wavelet packets. The db4 wavelet basis function is selected for decomposition, as it exhibits good temporal localization characteristics in high-frequency signal processing. The decomposition process is implemented according to the following formula:

[0054] Where T is the wavelet packet tree, S i The original voltage signal is acquired within the i-th sampling period, N is the number of decomposition levels (ranging from 3 to 5), and wname is the wavelet function type (set to 'db4' here). Then, the wavelet coefficients of each decomposed frequency band are calculated. An adaptive threshold is set according to the signal energy distribution. Noise coefficients below the threshold are set to zero, while effective coefficients above the threshold are retained and the signal is reconstructed. Finally, the denoised signal energy is obtained through the wavelet packet energy calculation function, as shown in the following formula:

[0055] Among them, E i Let N(t) be the energy value of the denoised voltage signal in the i-th sampling period. This process can eliminate random noise caused by electromagnetic radiation (such as 50Hz power frequency interference generated by surrounding electrical equipment), improving the signal-to-noise ratio to over 35dB. Adaptive noise cancellation technology is achieved by setting up a reference sensor. The reference sensor is of the same model as the monitoring sensor and is deployed at a location far from the monitoring area but with consistent environmental interference. After acquiring the pure interference signal N(t), an adaptive filtering algorithm generates a cancellation signal -N'(t) with the same amplitude but opposite phase as the interference signal. This cancellation signal is then superimposed on the mixed signal acquired by the monitoring sensor.

[0056] S(t) is the effective damage signal, and the specific superposition process is achieved by the following formula:

[0057] When the filtering algorithm converges, N(t)≈N'(t), and at this time Y(t)≈S(t), effectively canceling the interference of environmental vibration (such as low-frequency vibration generated by vehicle driving and mechanical operation). This technology can make the attenuation rate of vibration interference signal reach more than 80%.

[0058] The signal conditioning and anti-interference system also features self-diagnostic capabilities, enabling real-time monitoring of the amplifier, filter, and compensation module's operating status. When the amplifier output signal exhibits continuous saturation or no output, the system determines an amplifier fault and automatically switches to the backup amplifier channel. When the actual passband of the filter deviates from the set passband by more than 10%, the system triggers filter parameter recalibration. If the signal deviation after multi-field coupling compensation still exceeds 5%, the system indicates that the temperature, humidity, or stress acquisition components of the core sensing unit may have failed and require maintenance or replacement. Fault information is uploaded to the remote terminal in real-time via a wireless self-organizing network, and the system automatically activates a temporary signal processing scheme to ensure uninterrupted monitoring.

[0059] For example, the parameter solving module embeds a multi-damage quantification identification and localization algorithm, which can realize the use of differentiated algorithms to obtain the corresponding damage parameters based on the damage identification results in the above embodiments. Among them: the crack width is obtained by inversion based on the width-attenuation rate correlation curve that has been experimentally calibrated in advance; the void area is calculated based on the interpolation algorithm, and the surface void depth is obtained by inversion through the correlation between the time difference and the burial depth difference of the acquisition point and the propagation speed, combined with the propagation speed of the stress wave in the concrete surface; the signal energy value of the grouting area is compared with the standard energy value when there is no damage, and the damage factor is calculated using the root mean square relative error method. Then, the quantitative evaluation is performed by combining the correspondence between the experimentally calibrated damage factor and the grouting density to obtain the grouting density. The specific steps are not described here.

[0060] For example, this monitoring system also introduces an active compensation and self-repair mechanism. The active compensation and self-repair mechanism runs through the entire service life of the intelligent adaptive piezoelectric ceramic network. In response to the performance degradation, parameter drift and node failure that may occur in the core sensing unit and edge sensing unit during long-term operation, the system ensures the continuous stability and data reliability of the monitoring network through three core functions: regular calibration, dynamic sensitivity compensation and redundant replacement of failed nodes. Its operating logic is deeply adapted to the storage module built into the core sensing unit and the node communication protocol of the wireless self-organizing network. At the same time, based on the existing piezoelectric ceramic monitoring technology's approach to ensuring signal accuracy and network integrity, a closed-loop maintenance system is formed.

[0061] The periodic calibration function is automatically activated at a fixed cycle (every 30 days). The network master node sends a standard excitation signal to all sensing units. The amplitude, frequency, and waveform parameters of this signal are pre-calibrated to ensure the stability of the excitation signal. After receiving the standard excitation signal, the core sensing unit collects the feedback signal through its built-in signal conditioning chip. It compares the amplitude and frequency characteristics of the feedback signal with the standard response curve stored at the factory and calculates the sensitivity attenuation coefficient. If the amplitude of the feedback signal is lower than 90% of the standard value, it is determined that the unit has sensitivity attenuation, and the attenuation magnitude is automatically recorded and a calibration command is generated. Because the edge sensing unit focuses more on large-area scanning in its monitoring scenarios, the calibration threshold is set to 85% of the standard value to avoid frequent calibration triggers due to slight attenuation, thus balancing monitoring efficiency and energy consumption. During the calibration process, the master node simultaneously collects the temperature, humidity, and stress environment parameters of each unit. Combined with the correction logic of the multi-field coupling compensation module, it eliminates the interference of environmental factors on the calibration results, ensuring the accuracy of the attenuation coefficient calculation.

[0062] Dynamic sensitivity compensation addresses the performance degradation of the core sensing unit by employing a dual compensation method: adjusting the driving voltage and signal amplification factor. When the sensitivity degradation of the core sensing unit is between 10% and 30%, the system increases the unit's driving voltage via a piezoelectric controller. The voltage adjustment range is controlled between 0-100V, and the adjustment magnitude is calculated linearly based on the attenuation coefficient. Following the linear correlation between voltage and stress, a compensation relationship between the driving voltage and sensitivity is established to ensure that the unit's output signal amplitude recovers to over 95% of the standard value after the voltage increase. If the attenuation exceeds 30%, adjusting the voltage alone is insufficient to meet accuracy requirements. The system simultaneously increases the gain factor of the built-in amplifier, with the gain adjustment range consistent with the variable gain range of the adaptive signal conditioning module (1-1000 times). Through coordinated voltage and gain compensation, the unit's sensitivity is restored to over 80% of its initial state, meeting basic monitoring needs. Compensation parameters are stored in real-time in the unit's micro-storage module as reference data for the next calibration and are also uploaded to the master node for record-keeping, facilitating the tracking of unit performance change trends.

[0063] The redundancy replacement mechanism for failed sensing units is automatically triggered through real-time communication status monitoring of network nodes for completely failed sensing units (e.g., no signal output, feedback signal deviation exceeding 50%). Both core and edge sensing units are configured with backup nodes in a 1:1 ratio. The model and deployment location of the backup nodes are consistent with the main nodes—backup nodes for core sensing units are deployed within 0.3m-0.5m of the main nodes to ensure overlapping monitoring areas; backup nodes for edge sensing units are arranged at intervals along the monitoring strip, with a distance of no more than 0.2m from adjacent main nodes to ensure continuous monitoring range. When a main node is determined to have failed, it sends a failure signal to the corresponding backup node via a wireless ad hoc network. The main node immediately sends a wake-up command to the corresponding backup node. After startup, the backup node automatically reads the historical configuration parameters of the main node (such as communication frequency, signal acquisition cycle, and compensation coefficient) to quickly adapt to the monitoring scenario. The replacement time is controlled within 10 seconds to avoid monitoring interruption.

[0064] Furthermore, the proactive compensation and self-repair mechanism also features maintenance early warning functionality. The system periodically (every 7 days) analyzes the performance degradation trend of each sensing unit, predicts the degradation level of the unit over the next 30 days through linear fitting, and sends a maintenance warning to the remote terminal in advance if the predicted degradation exceeds 50%, prompting staff to replace the unit. Simultaneously, the system logs all compensation and replacement operations, including operation time, involved unit number, compensation parameters, and signal comparison data before and after replacement, facilitating later traceability and performance analysis. For replaced failed units, the system uses a fault diagnosis algorithm to preliminarily determine the cause of failure (such as sensor component damage, poor wire contact, or storage module failure), and writes the fault type into the log, providing data support for subsequent improvements to the sensing unit design.

[0065] In extreme environments (such as temperatures below -30℃ and humidity above 95%RH), the active compensation and self-repair mechanism automatically increases the calibration frequency (from 30 days to 15 days) and increases the wake-up priority of backup nodes to ensure the stability of the monitoring network under harsh conditions. When the external power supply of the monitoring system is interrupted and switched to battery power, the mechanism automatically reduces the calibration frequency and compensation operation frequency of non-critical area sensing units, prioritizes the operation of core area units, balances energy consumption and monitoring needs, and adapts to the complex and ever-changing service environment throughout the entire life cycle of concrete structures.

[0066] In this embodiment, the general flow of the monitoring system performing the monitoring process at different stages is as follows: Construction phase: During the concrete pouring process, core sensing units and edge sensing units are pre-embedded. The spacing between core units is set to 0.5m-2m according to the structural dimensions, and the coverage rate of edge units is not less than 90%. The wireless self-organizing network is built and initially calibrated. During the curing stage: the monitoring system operates in low-power mode, collects data on concrete hydration heat and humidity changes in real time, monitors the initiation of early shrinkage cracks, and issues timely warnings when abnormalities are detected. During service: The system collects monitoring data at an adaptive frequency, extracts damage characteristic parameters through signal conditioning, anti-interference processing and multi-field coupling compensation, realizes damage classification, quantification and localization, and generates health monitoring reports; Maintenance phase: Regularly calibrate and maintain the monitoring system, develop structural maintenance plans based on monitoring data, and promptly repair damaged areas.

[0067] Therefore, this embodiment provides a concrete full life cycle damage monitoring system, which has the following advantages: First, a three-dimensional distributed network layout combined with an adaptive sensing mechanism enables synchronous monitoring of multiple regions and damage types in concrete structures, covering the entire stage from microcracks to macroscopic damage. Second, multi-field coupling compensation and anti-interference technology effectively reduce the impact of environmental factors on monitoring data and improve the stability and reliability of long-term monitoring. Third, the quantitative identification algorithm can accurately calculate damage parameters and perform three-dimensional imaging, providing data support for structural maintenance; Fourth, the active compensation and self-repair mechanism extends the service life of the monitoring system, reduces later maintenance costs, and adapts to the full life cycle monitoring needs of concrete structures.

[0068] In summary, this invention provides a method and system for monitoring concrete damage throughout its entire life cycle. This method and system can be widely applied to various concrete components, including concrete beams, slabs, and columns in building construction; box girders and piers in bridge engineering; lining structures in tunnel engineering; and concrete dams in water conservancy projects. It is particularly suitable for critical infrastructure projects requiring high monitoring accuracy, complex service environments, and long-term health monitoring. Compared to existing concrete damage monitoring methods, this monitoring method has the following advantages: The above embodiments are merely one of the implementation methods for achieving the technical solution of the present invention. The scope of protection claimed by the present invention is not limited to this embodiment, but also includes any variations, substitutions and other implementation methods that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention.

[0069] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method of monitoring damage of a concrete throughout its life cycle, characterized by, The method comprises the following steps: acquiring a stress wave signal of the concrete to be monitored; wherein the stress wave signal comprises signal energy, first wave sound time, main frequency offset and waveform distortion rate; inputting the stress wave signal into a pre-trained multi-damage classification and recognition model to output a damage recognition result; the basic model of the multi-damage classification and recognition model adopts a support vector machine classification model; the damage recognition result comprises no damage, micro crack, surface void and grouting non-compactness; for the damage recognition result, a differentiated algorithm is used to obtain corresponding damage parameters, wherein: based on the width-attenuation rate correlation curve calibrated in advance through experiments, crack width is obtained by inversion; the void area is calculated based on an interpolation algorithm, and combined with the propagation speed of the stress wave in the concrete surface layer, the surface void depth is obtained by inversion through the correlation between the time difference and the depth difference of the collection point and the propagation speed; the signal energy value of the grouting area is compared with the standard energy value in the non-damage state, the damage factor is calculated by the root mean square relative error method, and then the corresponding relationship between the damage factor and the grouting compactness is quantitatively evaluated through the calibration of the experiment, so as to obtain the grouting compactness; the damage parameters and the corresponding position coordinates are associated with the three-dimensional model of the concrete structure to generate a three-dimensional image of the damage area; the three-dimensional image is identified by differential color to output the damage monitoring result.

2. The method of claim 1, wherein, In the step of acquiring the stress wave signal of the concrete to be monitored, the specific acquisition method of the stress wave signal is as follows: the signal energy is obtained based on the coefficient energy accumulation after wavelet packet decomposition; the first wave sound time is obtained by determining the rising edge threshold of the signal; the main frequency offset is obtained by comparing the difference between the reference main frequency and the real-time main frequency in the non-damage state to quantify the frequency characteristic change caused by damage; the waveform distortion rate is obtained based on the waveform similarity between the real-time signal and the reference signal.

3. The method of claim 1, wherein, In the step of inputting the stress wave signal into the pre-trained multi-damage classification and recognition model to output the damage recognition result, the training samples of the multi-damage classification and recognition model are derived from historical monitoring data under different damage scenarios, including signal features of three types of damage (micro crack, surface void and grouting non-compactness) and non-damage state; the input layer of the multi-damage classification and recognition model is the standardized value of the signal features of micro crack, surface void, grouting non-compactness and non-damage state, the hidden layer adopts a radial basis kernel function, and the output layer corresponds to four types of recognition results (non-damage, micro crack, surface void and grouting non-compactness).

4. The concrete full life cycle damage monitoring method according to claim 1, wherein: the step of obtaining crack width based on the width-attenuation rate correlation curve calibrated in advance through experiments comprises the following steps: acquiring the stress wave propagation time difference corresponding to the three adjacent core sensing units in the crack area, and calculating the distance from the crack center to each sensing unit in combination with the propagation speed of the stress wave in the concrete; determining the crack position coordinates by the triangular positioning method; calculating the crack width by the signal amplitude attenuation rate of the two core sensing units, wherein the amplitude attenuation rate is the ratio of the signal amplitude in the damage state to the amplitude in the non-damage state. Combined with the width-attenuation rate correlation curve calibrated in advance by experiments, the crack width is obtained by inversion; The area of the void is calculated based on the interpolation algorithm, and the surface void depth is obtained by inversion through the correlation between the time difference and the depth difference of the collection point and the propagation speed of the stress wave in the surface layer of the concrete, including: The edge sensing unit signals around the void area are selected, the signal energy of each edge sensing unit is calculated, and the energy ratio of the edge sensing unit is set to be lower than the first preset threshold value, and the area is regarded as the suspected void area; An energy distribution cloud map of the void area is generated by the interpolation algorithm, and the area corresponding to the lowest energy value in the energy distribution cloud map is the center of the void; The area of the void is calculated based on the interpolation algorithm, and the surface void depth is obtained by inversion through the correlation between the time difference and the depth difference of the collection point and the propagation speed of the stress wave in the surface layer of the concrete, including: Based on the first embedding depth h1 of the core perception unit and the second embedding depth h2 of the edge perception unit, the time difference of the same stress wave signal, combined with the propagation speed v of the stress wave in the concrete surface layer s '; Based on the first embedding depth, the second embedding depth, the time difference and the propagation velocity, the void depth is inverted, and the specific formula is as follows: Δt=|h1-h2| / v s ' The signal energy value of the grouting area is compared with the standard energy value when there is no damage, the damage factor is calculated by the root mean square relative error method, and the corresponding relationship between the damage factor and the grouting density is calibrated by experiments to quantitatively evaluate the grouting density, and the specific formula is as follows: In the formula, n is the number of sampling times, is the average value of E i , and δ is the damage factor; E i is the signal energy value collected by the core perception unit of the grouting area.

5. The method of claim 1, wherein the method further comprises: After obtaining the stress wave signal of the concrete to be monitored, the following steps are further included: The stress wave signal is corrected by using a multi-field coupling compensation method, and the specific formula is as follows: ； In the formula, U corr is the corrected piezoelectric signal amplitude, U raw is the original signal amplitude collected by the sensing unit; is the temperature difference; is the humidity difference; is the stress difference; , , are the corresponding weight coefficients, respectively.

6. The method of claim 1-5, wherein, After obtaining the stress wave signal of the concrete to be monitored, the following steps are further included: The stress wave signal is subjected to anti-interference processing by using a wavelet packet threshold denoising and an adaptive noise cancellation dual mode; wherein the specific steps of the wavelet packet threshold denoising process are as follows: The stress wave signal is subjected to wavelet packet decomposition to obtain a plurality of decomposition frequency bands; the specific formula is as follows: ； Wherein, T is a wavelet packet tree; S i is a stress wave signal collected in the i sampling period; N is the decomposition layer number, taking a value of 3-5, and wname is a wavelet function type. The wavelet coefficients of each decomposition frequency band are calculated, the adaptive threshold is set according to the signal energy distribution, the noise coefficients below the threshold are set to zero, the effective coefficients above the threshold are retained and the signal is reconstructed, and finally the signal energy after denoising is obtained through the wavelet packet energy calculation function; The adaptive noise cancellation process is as follows: The collected pure interference signal is subjected to adaptive filtering algorithm to generate a cancellation signal with equal amplitude and opposite phase of the interference signal; The cancellation signal and the collected stress wave signal are superimposed until the filtering algorithm converges, and the stress wave signal after adaptive noise cancellation is output.

7. A concrete life cycle damage monitoring system, characterized by, It includes: A signal acquisition module is configured to acquire a stress wave signal of a concrete to be monitored, wherein the stress wave signal includes signal energy, first wave sound time, main frequency offset, and waveform distortion rate; A damage identification module is configured to input the stress wave signal into a pre-trained multi-damage classification and identification model to output a damage identification result, wherein a basic model of the multi-damage classification and identification model is a support vector machine classification model, and the damage identification result includes no damage, micro-crack, surface void, and non-dense grouting. The parameter solving module is configured to solve the damage parameters corresponding to the damage identification result by using a differential algorithm, wherein: the crack width is obtained by inversion based on a width-attenuation rate correlation curve calibrated in advance by experiments; the void area is calculated based on an interpolation algorithm, and the surface void depth is obtained by inversion based on the correlation between the time difference and the depth difference of the collection points and the propagation speed of the stress wave in the concrete surface layer; the signal energy value of the grouting area is compared with the standard energy value in the undamaged state, the damage factor is calculated by using a root mean square relative error method, and the grouting density is quantitatively evaluated by combining the corresponding relationship between the damage factor calibrated in the experiment and the grouting density, so as to obtain the grouting density; The image generation module is configured to associate the damage parameters and the corresponding position coordinates with a three-dimensional model of the concrete structure, so as to generate a three-dimensional image of the damage area. The result output module is configured to identify the three-dimensional image by using different colors, so as to output the damage monitoring result.

8. The concrete full life cycle damage monitoring system of claim 7, wherein, The parameter solving module includes: The crack width calculation unit is configured to perform the inversion of the crack width based on the width-attenuation rate correlation curve calibrated in advance by experiments, and includes: The stress wave propagation time difference corresponding to the three adjacent core sensing units in the crack area is obtained, and the distance from the crack center to each sensing unit is calculated based on the propagation speed of the stress wave in the concrete. The crack position coordinates are determined by using a triangular positioning method. The crack width is calculated based on the signal amplitude attenuation rate of the two core sensing units, and the amplitude attenuation rate is the ratio of the signal amplitude in the damaged state to the amplitude in the undamaged state. The crack width is obtained by inversion based on the width-attenuation rate correlation curve calibrated in advance by experiments. The void parameter calculation unit is configured to calculate the void area based on the interpolation algorithm, and obtain the surface void depth by inversion based on the correlation between the time difference and the depth difference of the collection points and the propagation speed of the stress wave in the concrete surface layer, and includes: The edge sensing unit signals around the void area are selected, the energy ratio of each edge sensing unit signal to the energy in the undamaged state is calculated, and the region with an energy ratio lower than a first preset threshold is set as a suspected void region. An energy distribution cloud map of the void area is generated by using an interpolation algorithm, and the region corresponding to the lowest energy value in the energy distribution cloud map is the void center. The covered area of the edge sensing unit with a statistical energy ratio lower than a second preset threshold is estimated in combination with the interpolation result, so as to obtain the void area. Based on the first embedding depth h1 of the core perception unit and the second embedding depth h2 of the edge perception unit, the time difference of the same stress wave signal, combined with the propagation speed v of the stress wave in the concrete surface layer s ' Based on the first embedding depth, the second embedding depth, the time difference and the propagation velocity, the void depth is inverted, and the specific formula is as follows: Δt=|h1-h2| / v s ' The grouting density calculation unit is configured to compare the signal energy value of the grouting area with the standard energy value in the undamaged state, calculate the damage factor by using a root mean square relative error method, and quantitatively evaluate the grouting density by combining the corresponding relationship between the damage factor calibrated in the experiment and the grouting density, so as to obtain the grouting density, and the specific formula is as follows: In the formula, n is the number of sampling times, E is the average value of the signal energy value collected by the core perception unit of the grouting area, i δ is the damage factor; E i is the signal energy value collected by the core perception unit of the grouting area.

9. The concrete full life cycle damage monitoring system of claim 7, wherein, The monitoring system further includes an adaptive signal conditioning and anti-interference module configured to correct the stress wave signals by using a multi-field coupling compensation method and to perform anti-interference processing on the stress wave signals by using a wavelet packet threshold denoising and adaptive noise cancellation dual method; wherein: The stress wave signals are corrected by using the multi-field coupling compensation method, and the specific formula is as follows: ； In the formula, U corr is the corrected piezoelectric signal amplitude, U raw is the original signal amplitude collected by the sensing unit; is the temperature difference; is the humidity difference; is the stress difference; , , are the corresponding weight coefficients, respectively. The specific steps of the wavelet packet threshold denoising process are as follows: The stress wave signal is decomposed by wavelet packet to obtain a plurality of decomposition frequency bands; the specific formula is as follows: ； Wherein, T is a wavelet packet tree; S i is a stress wave signal collected in the i sampling period; N is the decomposition layer number, taking a value of 3-5, and wname is a wavelet function type. Wavelet coefficients of each decomposition frequency band are calculated, an adaptive threshold is set according to signal energy distribution, noise coefficients lower than the threshold are zeroed, effective coefficients higher than the threshold are reserved and a signal is reconstructed, and finally a signal energy after denoising is obtained through a wavelet packet energy calculation function; The adaptive noise cancellation process is as follows: The collected pure interference signal is generated into a cancellation signal with equal amplitude and opposite phase through an adaptive filtering algorithm; The cancellation signal is superimposed with the collected stress wave signal until the filtering algorithm converges, and the stress wave signal after adaptive noise cancellation is output.

10. The concrete full life cycle damage monitoring system of claim 7, wherein, The signal acquisition module comprises a core perception layer and an edge perception layer; The core perception layer comprises a plurality of spherical piezoelectric ceramic array units, which are implanted in key stress areas of the concrete structure, and the key stress areas of the concrete structure comprise beam ends, column joints and peripheries of prestressed pipes; The edge perception layer is constructed by using flexible piezoelectric ceramic film, and the flexible piezoelectric ceramic film is attached to the surface of the concrete structure or embedded in the concrete protective layer in the non-key area; The core perception layer and the edge perception layer realize data interaction through wireless ad hoc network technology, the network adopts a star-mesh hybrid topology structure, the perception units of the core perception layer act as master nodes, the perception units of the edge perception layer act as slave nodes, and the communication frequency between the master nodes and the slave nodes is adaptively adjusted according to the intensity of the damage signal; wherein each perception unit is equipped with an independent identity code, and the identity code comprises unit type, layout position coordinates and installation time, which are used for the system to identify the position and function of each perception unit through the identity code.

Images