A concrete structure crack damage evaluation system based on acoustic emission sensing

Abstract

The present application relates to the technical field of concrete detection, in particular to a concrete structure crack damage evaluation system based on acoustic emission sensing, which comprises a signal acquisition module, a feature discrete evolution module, a probability inference module, an energy gradient analysis module and a boundary definition module.In the present application, a multi-dimensional evolution set is constructed by extracting the amplitude and energy standard deviation of acoustic emission signals within a continuous time window, the synchronous change trend of the multi-parameter standard deviation is analyzed using a Bayesian probability model, the non-uniformly expanded posterior probability is calculated to lock the key signal set of the dominant damage expansion, the energy fluctuation coefficient is calculated in combination with the spatial coordinates of the sensor and a spatial sequence is generated, the stable interval of the fluctuation coefficient is identified according to the gradient attenuation law of energy with distance, thereby quantitatively defining the physical boundary of the dynamically active area of crack damage, and the non-uniform expansion state and active range of the concrete crack are accurately evaluated.

Description

Technical Field

[0001] This invention relates to the field of concrete testing technology, and in particular to a concrete structure crack damage assessment system based on acoustic emission sensing. Background Technology

[0002] The field of concrete testing technology refers to a collection of technologies related to the identification of the condition and performance assessment of concrete materials and structures during production, construction and service. It includes the detection, analysis and evaluation of the evolution of internal defects, cracks and damage in concrete and the structural safety status, and usually involves sensor placement, signal acquisition, physical quantity characterization, feature extraction and condition assessment.

[0003] Among them, the concrete structure crack damage assessment system refers to a system that uses acoustic emission sensors deployed on or inside concrete components to collect elastic wave signals generated by crack initiation, propagation, and damage development during the concrete loading process, and analyzes and judges the degree and distribution of crack damage based on parameters such as the amplitude, energy duration, and arrival sequence of acoustic emission events.

[0004] Existing assessment methods rely solely on basic statistical parameters such as the amplitude and energy of acoustic emission events to determine the degree of damage, ignoring the dynamic changes in the dispersion of signals during temporal evolution. This makes it difficult to capture the critical state of crack transition from uniform micro-damage to non-uniform propagation, resulting in the inability to accurately isolate the key signals that dominate crack propagation. Furthermore, the lack of energy gradient attenuation analysis for high-risk signals in the spatial dimension leads to blurred boundaries of dynamically active crack damage areas, making it impossible to construct quantitative evaluation indicators that reflect the trend of non-uniform crack propagation and reducing the accuracy of identifying fracture risk areas. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing technologies and propose a concrete structure crack damage assessment system based on acoustic emission sensing.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a concrete structure crack damage assessment system based on acoustic emission sensing includes:

[0007] The signal acquisition module acquires the elastic wave signal from the surface of the concrete structure and converts it into a digital waveform data sequence, then extracts the acoustic emission waveform data segment from the digital waveform data sequence.

[0008] The feature discrete evolution module extracts the amplitude, energy, and duration of the acoustic emission waveform data segment and maps them to a time window, calculates the mean and standard deviation of the data within the time window, and constructs a multi-parameter discrete evolution feature set.

[0009] The probability inference module sets a prior probability representing the possibility of non-uniform expansion in the region, and calculates the posterior probability of the concrete structure crack being in the non-uniform expansion state by combining the changing trend of the multi-parameter discrete evolution feature set, and marks the non-uniform expansion dominant signal set according to the posterior probability.

[0010] The energy gradient analysis module determines the corresponding sensor of the non-uniformly extended dominant signal set in the acoustic emission sensor array, obtains the spatial coordinates between the corresponding sensors and the positioning coordinates of the acoustic emission source, calculates the energy fluctuation coefficient, and generates an energy fluctuation coefficient sequence.

[0011] The boundary delineation module identifies the boundaries of the dynamic active zones characterizing crack damage in concrete structures based on the changes in energy fluctuation coefficients in the energy fluctuation coefficient sequence.

[0012] As a further aspect of the present invention, the acoustic emission waveform data segment is specifically a data sequence of a specified length extracted from the digital waveform data sequence after the triggering time; the multi-parameter discrete evolution feature set includes the amplitude standard deviation, energy standard deviation, and duration standard deviation calculated for each continuous time window; the non-uniform expansion dominant signal set specifically refers to the set of acoustic emission waveform data segments associated with time windows where the posterior probability is higher than the judgment benchmark probability; the energy fluctuation coefficient sequence includes the dimensionless energy fluctuation coefficient calculated for each spatial propagation segment; and the boundary of the dynamic active zone of concrete structure crack damage is specifically the starting distance value of the spatial propagation segment closest to the acoustic emission source location coordinates in the energy attenuation stable zone.

[0013] As a further aspect of the present invention, the signal acquisition module includes:

[0014] The elastic wave capture and conversion submodule acquires the acoustic emission sensor array arranged on the surface of the concrete structure, monitors the mechanical vibration waves generated during the stress process of the concrete structure in real time, collects the mechanical vibration waves as the initial signal, sets the sampling frequency and quantization bits to perform analog-to-digital conversion on the initial signal, arranges the converted values ​​in chronological order, and generates a digital waveform data sequence.

[0015] The trigger time scanning and identification submodule sets a background noise reference value to distinguish the attributes of the initial signal, scans the voltage amplitude in the digital waveform data sequence point by point according to the time step, compares it with the background noise reference value, locates the first time node where the voltage amplitude exceeds the background noise reference value, marks it as the starting position of the waveform, and obtains the digital waveform trigger time.

[0016] The waveform data truncation and storage submodule locates the digital waveform trigger time and subsequent time interval on the time axis of the digital waveform data sequence, sets the waveform truncation length parameter, truncates all digital points within the waveform truncation length parameter range after the trigger time, and generates acoustic emission waveform data segments.

[0017] As a further aspect of the present invention, the feature discretization evolution module includes:

[0018] The basic parameter parsing and extraction submodule parses the waveform structure of the acoustic emission waveform data segment, searches for the point with the maximum absolute voltage value as the amplitude data by traversing the waveform data points, performs envelope detection operation on the acoustic emission waveform data segment to construct the waveform envelope, performs time integration operation on the waveform envelope to obtain the area value as the energy data, detects the time difference between the first and last crossing of the voltage threshold of the waveform as the duration data, and obtains the waveform basic parameter set;

[0019] The window statistical mean calculation submodule, based on the generation timestamps of amplitude, energy and duration data in the waveform basic parameter set, maps the amplitude, energy and duration data to multiple consecutive time windows of preset length in a time sequence, extracts all amplitude, energy and duration data in each independent time window and calculates their respective average values ​​to obtain the window parameter average value information;

[0020] The discrete feature set construction submodule calculates the standard deviation between the amplitude, energy, and duration data within each independent time window and the corresponding average values ​​in the window parameter average information. The standard deviations of multiple dimensions are combined and arranged to generate a multi-parameter discrete evolution feature set.

[0021] As a further aspect of the present invention, the probability inference module includes:

[0022] The trend direction determination statistics submodule extracts the standard deviation between the current time window and the previous adjacent time window from the multi-parameter discrete evolution feature set, performs difference comparison on the standard deviation of the two adjacent windows, determines the increase or decrease of the values, and counts the number of parameters that show a synchronous increase or decrease trend among the three parameters of amplitude, energy and duration, as the direction of the discrete parameter change trend.

[0023] The posterior probability Bayesian operation submodule sets a prior probability value representing the possibility of non-uniform expansion of the region, constructs a likelihood function based on the direction of the trend of the discrete parameter change, and inputs the prior probability and the likelihood function into the Bayesian statistical inference model for product and normalization operations to obtain the non-uniform expansion posterior probability.

[0024] The non-uniform dominant signal marking submodule sets a baseline probability threshold for determining the confidence level of the state, compares the non-uniform spread posterior probability with the baseline probability threshold, filters time windows that exceed the baseline probability threshold, traces and locks the original waveform data associated with the time window, and marks it as a non-uniform spread dominant signal set.

[0025] As a further aspect of the present invention, the energy gradient analysis module includes:

[0026] The spatial distance mapping and sorting submodule identifies the response sensor number corresponding to the non-uniform extended dominant signal set and retrieves the corresponding sensor spatial coordinates from the acoustic emission sensor array. It calculates the straight-line distance between the sensor spatial coordinates and the acoustic emission source positioning coordinates as the spatial distance from the sensor to the source.

[0027] The energy space segmentation submodule acquires the energy data corresponding to the non-uniform spread dominant signal set, sorts the energy data in ascending order according to the spatial distance from the sensor to the source end, and cuts and distributes the sorted energy data into continuous space according to a preset distance interval step size to generate continuous spatial propagation segments.

[0028] The fluctuation coefficient sequence generation submodule calculates the average value of energy data within each continuous spatial propagation segment, retrieves the maximum and minimum values ​​of energy within the continuous spatial propagation segment and calculates the difference, calculates the energy fluctuation coefficient based on the ratio of the difference to the average value, and constructs the energy fluctuation coefficient sequence.

[0029] As a further aspect of the present invention, the boundary demarcation module includes:

[0030] The coefficient stability interval initial screening submodule scans the energy fluctuation coefficient sequence item by item along the distance increasing direction, checks whether there are three adjacent energy fluctuation coefficients in the sequence that are all lower than the preset stability discrimination threshold, records the position index of the corresponding energy fluctuation coefficient in the energy fluctuation coefficient sequence, and obtains the candidate stable region index;

[0031] The gradient decay condition verification submodule extracts the continuous spatial propagation segment corresponding to the candidate stable region index, calculates the average energy difference between the continuous spatial propagation segments, and compares it with a preset gradient threshold. If the average energy difference is less than the gradient threshold, it is confirmed that the energy decay has become gradual, and an energy decay stable region is generated.

[0032] The active zone boundary positioning submodule retrieves all continuous spatial propagation segments within the energy attenuation stable zone, identifies the spatial propagation segment closest to the acoustic emission source's positioning coordinates, extracts the starting distance of the corresponding spatial propagation segment, defines it as the boundary point between the dynamic active zone and the stable zone of crack damage, and generates the boundary of the dynamic active zone of crack damage in the concrete structure.

[0033] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0034] In this invention, a multidimensional evolution set is constructed by extracting the amplitude and energy standard deviation of acoustic emission signals within a continuous time window. The synchronous change trend of the standard deviation of multiple parameters is analyzed using a Bayesian probability model. The posterior probability of non-uniform propagation is calculated to lock the key signal set that dominates damage propagation. The energy fluctuation coefficient is calculated by combining the sensor spatial coordinates and a spatial sequence is generated. The stable interval of the fluctuation coefficient is identified based on the gradient decay law of energy with distance, thereby quantitatively defining the physical boundary of the dynamic active area of ​​crack damage and realizing the accurate assessment of the non-uniform propagation state and active range of concrete cracks. Attached Figure Description

[0035] Figure 1 This is a system flowchart of the present invention;

[0036] Figure 2 This is a flowchart of the signal acquisition module of the present invention;

[0037] Figure 3 This is a flowchart of the feature discrete evolution module of the present invention;

[0038] Figure 4 This is a flowchart of the probability inference module of the present invention;

[0039] Figure 5 This is a flowchart of the energy gradient analysis module of the present invention;

[0040] Figure 6 This is a flowchart of the boundary delineation module of the present invention. Detailed Implementation

[0041] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0042] Please see Figure 1 A concrete structure crack damage assessment system based on acoustic emission sensing includes:

[0043] The signal acquisition module acquires the elastic wave signal from the surface of the concrete structure and converts it into a digital waveform data sequence, then extracts the acoustic emission waveform data segment from the digital waveform data sequence.

[0044] The feature discrete evolution module extracts the amplitude, energy, and duration of acoustic emission waveform data segments and maps them to a time window. It calculates the mean and standard deviation of the data within the time window and constructs a multi-parameter discrete evolution feature set.

[0045] The probability inference module sets a prior probability representing the possibility of non-uniform expansion in a region, and calculates the posterior probability of the concrete structure cracks being in a non-uniform expansion state by combining the changing trend of the multi-parameter discrete evolution feature set. Based on the posterior probability, it marks the dominant signal set of non-uniform expansion.

[0046] The energy gradient analysis module identifies the corresponding sensors in the acoustic emission sensor array for the non-uniformly extended dominant signal set, obtains the spatial coordinates between the corresponding sensors and the positioning coordinates of the acoustic emission source, calculates the energy fluctuation coefficient, and generates an energy fluctuation coefficient sequence.

[0047] The boundary delineation module identifies the boundaries of the dynamic active zones characterizing crack damage in concrete structures based on the changes in energy fluctuation coefficients in the energy fluctuation coefficient sequence.

[0048] The acoustic emission waveform data segment is specifically a data sequence of a specified length extracted from the digital waveform data sequence after the triggering time. The multi-parameter discrete evolution feature set includes the amplitude standard deviation, energy standard deviation, and duration standard deviation calculated for each continuous time window. The non-uniform spread dominant signal set specifically refers to the set of acoustic emission waveform data segments associated with time windows where the posterior probability is higher than the judgment benchmark probability. The energy fluctuation coefficient sequence includes the dimensionless energy fluctuation coefficient calculated for each spatial propagation segment. The boundary of the dynamic active zone of concrete structure crack damage is specifically the starting distance value of the spatial propagation segment closest to the acoustic emission source location coordinates in the energy attenuation stable zone.

[0049] Please see Figure 2 The signal acquisition module includes:

[0050] The elastic wave capture and conversion submodule acquires the acoustic emission sensor array arranged on the surface of the concrete structure, monitors the mechanical vibration waves generated during the stress process of the concrete structure in real time, collects the mechanical vibration waves as the initial signal, sets the sampling frequency and quantization bits to perform analog-to-digital conversion on the initial signal, arranges the converted values ​​in chronological order, and generates a digital waveform data sequence.

[0051] Monitoring of a three-point bending load test on a C50 strength grade concrete beam was initiated. Broadband piezoelectric ceramic sensors deployed on the concrete surface detected the initiation and propagation of microcracks during the stress process in real time, converting the received mechanical vibration waves into analog voltage signals. After a preamplifier applied a 40 dB gain to the analog voltage signals, an analog-to-digital converter continuously discretized the signals at a sampling frequency of 2 MHz, generating 2,000,000 sampling points per second. With a quantization bit depth of 16 bits, the amplitude of the analog voltage signals was mapped to integer values ​​between -32768 and +32767. The converted digital values ​​were then arranged strictly in chronological order to generate a digital waveform data sequence containing timestamps and voltage amplitudes.

[0052] The trigger moment scanning and identification submodule sets a background noise reference value to distinguish the attributes of the initial signal, scans the voltage amplitude in the digital waveform data sequence point by point according to the time step, compares it with the background noise reference value, locates the first time node where the voltage amplitude exceeds the background noise reference value, marks it as the starting position of the waveform, and obtains the digital waveform trigger moment.

[0053] First, a background noise baseline value for distinguishing valid signals from noise is determined. This baseline value is based on statistical analysis of environmental noise data during the silent period before loading. Specifically, waveform data for a 1-second silent period is collected. This dataset contains 2,000,000 noise voltage points. The absolute average value of all voltage amplitudes in this dataset is calculated; for example, the calculated absolute average noise voltage is 0.005 volts. Considering the balance between signal recognition reliability and false trigger rate, the background noise baseline value is set to three times the absolute average value, i.e., Volts. After setting, scan point by point from the first data point of the digital waveform data sequence according to a time step of 0.5 microseconds (corresponding to the reciprocal of the 2 MHz sampling rate). Read the voltage amplitude at the current time step and compare it with the background noise reference value of 0.015 volts. If the current voltage amplitude is less than or equal to 0.015 volts, it is determined to be background noise, and the data of the next time step continues to be scanned; if the current voltage amplitude is greater than 0.015 volts, it is determined that an acoustic emission waveform has arrived at that moment. For example, when scanning to the 1050th sampling point, the voltage amplitude is read as 0.018 volts, because... The trigger time scanning and recognition submodule immediately locates the time node and marks it as the starting position of the waveform arrival, thereby obtaining the digital waveform trigger time.

[0054] The waveform data truncation and storage submodule locates the digital waveform trigger moment and subsequent time interval on the time axis of the digital waveform data sequence, sets the waveform truncation length parameter, truncates all digital points within the waveform truncation length parameter range after the trigger moment, and generates acoustic emission waveform data segments.

[0055] On the time axis of the digital waveform data sequence, a continuous time interval is locked forward from the trigger moment. The waveform truncation length parameter is set based on the typical attenuation period of the acoustic emission signal in the concrete medium and the reflection wave interference window. By analyzing the duration distribution of historical typical fracture signals, a duration covering 95% of the signal energy is selected, which is set to 2048 sampling points (corresponding to 1024 microseconds). Starting from the trigger moment, the submodule continuously reads the voltage values ​​of the subsequent 2048 digital points. If the number of remaining points at the end of the data sequence is less than 2048, zero values ​​are padded at the end to complete the length. Finally, these 2048 voltage values ​​arranged in time sequence are encapsulated into an independent acoustic emission waveform data segment.

[0056] Please see Figure 3 The feature discrete evolution module includes:

[0057] The basic parameter parsing and extraction submodule parses the waveform structure of the acoustic emission waveform data segment, searches for the point with the maximum absolute voltage value as the amplitude data by traversing the waveform data points, performs envelope detection operation on the acoustic emission waveform data segment to construct the waveform envelope, performs time integration operation on the waveform envelope to obtain the area value as the energy data, detects the time difference between the first and last crossing of the voltage threshold of the waveform as the duration data, and obtains the waveform basic parameter set;

[0058] Constructing the waveform envelope by performing envelope detection on the acoustic emission waveform data segment includes:

[0059] Obtain the time-domain discrete voltage numerical sequence composed of acoustic emission waveform data segments;

[0060] The Fast Fourier Transform algorithm is invoked to perform a mapping process from the time domain to the frequency domain on the discrete voltage numerical sequence in the time domain, resulting in a frequency domain signal sequence;

[0061] Identify positive and negative frequency components in a frequency domain signal sequence;

[0062] The values ​​of the negative frequency components are set to zero, and the adjusted one-sided spectrum sequence is calculated based on the positive frequency components and the preset weighting coefficient.

[0063] The inverse fast Fourier transform algorithm is called to perform an inverse transform on the one-sided spectral sequence, generating an analytic signal sequence in complex form;

[0064] For each sampling moment in the analytical signal sequence, the corresponding real part data value and imaginary part data value are separated;

[0065] Calculate the square of the real part based on the real part data value, and calculate the square of the imaginary part based on the imaginary part data value;

[0066] Calculate the sum of squares of the modulus based on the square values ​​of the real part and the imaginary part;

[0067] The instantaneous amplitude of the analytical signal sequence at the sampling time is calculated based on the sum of squared moduli.

[0068] Connect the instantaneous amplitude values ​​corresponding to all sampling times in chronological order to generate the waveform envelope;

[0069] First, iterate through all 2048 voltage values ​​in the data segment, compare their absolute values, and select the value with the largest absolute value as the amplitude data. For example, the maximum positive voltage in the data segment is 1.5 volts, and the minimum negative voltage is -1.8 volts. After comparison... Therefore, 1.8 volts was selected as the amplitude data. Envelope detection was then performed to construct the waveform envelope: the time-domain discrete voltage value sequence of the acoustic emission waveform data segment was obtained, and the sequence was mapped to the frequency domain using a Fast Fourier Transform (FFT). In the frequency domain signal sequence, the positive frequency components were retained and their amplitudes were multiplied by 2, while the amplitudes of the negative frequency components were set to zero, generating a one-sided spectrum sequence. An inverse Fast Fourier Transform was performed on the one-sided spectrum sequence to obtain a complex analytic signal sequence. For each sampling moment of the analytic signal sequence, the real and imaginary parts were extracted, the sum of the squares of the real and imaginary parts was calculated, and the square root of this sum was then performed to obtain the instantaneous amplitude value at that moment. All instantaneous amplitude values ​​were connected in chronological order to generate the waveform envelope. Time integration was performed on the waveform envelope, that is, the instantaneous amplitude value at each sampling point was multiplied by the sampling interval of 0.5 microseconds, and all products were summed to obtain the area value, for example, a result of 500 volts / microseconds, which was used as the energy data. To obtain duration data, a voltage threshold needs to be set. This threshold is set to 10% of the current waveform amplitude data. Volts. Retrieve the time point in the waveform data segment where the voltage first exceeds 0.18 volts and the time point where it last exceeds 0.18 volts, and calculate the difference between them. For example, a difference of 150 microseconds represents the duration data. This yields a set of basic waveform parameters containing an amplitude of 1.8 volts, an energy of 500 volts / microseconds, and a duration of 150 microseconds.

[0070] The window statistical mean calculation submodule, based on the generation timestamps of amplitude, energy and duration data in the waveform basic parameter set, maps the amplitude, energy and duration data to multiple consecutive time windows of preset length according to time sequence, extracts all amplitude, energy and duration data in each independent time window and calculates their respective average values ​​to obtain the window parameter average value information;

[0071] The entire monitoring process is divided into multiple consecutive time windows, each with a length of 1 second. Based on the generated timestamp, the waveform's basic parameter set is mapped to the corresponding time window. Taking the time window from the 10th to the 11th second as an example, a total of 50 acoustic emission signals are captured within this window. The submodule extracts the amplitude, energy, and duration data of these 50 signals respectively. The average amplitude is calculated by summing the 50 amplitude data and dividing by 50, assuming the result is 0.5 volts; the average energy is calculated by summing the 50 energy data and dividing by 50, assuming the result is 120 volts / microseconds; and the average duration is calculated by summing the 50 duration data and dividing by 50, assuming the result is 80 microseconds. These three values ​​(0.5 volts, 120 volts / microseconds, and 80 microseconds) constitute the average window parameter information for this window.

[0072] The discrete feature set construction submodule calculates the standard deviation between the amplitude, energy, and duration data in each independent time window and the corresponding average values ​​in the window parameter average information. The standard deviations of multiple dimensions are combined and arranged to generate a multi-parameter discrete evolution feature set.

[0073] Calculate the standard deviation of each parameter within each independent time window. A standard deviation calculation algorithm based on weighted Euclidean distance is introduced here. This algorithm can adaptively adjust the calculation weights according to signal intensity differences, thus more sensitively capturing the fluctuation characteristics of high-energy damage signals. Taking the amplitude parameter as an example, the standard deviation is set... The calculation formula is as follows: ,in, The standard deviation of a specific parameter (such as amplitude) within the current window; This represents the total number of signals captured within the current time window, in this example... ; For the first The specific parameter values ​​of the signal (e.g., the first) (amplitude of each signal) This is the average value of the parameter within this window; in this example, the average amplitude is 0.5 volts. For the first The weighting coefficient of each signal is set based on the ratio of the signal amplitude to the average amplitude, i.e. Since large-amplitude signals often correspond to the propagation of the main crack, a larger weight indicates a higher contribution to the overall standard deviation. Now, let's substitute specific values ​​into the calculation: assuming that for these 50 signals, the weighted squared differences are calculated one by one... And the sum after accumulation It is 0.725; at the same time, the sum of all weight coefficients is calculated. The value is 50.347. The calculation process is as follows: Thus, the amplitude standard deviation for the 10-second window is 0.12 volts. Similarly, the same algorithm is used to calculate the standard deviations for energy data (average 120 volts / µs) and duration data (average 80 µs), assuming an energy standard deviation of 35 volts / µs and a duration standard deviation of 18 µs, respectively. Finally, the three calculated standard deviations are combined to generate the multi-parameter discrete evolution feature set (0.12, 35, 18) for this time window.

[0074] Please see Figure 4 The probability inference module includes:

[0075] The trend direction determination statistics submodule extracts the standard deviation between the current time window and the previous adjacent time window from the multi-parameter discrete evolution feature set, performs difference comparison on the standard deviation of the two adjacent windows, determines the increase or decrease of the values, and counts the number of parameters that show a synchronous increase or decrease trend among the three parameters of amplitude, energy and duration, as the direction of the discrete parameter change trend.

[0076] Extract the multi-parameter discrete evolution feature sets of the current Nth time window (e.g., the 10th second window) and the preceding N-1th time window (the 9th second window). Assume the standard deviations of the 9th second window are: amplitude standard deviation 0.08 volts, energy standard deviation 25 volt-microseconds, and duration standard deviation 12 microseconds; the corresponding values ​​for the 10th second window are: amplitude standard deviation 0.12 volts, energy standard deviation 35 volt-microseconds, and duration standard deviation 18 microseconds. The submodule compares the differences item by item: amplitude standard deviation... The energy standard deviation is increasing. The standard deviation of duration is increasing; The values ​​show an increasing trend. Statistical results show that the amplitude, energy, and duration of all three parameters show an increasing trend, with a total of 3 parameters increasing synchronously. If the value of a parameter decreases or remains unchanged in the comparison results, it is not included in the synchronous increase count. According to the statistical results, there are four possibilities for the synchronous increase count: 0, 1, 2, and 3. In this example, the calculated result is 3, which is recorded as the direction of the discrete parameter change trend.

[0077] The posterior probability Bayesian operation submodule sets the prior probability value representing the possibility of non-uniform expansion of the region, constructs the likelihood function based on the direction of the discrete parameter change trend, and inputs the prior probability and the likelihood function into the Bayesian statistical inference model for product and normalization operations to obtain the non-uniform expansion posterior probability.

[0078] The posterior probability of non-uniform expansion is calculated based on the direction of the discrete parameter variation trend. Here, a Bayesian inference algorithm with dynamically weighted state intensity based on the synchronous change of multi-parameter standard deviations is introduced. This algorithm adjusts the weights of the likelihood function by quantifying the significance of the current trend change, thereby improving the sensitivity of the inference results to sudden non-uniform expansion. The calculation formula is as follows: ,in, In order to observe the current state of synchronous change in the standard deviation of multiple parameters Non-uniform expansion occurs under conditions where all three parameters increase simultaneously. The posterior probability; To determine the prior probability of non-uniform propagation, based on statistical results from life-cycle failure tests of similar concrete beams, the non-uniform propagation stage accounts for approximately 20% of the total stress process; therefore, we set... ; To observe the synchronous changes in the standard deviation of multiple parameters during non-uniform expansion. The likelihood probability is set to 0.9 based on the expert experience base; To observe the synchronous changes in the standard deviations of multiple parameters when non-uniform expansion (i.e., uniform damage or a stable period) has not occurred. The likelihood probability is set to 0.2; This is a weighting factor for the intensity of a state where the standard deviations of multiple parameters change synchronously. The calculation needs to be based on the relative increment of the standard deviation: calculate the relative increment of the amplitude. Relative increase in energy The relative increase in duration Calculate the average relative increment Set a reference incremental threshold. (This value is derived from the upper limit statistics of the normal loading fluctuation range). Set as Its physical meaning is that the greater the relative increment exceeds the reference threshold, the greater the weight of the state intensity of the synchronous change in the multi-parameter standard deviation. Calculation Substitute the above values ​​into the formula for calculation: First, calculate the weighted likelihood term: , Calculate the numerator: Calculate the denominator: The final posterior probability is obtained as follows: .

[0079] The non-uniform dominant signal marking submodule sets a baseline probability threshold for determining the confidence level of the state, compares the non-uniform spread posterior probability with the baseline probability threshold, filters time windows that exceed the baseline probability threshold, traces and locks the original waveform data associated with the time window, and marks it as a set of non-uniform spread dominant signals.

[0080] The calculated posterior probability is then used for assessment. A baseline probability threshold for determining the confidence level of the state is set. This threshold is based on ROC curve (Receiver Operating Characteristic) analysis, selecting the probability value corresponding to the maximum Youden index as the optimal classification threshold, typically set to 0.5. The calculated posterior probability of 0.974 is then compared with the baseline probability threshold of 0.5. Since... The module determines that the current 10-second time window is in a non-uniform spread dominant state. The submodule then traces all 50 original acoustic emission waveforms contained within the 10-second window and marks these waveforms and their parameters as the non-uniform spread dominant signal set.

[0081] Please see Figure 5 The energy gradient analysis module includes:

[0082] The spatial distance mapping and sorting submodule identifies the response sensor number corresponding to the non-uniformly extended dominant signal set and retrieves the corresponding sensor spatial coordinates from the acoustic emission sensor array. It calculates the straight-line distance between the sensor spatial coordinates and the acoustic emission source positioning coordinates as the spatial distance from the sensor to the source.

[0083] Calculating the straight-line distance between the sensor's spatial coordinates and the acoustic emission source's location coordinates includes:

[0084] The horizontal, vertical, and longitudinal coordinate components of the sensor spatial coordinates and the acoustic emission source positioning coordinates are obtained in a three-dimensional rectangular coordinate system.

[0085] Calculate the squared values ​​of the corresponding coordinate differences based on the horizontal coordinate components, vertical coordinate components, and vertical coordinate components respectively;

[0086] Calculate the sum of the squares of the spatial distances based on the squared values ​​of the coordinate differences corresponding to the three dimensions, and then perform a square root operation on the sum of the squared spatial distances to obtain the straight-line distances.

[0087] Identify the sensor number corresponding to a specific signal in the signal set, for example, sensor S1. Retrieve the three-dimensional spatial coordinates of S1 from the sensor layout plan; the coordinate values ​​are... (Unit: mm). The Geiger localization algorithm was used to calculate the coordinates of the acoustic emission source corresponding to this signal. The calculation result is as follows: (Unit: mm). Calculate the difference between the three coordinate components: horizontal. millimeters, longitudinal millimeters, vertical Millimeters. Calculate the square of the difference between each component: Calculate the sum of the squares of the coordinate differences: Perform the arithmetic square root operation on the sum: Millimeters. This value of 86.6 millimeters represents the spatial distance between the sensor and the source of this signal. The same calculation is performed on all 50 signals in the signal set to obtain a set of distance data.

[0088] The energy space segmentation submodule acquires the energy data corresponding to the non-uniform spread dominant signal set, sorts the energy data in ascending order according to the spatial distance from the sensor to the source, and cuts and distributes the sorted energy data into continuous space according to the preset distance interval step size to generate continuous spatial propagation segments.

[0089] Acquire the energy data corresponding to each signal in the signal set (e.g., the aforementioned 500 volts / microseconds). Sort the energy data in ascending order based on the calculated spatial distance from the sensor to the source, i.e., the closer the distance, the higher the ranking. Set the distance interval step size to 50 millimeters. This step size is based on the wavelength of elastic waves in concrete (approximately 50 to 100 millimeters), taking the lower limit to ensure spatial resolution. Divide the sorted data into consecutive spatial propagation segments: 0 to 50 millimeters is segment 1, 50 to 100 millimeters is segment 2, 100 to 150 millimeters is segment 3, and so on. Each segment contains several energy data points falling within that distance range.

[0090] The fluctuation coefficient sequence generation submodule calculates the average value of energy data within each continuous spatial propagation segment, retrieves the maximum and minimum values ​​of energy within the continuous spatial propagation segment and calculates the difference, calculates the energy fluctuation coefficient based on the ratio of the difference to the average value, and constructs the energy fluctuation coefficient sequence.

[0091] Calculate the energy fluctuation coefficient for each continuous spatial propagation segment. Taking segment 2 (50 to 100 mm) as an example, assume this segment contains 5 energy data points: 400, 420, 450, 410, and 430 (unit: volts / microseconds). First, calculate the average value: The maximum value is 450 and the minimum value is 400. Calculate the difference: Calculate the energy fluctuation coefficient: divide the difference by the average value, i.e. This calculation is performed sequentially on all segments. Assume the energy fluctuation coefficients calculated for segments 1 to 5 are 0.15, 0.118, 0.09, 0.05, and 0.04, respectively. This set of spatially ordered values ​​(0.15, 0.118, 0.09, 0.05, 0.04) constitutes the energy fluctuation coefficient sequence.

[0092] Please see Figure 6 The boundary delineation module includes:

[0093] The coefficient stability interval initial screening module scans the energy fluctuation coefficient sequence item by item along the distance increasing direction, checks whether there are three adjacent energy fluctuation coefficients in the sequence that are all lower than the preset stability discrimination threshold, records the position index of the corresponding energy fluctuation coefficient in the energy fluctuation coefficient sequence, and obtains the candidate stable region index;

[0094] A preset stability threshold of 0.1 is set. This threshold is based on statistical analysis of acoustic wave transmission experiments on intact, undamaged concrete specimens, where the coefficient of variation of acoustic wave energy fluctuations in a homogeneous medium is typically below 0.1. The submodule scans the sequence (0.15, 0.118, 0.09, 0.05, 0.04) sequentially along the direction of increasing distance. It checks for any three consecutive values ​​below 0.1. The scanning process is as follows: the first value... The condition is not met; the second value. The condition is not met; the third value. The fourth value is satisfied. The 5th value is satisfied. The condition is met. It was found that starting from the third value, three consecutive values ​​(0.09, 0.05, 0.04) are all below the preset stability threshold. The position indices corresponding to these three values ​​are recorded, i.e., segments 3, 4, and 5. These segments are determined to be candidate stable regions, and the starting index of the candidate stable region is recorded as segment 3.

[0095] The gradient decay condition verification submodule extracts the continuous spatial propagation segment corresponding to the candidate stable region index, calculates the average energy difference between the continuous spatial propagation segments, and compares it with the preset gradient threshold. If the average energy difference is less than the gradient threshold, it is confirmed that the energy decay has become gradual, and an energy decay stable region is generated.

[0096] The candidate stable regions are then re-confirmed, and the average energy data of the continuous spatial propagation segments corresponding to the candidate stable region indices are extracted. Assume the average energy of segment 3 is 300 volts / microseconds and the average energy of segment 4 is 290 volts / microseconds. The average energy difference between adjacent segments is calculated: Volts / microseconds. The gradient threshold is set to 20 volts / microseconds, based on the range of random energy fluctuations caused by background noise, taken as twice the standard deviation of the background noise energy. The average energy difference of 10 is compared with the gradient threshold of 20. Because... The energy decay was determined to have leveled off, with no abnormal abrupt changes. The verification was successful, confirming that the third segment and its subsequent regions are stable energy decay zones.

[0097] The active zone boundary positioning submodule retrieves all continuous spatial propagation segments within the energy attenuation stable zone, identifies the spatial propagation segment closest to the acoustic emission source's positioning coordinates, extracts the starting distance of the corresponding spatial propagation segment, defines it as the boundary point between the dynamic active zone and the stable zone of crack damage, and generates the boundary of the dynamic active zone of crack damage in the concrete structure.

[0098] The confirmed stable energy attenuation regions (segment 3 and subsequent segments) were retrieved, and the segment closest to the acoustic emission source's location coordinates was identified as segment 3. The distance range corresponding to segment 3 is 100 mm to 150 mm (based on the aforementioned 50 mm step division: segment 1 0-50, segment 2 50-100, segment 3 100-150). The starting distance value of 100 mm for this segment was extracted and defined as the boundary between the dynamically active and stable regions of crack damage. This result indicates that the region within a 100 mm radius of the acoustic emission source is the dynamically active region of crack damage, characterized by severe energy fluctuations (fluctuation coefficient greater than 0.1) and non-uniform damage propagation; while the region beyond 100 mm is the stable region. Finally, 100 mm is output as the boundary of the dynamically active region of crack damage in the concrete structure.

[0099] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A concrete structure crack damage assessment system based on acoustic emission sensing, characterized in that, The system includes: The signal acquisition module acquires the elastic wave signal from the surface of the concrete structure and converts it into a digital waveform data sequence, then extracts the acoustic emission waveform data segment from the digital waveform data sequence. The feature discrete evolution module extracts the amplitude, energy, and duration of the acoustic emission waveform data segment and maps them to a time window, calculates the mean and standard deviation of the data within the time window, and constructs a multi-parameter discrete evolution feature set. The probability inference module sets a prior probability representing the possibility of non-uniform expansion in the region, and calculates the posterior probability of the concrete structure crack being in the non-uniform expansion state by combining the changing trend of the multi-parameter discrete evolution feature set, and marks the non-uniform expansion dominant signal set according to the posterior probability. The energy gradient analysis module determines the corresponding sensor of the non-uniformly extended dominant signal set in the acoustic emission sensor array, obtains the spatial coordinates between the corresponding sensors and the positioning coordinates of the acoustic emission source, calculates the energy fluctuation coefficient, and generates an energy fluctuation coefficient sequence. The boundary delineation module identifies the boundary of the dynamic active zone characterizing crack damage in the concrete structure based on the changes in the energy fluctuation coefficient in the energy fluctuation coefficient sequence. The probability inference module includes: The trend direction determination statistics submodule extracts the standard deviation between the current time window and the previous adjacent time window from the multi-parameter discrete evolution feature set, performs difference comparison on the standard deviation of the two adjacent windows, determines the increase or decrease of the values, and counts the number of parameters that show a synchronous increase or decrease trend among the three parameters of amplitude, energy and duration, as the direction of the discrete parameter change trend. The posterior probability Bayesian operation submodule sets a prior probability value representing the possibility of non-uniform expansion of the region, constructs a likelihood function based on the direction of the trend of the discrete parameter change, and inputs the prior probability and the likelihood function into the Bayesian statistical inference model for product and normalization operations to obtain the non-uniform expansion posterior probability. The non-uniform dominant signal marking submodule sets a baseline probability threshold for determining the confidence level of the state, compares the non-uniform spread posterior probability with the baseline probability threshold, filters time windows that exceed the baseline probability threshold, traces and locks the original waveform data associated with the time window, and marks it as a non-uniform spread dominant signal set. The energy gradient analysis module includes: The spatial distance mapping and sorting submodule identifies the response sensor number corresponding to the non-uniform extended dominant signal set and retrieves the corresponding sensor spatial coordinates from the acoustic emission sensor array. It calculates the straight-line distance between the sensor spatial coordinates and the acoustic emission source positioning coordinates as the spatial distance from the sensor to the source. The energy space segmentation submodule acquires the energy data corresponding to the non-uniform spread dominant signal set, sorts the energy data in ascending order according to the spatial distance from the sensor to the source end, and cuts and distributes the sorted energy data into continuous space according to a preset distance interval step size to generate continuous spatial propagation segments. The fluctuation coefficient sequence generation submodule calculates the average value of energy data within each continuous spatial propagation segment, retrieves the maximum and minimum values ​​of energy within the continuous spatial propagation segment and calculates the difference, calculates the energy fluctuation coefficient based on the ratio of the difference to the average value, and constructs the energy fluctuation coefficient sequence.

2. The concrete structure crack damage assessment system based on acoustic emission sensing according to claim 1, characterized in that, The acoustic emission waveform data segment is specifically a data sequence of a specified length extracted from the digital waveform data sequence after the triggering time. The multi-parameter discrete evolution feature set includes the amplitude standard deviation, energy standard deviation, and duration standard deviation calculated for each continuous time window. The non-uniform expansion dominant signal set specifically refers to the set of acoustic emission waveform data segments associated with time windows where the posterior probability is higher than the judgment benchmark probability. The energy fluctuation coefficient sequence includes the dimensionless energy fluctuation coefficient calculated for each spatial propagation segment. The boundary of the dynamic active zone of concrete structure crack damage is specifically the starting distance value of the spatial propagation segment closest to the acoustic emission source location coordinates in the energy attenuation stable zone.

3. The concrete structure crack damage assessment system based on acoustic emission sensing according to claim 1, characterized in that, The signal acquisition module includes: The elastic wave capture and conversion submodule acquires the acoustic emission sensor array arranged on the surface of the concrete structure, monitors the mechanical vibration waves generated during the stress process of the concrete structure in real time, collects the mechanical vibration waves as the initial signal, sets the sampling frequency and quantization bits to perform analog-to-digital conversion on the initial signal, arranges the converted values ​​in chronological order, and generates a digital waveform data sequence. The trigger time scanning and identification submodule sets a background noise reference value to distinguish the attributes of the initial signal, scans the voltage amplitude in the digital waveform data sequence point by point according to the time step, compares it with the background noise reference value, locates the first time node where the voltage amplitude exceeds the background noise reference value, marks it as the starting position of the waveform, and obtains the digital waveform trigger time. The waveform data truncation and storage submodule locates the digital waveform trigger time and subsequent time interval on the time axis of the digital waveform data sequence, sets the waveform truncation length parameter, truncates all digital points within the waveform truncation length parameter range after the trigger time, and generates acoustic emission waveform data segments.

4. The concrete structure crack damage assessment system based on acoustic emission sensing according to claim 3, characterized in that, The feature discretization evolution module includes: The basic parameter parsing and extraction submodule parses the waveform structure of the acoustic emission waveform data segment, searches for the point with the maximum absolute voltage value as the amplitude data by traversing the waveform data points, performs envelope detection operation on the acoustic emission waveform data segment to construct the waveform envelope, performs time integration operation on the waveform envelope to obtain the area value as the energy data, detects the time difference between the first and last crossing of the voltage threshold of the waveform as the duration data, and obtains the waveform basic parameter set; The window statistical mean calculation submodule, based on the generation timestamps of amplitude, energy and duration data in the waveform basic parameter set, maps the amplitude, energy and duration data to multiple consecutive time windows of preset length in a time sequence, extracts all amplitude, energy and duration data in each independent time window and calculates their respective average values ​​to obtain the window parameter average value information; The discrete feature set construction submodule calculates the standard deviation between the amplitude, energy, and duration data within each independent time window and the corresponding average values ​​in the window parameter average information. The standard deviations of multiple dimensions are combined and arranged to generate a multi-parameter discrete evolution feature set.

5. The concrete structure crack damage assessment system based on acoustic emission sensing according to claim 4, characterized in that, Constructing the waveform envelope by performing envelope detection on the acoustic emission waveform data segment includes: Obtain the time-domain discrete voltage numerical sequence composed of acoustic emission waveform data segments; The Fast Fourier Transform algorithm is invoked to perform a mapping process from the time domain to the frequency domain on the discrete voltage numerical sequence in the time domain, resulting in a frequency domain signal sequence; Identify positive and negative frequency components in a frequency domain signal sequence; The values ​​of the negative frequency components are set to zero, and the adjusted one-sided spectrum sequence is calculated based on the positive frequency components and the preset weighting coefficient. The inverse fast Fourier transform algorithm is called to perform an inverse transform on the one-sided spectral sequence, generating an analytic signal sequence in complex form; For each sampling moment in the analytical signal sequence, the corresponding real part data value and imaginary part data value are separated; Calculate the square of the real part based on the real part data value, and calculate the square of the imaginary part based on the imaginary part data value; Calculate the sum of squares of the modulus based on the square values ​​of the real part and the imaginary part; The instantaneous amplitude of the analytical signal sequence at the sampling time is calculated based on the sum of squared moduli. Connect the instantaneous amplitude values ​​corresponding to all sampling times in chronological order to generate the waveform envelope.

6. The concrete structure crack damage assessment system based on acoustic emission sensing according to claim 1, characterized in that, For non-uniformly expanded posterior probabilities, the formula is used: ； in, Let E be the posterior probability of non-uniform expansion given that the current state H of simultaneous changes in the standard deviations of multiple parameters is observed. This represents the prior probability of non-uniform spread. To determine the likelihood probability of observing a state H with synchronized changes in the standard deviations of multiple parameters during non-uniform expansion, To determine the likelihood probability of observing a state H with synchronized changes in the standard deviations of multiple parameters without non-uniform expansion. This is a weighting factor for the intensity of a state where the standard deviations of multiple parameters change synchronously.

7. The concrete structure crack damage assessment system based on acoustic emission sensing according to claim 1, characterized in that, Calculating the straight-line distance between the sensor's spatial coordinates and the acoustic emission source's location coordinates includes: The horizontal, vertical, and longitudinal coordinate components of the sensor spatial coordinates and the acoustic emission source positioning coordinates are obtained in a three-dimensional rectangular coordinate system. Calculate the squared values ​​of the corresponding coordinate differences based on the horizontal coordinate components, vertical coordinate components, and vertical coordinate components respectively; The sum of the squares of the spatial distances is calculated based on the squared values ​​of the coordinate differences corresponding to the three dimensions. The square root of the sum of the squared spatial distances is then performed to obtain the straight-line distance.

8. The concrete structure crack damage assessment system based on acoustic emission sensing according to claim 1, characterized in that, The boundary delineation module includes: The coefficient stability interval initial screening submodule scans the energy fluctuation coefficient sequence item by item along the distance increasing direction, checks whether there are three adjacent energy fluctuation coefficients in the sequence that are all lower than the preset stability discrimination threshold, records the position index of the corresponding energy fluctuation coefficient in the energy fluctuation coefficient sequence, and obtains the candidate stable region index; The gradient decay condition verification submodule extracts the continuous spatial propagation segment corresponding to the candidate stable region index, calculates the average energy difference between the continuous spatial propagation segments, and compares it with a preset gradient threshold. If the average energy difference is less than the gradient threshold, it is confirmed that the energy decay has become gradual, and an energy decay stable region is generated. The active zone boundary positioning submodule retrieves all continuous spatial propagation segments within the energy attenuation stable zone, identifies the spatial propagation segment closest to the acoustic emission source's positioning coordinates, extracts the starting distance of the corresponding spatial propagation segment, defines it as the boundary point between the dynamic active zone and the stable zone of crack damage, and generates the boundary of the dynamic active zone of crack damage in the concrete structure.

Images