Urban underground comprehensive pipe gallery concrete engineering quality evaluation method

Abstract

This invention belongs to the technical field of non-destructive testing of concrete structures, and relates to a method for evaluating the quality of concrete engineering projects in urban underground integrated pipe corridors. The method includes: acquiring raw acoustic emission signals and performing structured processing to generate a structured acoustic emission event data object containing multi-dimensional feature information; combining environmental data to perform feature correction and verification of the data object, generating a focused scanning command containing the target's three-dimensional coordinates and risk confidence; generating a parameterized adaptive focused scanning configuration package through inverse mapping of physical parameters based on energy spectrum features, driving a phased array ultrasonic system to target and image the target area and extract acoustic image features of defects; finally, comparing the correlation between the data object and image features, performing closed-loop verification, and adaptively updating the discrimination parameters used for risk assessment. This invention solves the technical problem in the prior art where passive monitoring and active detection are disconnected, leading to insufficient evaluation accuracy and delayed verification.

Description

Technical Field

[0001] This invention belongs to the technical field of non-destructive testing of concrete structures, and relates to a method for evaluating the quality of concrete engineering projects in urban underground integrated pipe corridors. Background Technology

[0002] Accurate evaluation of the early engineering quality is a prerequisite for ensuring the long-term safety and durability of large concrete structures such as urban underground utility tunnels during construction and operation.

[0003] In existing technological practices, two methods are commonly used: passive acoustic emission monitoring and active ultrasonic testing. Passive acoustic emission monitoring uses sensors deployed in concrete structures to listen to transient elastic wave signals generated by stress release behaviors such as the initiation and propagation of microcracks inside the material, thereby enabling early warning of potential damage. Active ultrasonic testing, on the other hand, uses active emission and reception of ultrasonic waves to image the interior of the structure, in order to locate and quantify the size and shape of defects.

[0004] However, the aforementioned technologies have limitations in practical applications. Acoustic emission monitoring, as a passive method, has complex signal sources. In addition to structural damage, non-structural factors such as changes in environmental temperature and humidity and surrounding vibrations can also generate similar signals, resulting in low signal-to-noise ratios in the raw data and making it difficult to directly correlate detected events with actual defects. Although active ultrasonic testing has high resolution, it consumes a lot of time and resources for comprehensive scanning of large-volume linear projects and is usually only used as a sampling or confirmation method, resulting in intervention lag. Furthermore, passive monitoring and active detection are relatively independent in terms of data and workflow, lacking effective data interfaces and collaborative mechanisms. This makes it difficult for early warning information to automatically and accurately guide active detection equipment for targeted verification, and existing static evaluation models are difficult to adaptively adjust according to the environmental conditions and material properties of specific pipe gallery sections, affecting the evaluation accuracy in complex field environments. Summary of the Invention

[0005] To address the aforementioned problems, this invention provides a method for evaluating the quality of concrete engineering in urban underground utility tunnels.

[0006] A method for evaluating the quality of concrete engineering in urban underground utility tunnels includes the following steps:

[0007] S1. Obtain the original acoustic emission signal inside the concrete structure, perform wavelet packet transform on the original acoustic emission signal, decompose it into a preset number of energy sub-bands, and calculate the normalized energy value of each energy sub-band to construct an energy spectral density vector; calculate the Shannon entropy value, which characterizes the signal complexity, based on the energy distribution probability in the energy spectral density vector; use the time difference of arrival (TDOA) localization algorithm to solve the source three-dimensional spatial coordinates of the location where the original acoustic emission signal occurred, and generate a unique event identifier based on the original waveform data; encapsulate the source three-dimensional spatial coordinates, energy spectral density vector, Shannon entropy value, and event unique identifier to generate a structured acoustic emission event data object containing multi-dimensional feature information;

[0008] S2. Receive structured acoustic emission event data objects, combine them with environmental data to perform feature correction and multi-dimensional verification, and generate a focused scan command to be verified, which includes the target's three-dimensional spatial coordinates and risk confidence.

[0009] S3. Based on the focused scan command to be verified, a parameterized adaptive focused scan configuration package is generated by inverse mapping of physical parameters based on energy spectrum characteristics.

[0010] S4. In response to the focused scanning configuration package, drive the phased array ultrasound diagnostic system to perform targeted imaging of the target area and extract the acoustic image features of the defects.

[0011] S5. Compare the correlation between the structured acoustic emission event data object and the acoustic image features of the defect, perform closed-loop verification, and adaptively update the discrimination parameters used for risk determination in step S2.

[0012] A further aspect of the present invention, step S2, includes the following steps:

[0013] Acquire local concrete temperature data and overall structural strain data that are synchronized with the structured acoustic emission event data object in time, and perform weighted correction on the energy spectral density vector in the data object;

[0014] The corrected energy spectral density vector and the Shannon entropy value in the data object are compared with the preset background noise model to calculate the abnormal deviation of the current event.

[0015] The trend of the change in the slope of the rising edge of the signal of acoustic emission events that occur continuously within a preset time window is detected.

[0016] When the abnormal deviation exceeds the first risk threshold and the slope of the rising edge of the signal shows an overall upward trend, it is judged as a high-risk event, and a focused scan command to be verified is generated.

[0017] A further aspect of the present invention, step S3, includes the following steps:

[0018] Extract the main frequency band energy value from the structured acoustic emission event data object corresponding to the focused scanning command to be verified, and map it to the target detection center frequency through a preset physical inversion function;

[0019] The pulse width is selected from the preset mapping table based on the risk confidence level in the focusing scan command to be verified, and the scanning aperture angle that can cover the error area is calculated based on the positioning error radius.

[0020] The target detection center frequency, pulse width, scanning aperture angle, and the event unique identifier in the corresponding data object are encapsulated to generate a focused scanning configuration package.

[0021] A further aspect of the present invention, step S4, includes the following steps:

[0022] After verifying that the unique event identifier in the focused scan configuration package matches the pre-stored focused scan command to be verified, the emission delay time of each element of the two-dimensional array transducer is calculated according to the parameters in the configuration package.

[0023] A two-dimensional array transducer is driven to emit a focused ultrasonic beam and receive echo signals to synthesize an acoustic tomographic image of the target area.

[0024] Abnormal response regions are identified from acoustic tomographic images, and the equivalent diameter, maximum echo amplitude, and acoustic attenuation coefficient of these regions are calculated to constitute the acoustic image features of the defects.

[0025] A further aspect of the present invention, step S5, includes the following steps:

[0026] Calculate the correlation coefficient between the maximum echo amplitude in the acoustic image features of the defect and the total energy in the structured acoustic emission event data object that triggered this scan;

[0027] The correlation coefficient is compared with the preset confirmation threshold. If it is greater than the confirmation threshold, it is determined to be a real defect activity. If it is less than or equal to the confirmation threshold, it is determined to be environmental noise or a false signal.

[0028] Based on the judgment result, the risk judgment threshold used to calculate the abnormal deviation degree in step S2 is adjusted using probability statistics update logic.

[0029] A further aspect of this invention involves weighted correction of the energy spectral density vector by: calling a preset environmental impact compensation model to calculate the signal attenuation coefficient caused by environmental temperature changes and structural deformation; and using the signal attenuation coefficient to compensate the original energy spectral density vector in order to eliminate non-damaging signal drift introduced by environmental factors.

[0030] In a further embodiment of the present invention, the physical inversion function is configured to establish a cube root mapping relationship between the normalized energy value of the main frequency band of the acoustic emission signal and the center frequency of the ultrasonic target detection, so that the calculated target detection center frequency matches the physical scale of the potential defect.

[0031] A further aspect of this invention involves calculating the acoustic attenuation coefficient by estimating the logarithmic attenuation rate of the echo signal at different depth intervals below the abnormal response region. The depth intervals are selected to avoid near-field interference and bottom reflection regions.

[0032] A further aspect of this invention, adjusting the risk assessment threshold, specifically includes:

[0033] If the activity is determined to be a genuine defect, the structured acoustic emission event data object will be stored in the high-risk defect fingerprint database as a positive sample to lower the risk judgment threshold.

[0034] If the signal is determined to be environmental noise or a spurious signal, the structured acoustic emission event data object is stored in the noise database as a negative sample to increase the risk assessment threshold.

[0035] In summary, the present invention has the following beneficial technical effects:

[0036] 1. By constructing structured acoustic emission event data objects and combining energy spectrum, signal complexity, and temporal evolution trends for multi-dimensional feature verification, while simultaneously using synchronously acquired temperature and strain data to correct the signal, this mechanism can effectively suppress interference signals caused by non-structural factors such as environmental temperature changes and structural non-damage deformation. This allows for more accurate screening of events with high risk confidence from complex monitoring data, reducing the ineffective investment of subsequent detection resources.

[0037] 2. A physical parameter inverse mapping technique is employed to calculate the target detection center frequency based on the energy spectrum characteristics of acoustic emission events, and an adaptive focused scanning configuration package is generated by combining positioning error and risk confidence. This method enables the frequency, energy, and spatial coverage of the active ultrasonic beam to match the physical properties of potential defects. Compared to general parameter scanning, this method can improve the imaging signal-to-noise ratio for suspicious areas and obtain clearer acoustic image characterization of defects.

[0038] 3. A closed-loop verification mechanism based on unique identifiers (such as hash values) was established. By comparing the total energy of acoustic emission events with the amplitude of defect echoes obtained from focused imaging, the correlation between the two in terms of physical quantities was analyzed. This mechanism provides a quantitative basis for determining whether passive monitoring signals originate from real defect activity, realizing a logical closed loop from discovering suspicious signals to confirming physical defects, and solving the problem that single passive monitoring is insufficient to confirm the source of signals.

[0039] 4. A positive and negative sample fingerprint database is constructed using the results of closed-loop verification, and the threshold parameters for risk assessment are dynamically adjusted using probabilistic statistical logic (such as Bayesian updates). This continuous feedback mechanism enables the system to learn the background noise characteristics and real defect signal fingerprints under specific utility tunnel environments. As data accumulates, the evaluation model can iterate its parameters for the current environment, which helps to reduce the false alarm rate and false negative rate during long-term monitoring. Attached Figure Description

[0040] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. The drawings are used to provide a further understanding of the present invention.

[0041] Figure 1 A flowchart illustrating an embodiment of this application is disclosed.

[0042] Figure 2 Structural schematic diagrams of embodiments of this application are disclosed. Detailed Implementation

[0043] The following is in conjunction with the appendix Figure 1 - Figure 2 A preferred description of the present invention is provided below.

[0044] See attached document Figure 1 This invention proposes a method for evaluating the quality of concrete engineering in urban underground utility tunnels, comprising the following steps:

[0045] S1. Obtain the original acoustic emission signal inside the concrete structure, and perform structured processing on the original acoustic emission signal to generate a structured acoustic emission event data object containing multi-dimensional feature information.

[0046] S2. Receive structured acoustic emission event data objects, combine them with environmental data to perform feature correction and multi-dimensional verification, and generate a focused scan command to be verified, which includes the target's three-dimensional spatial coordinates and risk confidence.

[0047] S3. Based on the focused scan command to be verified, a parameterized adaptive focused scan configuration package is generated by inverse mapping of physical parameters based on energy spectrum characteristics.

[0048] S4. In response to the focused scanning configuration package, drive the phased array ultrasound diagnostic system to perform targeted imaging of the target area and extract the acoustic image features of the defects.

[0049] S5. Compare the correlation between the structured acoustic emission event data object and the acoustic image features of the defect, perform closed-loop verification, and adaptively update the discrimination parameters used for risk determination in step S2.

[0050] In one embodiment of the present invention, step S1 includes the following steps:

[0051] The original acoustic emission signal is subjected to wavelet packet transform to decompose it into a preset number of energy sub-bands, and the normalized energy value of each energy sub-band is calculated to construct the energy spectral density vector.

[0052] Based on the energy distribution probability in the energy spectral density vector, calculate the Shannon entropy value, which characterizes the signal complexity;

[0053] The three-dimensional spatial coordinates of the source of the original acoustic emission signal are calculated using the time difference of arrival localization algorithm, and a unique event identifier is generated based on the original waveform data.

[0054] The three-dimensional spatial coordinates of the earthquake source, the energy spectral density vector, the Shannon entropy value, and the unique identifier of the event are encapsulated to construct a structured acoustic emission event data object.

[0055] Specifically, the data processing unit continuously captures transient elastic wave signals generated by the release of internal stress in concrete using a distributed acoustic emission sensor array deployed inside or on the surface of the concrete structure of urban underground utility tunnels at a preset high sampling rate. It should be noted that the distributed acoustic emission sensor array consists of at least four sensor nodes deployed in a spatial topology to ensure a unique solution for the time difference of arrival (TDOA) localization algorithm; each sensor node achieves microsecond-level time alignment through a unified time synchronization protocol, which can adopt the IEEE 1588 precise time protocol; the preset high sampling rate is on the order of megahertz to ensure the capture of high-frequency components in the acoustic emission signal. Specifically, the high sampling rate is set to 2 MHz to 5 MHz, a range that covers the main frequency band of acoustic emission signals released by typical microcracks in concrete, typically 20 kHz to 1 MHz, while allowing sufficient Nyquist margin to suppress aliasing. The sampling rate is selected based on the highest frequency component with 99% energy concentration in the spectrum analysis of the signal collected during fracture experiments on C50 concrete standard test blocks, multiplied by a safety factor. The sensor node digitizes the captured raw voltage signal via an analog-to-digital converter and then uploads it to the data processing unit in real time via an industrial Ethernet or wireless data transmission network.

[0056] After receiving the raw acoustic emission signal, the data processing unit first performs a digital bandpass filter to remove out-of-band noise. The passband frequency range of the bandpass filter is set to 50 kHz to 800 kHz. This range is an effective frequency band obtained from a large number of measured and statistically analyzed concrete acoustic emission signals, which can effectively retain damage-related signals while suppressing low-frequency environmental vibrations and high-frequency electronic noise. Then, a sliding time window detection method is used to identify independent pulse events in the signal whose amplitude exceeds a preset noise floor threshold. This noise floor threshold is set to 3-5 times the root mean square value of the amplitude of the background signal continuously acquired for 10 minutes under no external interference. The specific value is obtained through on-site calibration, for example, a typical value of ±10 mV. For each identified pulse event, the data processing unit performs wavelet packet transform to decompose the time-domain waveform of the pulse event into multiple preset energy sub-frequency bands. Specifically, the Daubechies wavelet basis function is selected to perform wavelet packet decomposition of the truncated event waveform at a predetermined number of levels to obtain the time-domain signal of the corresponding number of sub-frequency bands. The number of wavelet packet decomposition layers is set to 4 to 6, corresponding to 16 to 64 sub-bands. This number of layers is determined based on a trade-off between the bandwidth of typical concrete acoustic emission signals and the required frequency resolution. Experiments have verified that this number of decomposition layers achieves a good balance between computational complexity and frequency band resolution.

[0057] Subsequently, the energy value of each sub-band signal is calculated, and then the energy values ​​of all sub-bands are normalized to form a probability vector of energy distribution. Based on this probability vector, the data processing unit calculates the Shannon entropy value, which characterizes the signal complexity of the pulse event. It should be noted that the formula for calculating the Shannon entropy value H is... ,in, In this embodiment, the total number of sub-bands is [number]. The number of wavelet packet decomposition levels determines the decomposition level; for example, a 4-level decomposition corresponds to... It is 16. For the first The normalized energy value of a sub-band represents the proportion of that sub-band's energy to the total energy of the event. The base of the Shannon entropy is typically set to 2, measuring the uncertainty of information in bits. Simultaneously with signal feature extraction, the data processing unit calls the time-difference-of-arrival (TDOA) localization algorithm. Utilizing the precise time difference between the arrival times of the same pulse event at different sensor nodes in the distributed acoustic emission sensor array, and combining this with the known three-dimensional coordinates of the sensor nodes and the average sound velocity in the concrete, the three-dimensional spatial coordinates of the source of the acoustic emission event are calculated. It should be noted that the TDOA localization algorithm used here employs the Geiger iterative method to calculate the source coordinates, where the average sound velocity in the concrete is preset to a typical value corresponding to the concrete grade. For common C30-C60 concrete, the average sound velocity is preset to 3800 m / s to 4500 m / s. Specific values ​​can be obtained by pre-embedding sound velocity calibration blocks during pipe gallery pouring or by on-site measurement using the penetration method, and are input as configurable parameters in the system. For example, the preset sound velocity for C50 concrete is typically 4000 m / s.

[0058] Finally, the data processing unit integrates the absolute timestamp of the event, the calculated three-dimensional spatial coordinates of the source, the energy spectral density vector composed of the normalized energy values ​​of each sub-band, and the calculated Shannon entropy value. Based on all the original waveform data of the event, it generates a unique event hash value using the SHA-256 hash algorithm. It should be understood that this unique event hash value is generated by calculating the event waveform data block using the SHA-256 algorithm and is used to uniquely identify and trace the structured acoustic emission event data object in subsequent processes. All the above data items are collectively encapsulated to construct a structured acoustic emission event data object. This data object is a multi-dimensional data object containing time, space, frequency domain energy, signal entropy, and a unique identifier, fully presenting the acoustic emission event.

[0059] The preset noise floor threshold is set by collecting a background signal under conditions of no external interference after system deployment, and then taking a number of times the standard deviation of the peak amplitude. Specifically, the system continuously collects background signals for at least 30 minutes during periods without construction or traffic interference, and calculates the standard deviation of the signal amplitude for that period. Set the noise floor threshold to to This multiplier is determined based on a 99.7% to 99.99% confidence interval under the assumption of a normal distribution, to minimize false triggering by random noise. The wavelet packet transform decomposition layer is multi-layered. This parameter is a typical value set after extensive experimental statistics, based on the main energy frequency band distribution width of the acoustic emission signals released by early-stage concrete microcracks. Increasing the number of decomposition layers increases the computational burden, while decreasing the number of layers may lead to insufficient frequency band resolution. Specific experimental methods include: prefabricating microcracks of different sizes, such as 0.1 mm to 2 mm, on standard laboratory specimens, collecting their acoustic emission signals, and analyzing the discriminative power of the energy spectral density vector for crack scale under different decomposition layer numbers. The results show that when the number of decomposition layers is 4 to 6, the signals of cracks of different scales exhibit the most significant differences in the energy spectrum, measured by the ratio of inter-class variance to intra-class variance; therefore, this range is considered the recommended setting.

[0060] For example, assume the sensor array consists of four sensors with coordinates S1(0, 0, 0), S2(5, 0, 0), S3(0, 5, 0), and S4(0, 0, 5) (unit: meters). The data processing unit receives a set of synchronized raw acoustic emission signals from the network interface and detects pulse events with amplitudes exceeding a preset noise floor threshold. The system captures waveform data within a predetermined time window after the event occurs. The time window length is typically 3 to 5 times the signal rise time, for example, 100 μs to 500 μs. First, a wavelet packet decomposition of, for example, 3 layers is performed to obtain, for example, 8 sub-bands. The energy of each sub-band is calculated, assuming the normalized energy probability vector is [0.40, 0.25, 0.15, 0.10, 0.05, 0.03, 0.01, 0.01]. The energy is calculated using the Shannon entropy formula: Bits. Simultaneously, the positioning algorithm, based on the relative time differences of the event's arrival at the four sensors [t1=0 μs, t2=1250 μs, t3=1250 μs, t4=1800 μs], combined with a preset sound velocity, iteratively calculates the source coordinates as (2.0, 2.0, 2.0) meters. The absolute timestamp is recorded as T=2023-10-27 14:30:05.123456. Next, a SHA-256 hash operation is performed on the captured raw waveform data to generate, for example, a 64-bit hexadecimal hash value, such as "a1b2c3d4e5f6…". Finally, these data items—timestamp T, coordinates (2.0, 2.0, 2.0), energy vector [0.40, 0.25, 0.15, 0.10, 0.05, 0.03, 0.01, 0.01], entropy value 2.22, and hash value “a1b2c3d4e5f6…”—are integrated to generate a structured acoustic emission event data object.

[0061] In one embodiment of the present invention, step S2 includes the following steps:

[0062] Acquire local concrete temperature data and overall structural strain data that are synchronized with the structured acoustic emission event data object in time, and perform weighted correction on the energy spectral density vector in the data object;

[0063] The corrected energy spectral density vector and the Shannon entropy value in the data object are compared with the preset background noise model to calculate the abnormal deviation of the current event.

[0064] The trend of the change in the slope of the rising edge of the signal of acoustic emission events that occur continuously within a preset time window is detected.

[0065] When the abnormal deviation exceeds the first risk threshold and the slope of the rising edge of the signal shows an overall upward trend, it is judged as a high-risk event, and a focused scan command to be verified is generated.

[0066] Specifically, firstly, the data processing unit retrieves data collected within the same time window from the temperature sensor and fiber Bragg grating strain sensor, which are deployed synchronously with the acoustic emission sensor array. The temperature data is used to obtain the real-time temperature value of the local concrete area, and the strain data is used to obtain the overall deformation of the concrete structure. It should be noted that the alignment error between the sampling timestamp of the synchronously acquired temperature and strain data and the absolute timestamp of the structured acoustic emission event data object must be less than a preset minimum time value, such as 1 ms.

[0067] The data processing unit uses a pre-defined environmental impact compensation model to correct the energy spectral density vector in the structured acoustic emission event data object. This compensation model is defined as multiplying each component of the original energy spectral density vector by an attenuation coefficient related to environmental temperature changes and overall strain. The attenuation coefficient in this environmental impact compensation model... The calibration was performed experimentally. The calibration method involved applying different temperatures and strain levels to concrete specimens in a laboratory constant-temperature chamber and measuring the energy changes of the acoustic emission signals generated by the standard fracture source. Compensation coefficients were obtained through multiple linear regression fitting. and The usual form is ,in The change is relative to a reference temperature. This represents the change in strain relative to the zero-stress state. and For material-related temperature compensation coefficients and strain compensation coefficients, such as for C50 concrete, The typical value range is 0.01 to 0.03 per degree Celsius. The typical value range is 0.1 to 0.5 per microstrain. The corrected energy spectral density vector aims to eliminate the non-destructive acoustic emission energy contribution caused by thermal expansion and contraction of the material and overall structural bending.

[0068] Next, the data processing unit accesses a pre-established database of background noise models for concrete during normal hydration. This database stores the mean values ​​of typical energy spectral density vectors and Shannon entropies, along with their covariance matrices, obtained through long-term monitoring and statistical analysis of defect-free pipe gallery sections under different curing ages and environmental conditions. The background noise model is established as follows: after the pipe gallery is poured, at least three known defect-free typical sections are selected as reference sections. During the standard curing period and subsequent operation and maintenance, at least 30 days of continuous data collection are conducted, accumulating over 100,000 acoustic emission events. Data is grouped by age and ambient temperature range, and the mean vector μ of the energy spectral density vector and Shannon entropy value for all events within each group, along with the covariance matrix S, are calculated and stored in the database. The data processing unit calculates the Mahalanobis distance between the corrected energy spectral density vector and Shannon entropy value of the current event and the corresponding age and condition data in the background noise model, quantifying this distance value as the abnormal deviation of the current event.

[0069] Simultaneously, the data processing unit retrieves all historical structured acoustic emission event data objects occurring within a preset time window, centered on the current event's source coordinates, and within a certain spatial radius. It extracts the rising slope of these event waveform signals and analyzes their trend over time. Here, the rising slope is obtained by linear regression of data within a certain percentage range of the peak amplitude at the beginning of the event waveform, such as 10% to 90%. The preset time window is typically set to 5 to 30 minutes, and the spatial radius is set to 2 to 3 times the positioning error radius, for example, 0.3 meters to 0.9 meters. If the calculated abnormal deviation exceeds the first risk threshold, and the rising slope of consecutive events within the time window, after linear fitting, has a positive slope value greater than the second trend threshold, then the current event is determined to be a high-risk event. Conversely, if only the abnormal deviation exceeds the first risk threshold but the rising slope does not show an overall upward trend (or a continuous increasing trend), then it is determined to be a low-risk event. The first risk threshold is based on the distribution of Mahalanobis distance values ​​for all normal events accumulated during the background noise model building phase. The 95th to 99th quantile of this distribution is used as the initial threshold, typically ranging from 3.0 to 5.0. For example, if the Mahalanobis distance for normal events follows a chi-square distribution with the degrees of freedom as the characteristic dimension, the chi-square value corresponding to the 99% confidence level is used as the threshold. The second trend threshold is set based on experimental data of acoustic emission signals simulating stable and accelerated crack propagation. When the rising slope trend is greater than 0.05 to 0.2, it usually indicates that the crack has entered the unstable propagation stage. This threshold is determined by analyzing the historical data of the rising slope changes of acoustic emission signals from known accelerated failure specimens, and taking the average slope of the initial acceleration point.

[0070] Finally, based on the judgment result, the data processing unit generates a focusing scan instruction to be verified. The data structure of this instruction includes the target's three-dimensional spatial coordinates, risk level identifier, calculated risk confidence, and event hash value in the structured acoustic emission event data object that triggered this judgment.

[0071]

[0072] in, This represents the calculated Mahalanobis distance, i.e., the degree of anomaly deviation; It is a combined feature vector, which is formed by concatenating the corrected energy spectral density vector of the current event with the Shannon entropy value; It is the feature mean vector retrieved from the background noise model under the corresponding conditions; It is the feature covariance matrix under corresponding conditions. This formula is used to quantify the multidimensional deviation between the features of the current event and the statistical features of the normal background.

[0073] The background noise model for concrete during normal hydration is established by continuously collecting tens of thousands of acoustic emission event data points during the standard curing period (7 to 28 days) after concrete pouring in a known defect-free pipe gallery demonstration section. The data are divided into age intervals based on predetermined time points, and the mean and covariance of the energy spectral density vector and Shannon entropy value are statistically analyzed. Within each age interval, further subdivision by ambient temperature is required to eliminate the significant impact of temperature on acoustic emission background noise. At least 1000 event data points are needed in each sub-interval to ensure statistical significance. The first risk threshold is based on the Mahalanobis distance distribution of normal events and known defective events in historical data, selecting the point with the highest discriminative power, typically set between 3.0 and 5.0. Specifically, this can be achieved by calculating the receiver operating characteristic curve of the Mahalanobis distance distribution of normal and defective events, and selecting the Mahalanobis distance value corresponding to the point with the largest Youden exponent as the threshold. The second trend threshold is used to determine whether the rising slope shows a significant increasing trend, typically set between 0.05 and 0.2, based on the physical law that the signal front steepens when simulating accelerated crack propagation. Risk confidence level is determined by abnormal deviation. Obtained through Sigmoid function mapping, the calculation formula is: ,in For the bias parameter related to the first risk threshold, such that when When equal to the threshold, Approximately 0.5. Bias parameter. It is directly set to the current value of the first risk threshold, i.e. This is the current first risk threshold. Thus, when... When the value equals the threshold, the exponent term is 0. The value is 0.5, which aligns with the intuitive definition of risk confidence level.

[0074] For example, the data processing unit receives a structured acoustic emission event data object with an event hash value of "a1b2c3d4e5f6…". The system synchronously retrieves the temperature data (25 degrees Celsius) and strain data (50 microstrain) collected at the time of the event (T=2023-10-27 14:30:05.123456). Assume a temperature compensation coefficient… strain compensation coefficient Then calculate the attenuation coefficient. The energy spectral density vector [0.40, 0.25, 0.15, 0.10, 0.05, 0.03, 0.01, 0.01] in the structured acoustic emission event data object is corrected using this coefficient, resulting in a corrected vector of approximately [0.35994, 0.22496, 0.13498, 0.08999, 0.04499, 0.02700, 0.00900, 0.00900]. The Shannon entropy value of 2.22 is temporarily unaffected by this compensation. Next, the system queries the background noise model database. Assuming the current concrete age is 21 days and environmental conditions are similar, the corresponding feature mean vector is retrieved. The values ​​are [0.05, 0.04, 0.10, 0.15, 0.20, 0.18, 0.16, 0.12, 1.8], where the first 8 dimensions are the energy spectrum and the last dimension is the entropy and covariance matrix. Given a 9x9 positive definite matrix, for simplicity, assume it is a diagonal matrix, with diagonal elements representing the variances of each eigencomponent, assumed to be 0.01. The current eigenvector... =[0.35994, 0.22496, 0.13498, 0.08999, 0.04499, 0.02700, 0.00900, 0.00900, 2.22] and and Substituting into the Mahalanobis distance formula, the abnormal deviation is obtained. The risk level is significantly higher than the set first risk threshold. Simultaneously, the system retrieves several historical events from the same area within a preset time period, with rising slopes of [0.15, 0.18, 0.22, 0.26, 0.31] (unit: V / μs). Linear fitting shows a slope of approximately 0.032 per event, exceeding the second trend threshold, for example, 0.02. Therefore, the current event is determined to be high-risk. The calculated risk confidence level is 0.92. Finally, a focused scan command to be verified is generated: {Target coordinates: (2.0, 2.0, 2.0), Risk level: High risk, Risk confidence level: 0.92, Trigger event hash value: “a1b2c3d4e5f6…”}.

[0075] In one embodiment of the present invention, step S3 includes the following steps:

[0076] Extract the main frequency band energy value from the structured acoustic emission event data object corresponding to the focused scanning command to be verified, and map it to the target detection center frequency through a preset physical inversion function;

[0077] The pulse width is selected from the preset mapping table based on the risk confidence level in the focusing scan command to be verified, and the scanning aperture angle that can cover the error area is calculated based on the positioning error radius.

[0078] The target detection center frequency, pulse width, scanning aperture angle, and the event unique identifier in the corresponding data object are encapsulated to generate a focused scanning configuration package.

[0079] Specifically, after receiving the focused scan command to be verified, the data processing unit initiates the process of generating the focused scan configuration package. First, the data processing unit extracts the trigger event hash value from the focused scan command to be verified, and uses this hash value as an index to retrieve the corresponding structured acoustic emission event data object from the local cache or database, thereby obtaining the stored energy spectral density vector. The data processing unit analyzes this energy spectral density vector to identify the dominant frequency band with the highest energy value. Next, the data processing unit calls a preset physical inversion function, which encapsulates the empirical mapping relationship between the dominant frequency energy of the acoustic emission signal and the center frequency of the ultrasonic target detection. It should be understood that this function is specifically embodied in the aforementioned empirical formula. The formula is used to calculate the target detection center frequency. . This represents the normalized energy value of the main frequency band extracted from the energy spectral density vector. It is a constant related to the acoustic properties of concrete materials, and its dimension is frequency.

[0080] The data processing unit takes the normalized energy value of the main frequency band as input and calculates the target detection center frequency that physically matches the scale of the potential defect using a physical inversion function. (Constant) The value ranges from 0.5MHz to 2.0MHz, and its specific value is obtained through laboratory calibration. The calibration method is as follows: a series of concrete test blocks containing artificial holes or cracks of different diameters (e.g., 1mm, 2mm, 5mm) are prepared, and acoustic emission events are collected during loading, recording the normalized energy of the dominant frequency band. Simultaneously, phased array ultrasonic probes with different center frequencies (e.g., 0.5 MHz, 1.0 MHz, 1.5 MHz, 2.0 MHz) were used to scan the defect to find the center frequency with the highest signal-to-noise ratio. For multiple groups ( , Power-law fitting of data points ( = × ), determine coefficients Sum of Indices Typically, n is close to 1 / 3. For ordinary aggregate concrete, The typical value is about 1.0 MHz.

[0081] Simultaneously, the data processing unit extracts the target's three-dimensional spatial coordinates and risk confidence level from the focusing scan command to be verified. Based on the target's three-dimensional spatial coordinates and the positioning error model implicit in the geometry of the sensor array used for positioning, the data processing unit calculates the adaptive scanning aperture angle. This angle calculation ensures that the generated ultrasonic beam's sound field coverage at the target depth completely encompasses the positioning error region. The calculation formula is as follows: .in, It is the source coordinate error radius estimated based on the location model. The estimation formula is: ,in For the speed of sound, Typical values ​​for sensor time synchronization and pickup errors (typically 0.1 μs to 0.5 μs), accuracy attenuation factor. It is a function of the sensor array geometry and can be obtained through simulation calculations. For a tetrahedral array, typically... The value is between 1.5 and 3.0, therefore The typical range is 0.1 meters to 0.3 meters. It is the depth of the target point relative to the surface of the ultrasonic transducer array.

[0082] Furthermore, the data processing unit dynamically selects the pulse width from a preset mapping table based on the risk confidence level. This mapping table is set based on a trade-off between imaging quality, resolution, and penetration: when the risk confidence level P ≥ 0.8, indicating a clear target, a short pulse mode is selected to obtain high axial resolution; when 0.5 ≤ P < 0.8, a medium pulse mode is selected to balance resolution and penetration; and when P < 0.5, a long pulse mode is selected to enhance energy penetration into deep concrete. The specific value of the pulse cycle number must match the ultrasonic center frequency, and the duration of one cycle is... .

[0083] Finally, the data processing unit encapsulates the calculated target detection center frequency, scanning aperture angle, pulse width, and the unique identifier (hash value) of the trigger event used for authentication and traceability into a structured focused scan configuration package with unique execution conditions. The data format of this configuration package is configured as a sequence of instructions that can be parsed by the downstream phased array ultrasound diagnostic system. The focused scan configuration package is a digital data structure that contains at least the following fields: protocol version number, opcode, and center frequency. Aperture angle Pulse cycle count, target coordinates And the hash value of the event that triggered it.

[0084] Among them, scanning aperture angle The calculation depends on the positioning error radius. , The estimation is based on the sensor baseline length and timing measurement error used in the time difference of arrival (TDOA) localization algorithm, typically set to 0.1 to 0.3 meters. Monte Carlo simulations show that, under typical sensor arrangements and timing errors, the semi-major axis of the error ellipse for the seismic source coordinates is mainly distributed within this range.

[0085] For example, the data processing unit receives a focusing scan command to be verified, which includes the trigger event hash value "a1b2c3d4e5f6…". The system uses this hash value to retrieve the corresponding structured acoustic emission event data object and reads its energy spectral density vector as [0.35994, 0.22496, 0.13498, 0.08999, 0.04499, 0.02700, 0.00900, 0.00900], which is the corrected vector. The maximum energy value is then identified. This corresponds to the first sub-frequency band. Assume the calibration constants of the current utility tunnel concrete are retrieved from the material parameter database. Substitute into the formula to calculate the target detection center frequency: Next, the target coordinates (2.0, 2.0, 2.0) meters in the processing command are determined. Assuming the ultrasonic transducer array has moved above this area, its surface coordinates are approximately (2.0, 2.0, 0), therefore the target depth is... Based on the positioning error model, set... Calculate the scanning aperture angle: The risk confidence level in the instruction is 0.92, which is greater than 0.8. Based on the mapping rules, a short pulse mode is selected, and the pulse width is set to 3 cycles. Finally, the focused scan configuration package is generated, and its contents can be represented as: {ver:1.0, cmd:FOCUS_SCAN, fc:0.85, theta:8.58, pulses:3, target:(2.0, 2.0, 2.0), hash:“a1b2c3d4e5f6…”}.

[0086] In one embodiment of the present invention, step S4 includes the following steps:

[0087] After verifying that the unique event identifier in the focused scan configuration package matches the pre-stored focused scan command to be verified, the emission delay time of each element of the two-dimensional array transducer is calculated according to the parameters in the configuration package.

[0088] A two-dimensional array transducer is driven to emit a focused ultrasonic beam and receive echo signals to synthesize an acoustic tomographic image of the target area.

[0089] Abnormal response regions are identified from acoustic tomographic images, and the equivalent diameter, maximum echo amplitude, and acoustic attenuation coefficient of these regions are calculated to constitute the acoustic image features of the defects.

[0090] Specifically, the data processing unit sends the focused scan configuration packet to the phased array ultrasound diagnostic system mounted on the mobile platform via a wireless LAN or industrial bus network. The phased array ultrasound diagnostic system continuously monitors the network port in standby mode. Upon receiving the focused scan configuration packet, it first parses the packet structure and extracts the trigger event hash value. It should be noted that the phased array ultrasound diagnostic system includes a two-dimensional array transducer, whose element spacing is based on the wavelength corresponding to the target detection center frequency. The setting is usually no greater than To avoid grating lobes, for a center frequency of 0.85 MHz and a sound speed of 4000 m / s, the wavelength is approximately 4.7 mm, and the typical element spacing is set to 1 mm to 2 mm. The specific choice of element spacing needs to balance beam deflection capability and manufacturing cost. In this embodiment, the element spacing d satisfies d ≤ For a center frequency of 0.85 MHz, The diameter is approximately 2.35 mm. Choosing a d value of 1.5 mm can provide sufficient element density to achieve effective electronic deflection within ±30° while avoiding grating lobes. The system compares this hash value with the hash values ​​in its stored list of instructions to be verified. If a match is found, the configuration package is deemed valid, and the scan execution thread is activated; if a match fails, the configuration package is discarded, and an error code is returned to the data processing unit.

[0091] After successful verification, the beamforming controller of the phased array ultrasound diagnostic system reads the target detection center frequency, scanning aperture angle, and pulse width parameters defined in the focused scanning configuration package. Based on the target detection center frequency and scanning aperture angle, combined with the known geometric parameters of the transducer array and the preset sound velocity in the concrete, the controller calculates the precise delay time used to control the emission timing of each element of the two-dimensional area array transducer. It should be understood that the preset sound velocity... The velocity of sound is set according to the grade of the concrete in the utility tunnel; for example, C50 concrete is typically set to 4000 m / s to 4500 m / s. The specific velocity of sound can be obtained on-site after system installation by measuring the ultrasonic flight time between two points at a known distance, and then used as the system parameter input.

[0092] Subsequently, the controller drives the high-voltage pulse transmitting circuit to excite each element of the two-dimensional array transducer to emit phase-modulated ultrasonic pulses according to the calculated delay sequence and set pulse width. These pulses coherently superimpose in the concrete medium, forming an ultrasonic beam with focused energy on the target's three-dimensional spatial coordinates. The system controls the moving platform or internal electronic scanning logic to perform a gridded scan of the area centered on the target coordinates along a preset path. This scanning path is typically a rectangular grid, with the grid spacing set according to the required imaging resolution. The resolution requirement is usually half the minimum detectable defect size; for example, if the minimum detectable defect is 2mm, then the resolution needs to reach 1mm. The grid spacing Δx and Δy should satisfy the sampling theorem and are generally no greater than half the wavelength. For 0.85MHz, It is approximately 2.35mm, with a typical grid spacing of 1mm to 2mm.

[0093] During the scanning process, the same transducer array switches to receiving mode to acquire echo signals reflected and scattered from inside the concrete. The received analog echo signals are amplified, filtered, and converted from analog to digital before being sent to the image processing unit. The image processing unit employs a time-delay superposition algorithm, also known as beamforming, which essentially calculates the propagation time of the received signal from each array element to that point for each imaging pixel and then aligns and superimposes the signals to enhance the signal from that point. The image processing unit performs time-delay superposition processing on the echo amplitude data from all scan points and arranges them according to the scan position, ultimately synthesizing a B-Scan (depth-horizontal position) or C-Scan (horizontal plane) acoustic image of the region.

[0094] Finally, the system identifies anomalous response regions (i.e., suspected defect regions) from the synthesized acoustic image using edge detection and connected component analysis algorithms. It then extracts the geometric dimensions, maximum echo amplitude, and acoustic attenuation coefficient (obtained by calculating the energy attenuation of the echo signal within a specific depth range) of these regions, collectively forming the acoustic image feature data block of the defect. Edge detection can employ the Canny operator, with its high and low thresholds adaptively determined by analyzing the grayscale histogram of the entire image: high threshold... Set as the top 20% quantiles of gray values ​​among pixels whose gray values ​​are greater than the mean of all pixels plus twice the standard deviation; low threshold. Set as The sound attenuation coefficient here is 40% to 50%. The energy logarithmic attenuation rate of the echo signal at different depths below the defect region is estimated by calculating the following formula: ,in and These are the echo amplitudes at two different depths. This represents the depth difference between the two. Depth range. The selection of the depth should avoid near-field interference and bottom surface reflection effects, and is usually taken in the range of 20mm to 50mm below the defect area. The diameter should generally be no less than 10 mm to ensure the stability of attenuation measurement.

[0095]

[0096] in, It is the calculated number of The transmission delay time of each array element relative to the reference array element. It is the first The coordinates of each array element on the array plane It refers to the target's three-dimensional spatial coordinates specified in the focused scan configuration package, i.e., the focal point position. It's the focal length, in this dynamic focus mode. equal to target depth . This is the propagation speed of ultrasound in concrete, a preset constant. This formula enables precise adjustment of the wavefront curvature center to the target point. Phased focusing.

[0097] The scanning path consists of a rectangular grid centered on the target coordinates, with the grid spacing set according to the required imaging resolution, typically ranging from 0.5 mm to 2 mm. The edge detection algorithm can employ the Canny operator, with its high and low thresholds adaptively determined by analyzing the grayscale histogram of the entire image. The acoustic image feature data block of the defect is a structured record containing the defect ID, the set of boundary pixel coordinates, the equivalent diameter (or major axis / minor axis), area, maximum echo amplitude (in dB), and the calculated acoustic attenuation coefficient (in dB / cm).

[0098] For example, a phased array ultrasound diagnostic system receives a focused scan configuration packet containing {fc:0.85, theta:8.58, pulses:3, target:(2.0, 2.0, 2.0), hash:“a1b2c3d4e5f6…”}. The system first verifies that the hash value “a1b2c3d4e5f6…” exists in its list of pending instructions, and the verification passes. The system moves to the approximate location based on the target coordinates (2.0, 2.0, 2.0), assuming the center coordinates of the transducer array surface are adjusted to (2.0, 2.0, 0). The system sets the sound velocity in concrete. For the target point (2.0, 2.0, 2.0), the focal length is... Using the array element (2.0, 2.0, 0) as the reference point, calculate the delay time of its adjacent array element (2.001, 2.0, 0):

[0099] .

[0100] The controller calculates similar delay times for all array elements and loads them into the transmit timing controller. Subsequently, the system transmits a three-cycle modulated pulse with a center frequency of 0.85 MHz, as specified in the configuration package. The system controls an electronic scan, scanning a square region with sides of 50 mm in 1 mm steps in the x and y directions, acquiring echo data at a corresponding number of points. The image processing unit performs beamforming processing on the echo data to synthesize a C-Scan image. Assuming a bright region is identified near image coordinates (2.0, 2.0), edge extraction and measurement exemplarily yield an equivalent diameter of 8 mm and an area of ​​50.24 mm². 2 The maximum echo amplitude extracted from this region, after calibration, is -12 dB relative to the total reflection surface. The acoustic attenuation coefficient is calculated by analyzing the echo amplitudes at two depth points below this region. Finally, the acoustic image features forming the defect are: {equivalent diameter: 8 mm, area: 50.24 mm²}. 2 Maximum echo amplitude: -12 dB, sound attenuation coefficient: 15 dB / cm.

[0101] In one embodiment of the present invention, step S5 includes the following steps:

[0102] Calculate the correlation coefficient between the maximum echo amplitude in the acoustic image features of the defect and the total energy in the structured acoustic emission event data object that triggered this scan;

[0103] The correlation coefficient is compared with the preset confirmation threshold. If it is greater than the confirmation threshold, it is determined to be a real defect activity. If it is less than or equal to the confirmation threshold, it is determined to be environmental noise or a false signal.

[0104] Based on the judgment result, the risk judgment threshold used to calculate the abnormal deviation degree in step S2 is adjusted using probability statistics update logic.

[0105] Specifically, after receiving the acoustic image features of the defect, the data processing unit initiates a closed-loop verification process. First, the data processing unit extracts the maximum echo amplitude from the acoustic image features of the defect. Simultaneously, using the trigger event hash value carried in the focused scan configuration package as a clue, it reverse-searches and obtains the original structured acoustic emission event data object that triggered this scan. From the structured acoustic emission event data object, the sum of all components of its energy spectral density vector is extracted as the total energy of the acoustic emission event. It should be noted that the total energy of the acoustic emission event is obtained by summing all components of the energy spectral density vector, i.e., the absolute energy value before normalization or the relative energy ratio represented by the normalized value. In this embodiment, the sub-band energy value before normalization is used for summation.

[0106] The data processing unit calculates the Pearson correlation coefficient between the maximum echo amplitude and the total energy of the acoustic emission event, using it as a quantitative indicator for correlation comparison. The system pre-stores a confirmation threshold; if the calculated correlation coefficient is greater than this threshold, the acoustic emission event is determined to be caused by a genuine defect activity, thus completing the confirmation; otherwise, it is determined to be environmental noise or a spurious signal. Pearson Correlation Coefficient The confirmation threshold is typically set between 0.6 and 0.8. This threshold is based on the following: under laboratory conditions, at least 100 sets of paired acoustic emission-ultrasound experiments are conducted on known real defects and known noise sources, respectively, and the correlation coefficient for each set is calculated. (The real defect group...) The values ​​are mainly distributed above 0.7, while the noise group The values ​​are mainly distributed below 0.4. Selecting a threshold of 0.6 to 0.8 can effectively control the false positive rate while ensuring a sufficient detection rate. The initial threshold of the system can be set to 0.7.

[0107] If the activity is confirmed as a genuine defect, the data processing unit performs a model update operation: The energy spectral density vector and Shannon entropy value from the structured acoustic emission event data object are stored as positive sample pairs in the high-risk defect fingerprint database. Simultaneously, the probabilistic statistical update logic is invoked; in this embodiment, a Bayesian update logic is specifically employed. This logic, based on the currently stored historical positive sample data, recalculates the posterior probability distribution of the first risk threshold used for risk assessment, and uses its mean as the new, reduced first risk threshold for subsequent assessments of similar events. In the Bayesian update logic, it is assumed that the first risk threshold... The prior distribution is a normal distribution, with its mean initialized to an empirical value, such as 4.0, and its variance is relatively large. The initial variance is set to 2.0 to reflect the uncertainty of initial perception. Whenever a new positive or negative sample is stored, a likelihood function is constructed based on the relationship between the sample features and the current threshold, and then the likelihood function is calculated. The posterior distribution. and This is the mean of the posterior distribution. The specific Bayesian update process is as follows: assuming a threshold... The prior distribution is For positive samples, it is assumed that their eigenvalues ​​(such as Mahalanobis distance) should be greater than an ideal threshold, and the likelihood function is constructed as follows: ,in The standard normal cumulative distribution function is... Let be the standard deviation of the likelihood function, reflecting the variability of the sample, and set to 1.0. For negative samples, the likelihood function is constructed as follows: The posterior distribution is obtained by updating using Bayes' theorem. New threshold Take immediately Learning rate Used for smooth updates = ( ) + .

[0108] If the signal is confirmed to be noise or a spurious signal, the data processing unit stores the energy spectral density vector and Shannon entropy value from the structured acoustic emission event data object as negative sample pairs in the environmental noise database. Similarly, based on Bayesian update logic, it recalculates the posterior probability distribution of the first risk threshold using the negative sample data and uses its mean as the new, improved first risk threshold. Through this mechanism, the system achieves continuous learning of the background noise and defect signal characteristics of a specific utility tunnel environment, driving the adaptive evolution of the evaluation model's discrimination parameters.

[0109]

[0110] (Positive sample update);

[0111] (Negative sample update);

[0112] The first formula is the Pearson correlation coefficient. The calculation formula is used for correlation comparison. Among them, and Representing the first The total energy of acoustic emission events and the corresponding defect echo amplitude in each confirmation cycle. and These are their means, This refers to the number of historical data pairs used to calculate the correlation coefficient. Typically, a few recent data points are selected, such as 10 to 50, to ensure statistical significance while reflecting recent trends. During initial system startup, the cumulative data pairs are insufficient. At that time, all available data is used for calculation. The second and third formulas describe the first risk threshold. The adaptive update logic. It is the posterior mean of the ideal threshold estimated from all positive samples. It is the posterior mean of the ideal threshold estimated from all negative samples. This is the learning rate, a preset constant ranging from 0.01 to 0.1. It controls the magnitude of each update, preventing excessive adjustments to the threshold from a single sample and ensuring smooth evolution of system parameters. Learning rate A typical initial value is 0.05. An adaptive learning rate strategy can also be used, for example, gradually decreasing the learning rate as the number of samples increases. ,in The initial learning rate, such as 0.1, This represents the total number of samples in the corresponding sample library. Updates with positive samples will lower the threshold. This makes the system more sensitive to similar signals; negative sample updates increase the threshold. This makes the system more conservative in order to reduce false alarms.

[0113] Among them, the high-risk defect fingerprint database and the environmental noise database are both time-series databases. In addition to storing the energy spectral density vector and Shannon entropy value, each record is also associated with its event hash value, timestamp, spatial coordinates and confirmation conclusion.

[0114] For example, the data processing unit receives the acoustic image features of a defect with a maximum echo amplitude of -12dB. The system traces the configuration packet from this scan to find the trigger hash value "a1b2c3d4e5f6…", and then retrieves the corresponding structured acoustic emission event data object. From this data object, the energy spectral density vector is obtained, and the energy values ​​of each sub-band are calculated based on the original energy (assuming they are in arbitrary units [850, 530, 320, 215, 107, 64, 21, 21]). These values ​​are then summed to obtain the total energy of the acoustic emission event. Unit. The system retrieves the most recent few similar confirmation cycles of data, forming... Based on the data pairs, assuming that the energy and echo amplitude sequences of 9 historical pairs are calculated, the current correlation coefficient is obtained. Let the confirmation threshold be 0.7. Since 0.75 > 0.7, the system confirms the event as a genuine defect activity. Next, the system stores the energy spectral density vector, the normalized vector [0.40, 0.25, 0.15, 0.10, 0.05, 0.03, 0.01, 0.01], and the Shannon entropy value of 2.22 as positive samples in the high-risk defect fingerprint database. Assume the current first risk threshold... The posterior mean calculated based on all positive samples Learning rate The updated threshold In fact, due to After calculation Slightly greater than 4.0, but the trend is towards 3.5, and it may decrease in the next update. If the current event is confirmed to be noise, for example... If the sample is negative, it is stored as a negative sample. Assume the posterior mean is based on the negative sample. Then update The trend is towards a higher 4.8, with the actual value changing slowly due to the learning rate. Through this continuous closed-loop verification and parameter updates, the system gradually adapts to the unique acoustic environment of this utility tunnel section.

[0115] See appendix Figure 2 The present invention also proposes a quality evaluation system for concrete engineering of urban underground integrated pipe corridors, comprising the following modules:

[0116] The acoustic emission data structuring module is used to acquire the original acoustic emission signals inside the concrete structure and perform structuring processing on the original acoustic emission signals to generate a structured acoustic emission event data object containing multi-dimensional feature information.

[0117] The risk prediction and command generation module is used to receive structured acoustic emission event data objects, combine them with environmental data for feature correction and multi-dimensional verification, and generate a focused scanning command to be verified, which includes the target's three-dimensional spatial coordinates and risk confidence.

[0118] The scanning protocol reverse generation module generates a parameterized adaptive focused scanning configuration package based on the focused scanning command to be verified through reverse mapping of physical parameters based on energy spectrum characteristics.

[0119] The focused imaging and feature extraction module is used to drive the phased array ultrasound diagnostic system to perform targeted imaging of the target area and extract the acoustic image features of the defects in response to the focused scanning configuration package.

[0120] The verification and model adaptation module is used to compare the correlation between the structured acoustic emission event data object and the acoustic image features of the defect, perform closed-loop verification, and adaptively update the discrimination parameters used for risk determination in step S2.

[0121] Each of the modules can be implemented in whole or in part through software, hardware, or a combination thereof. It supports hardware embedded in or independent of the processor in the computer device, and also supports software stored in the memory of the computer device, so that the processor can call and execute the operations corresponding to each of the above modules.

[0122] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A method for evaluating the quality of concrete engineering in urban underground utility tunnels, characterized in that, Includes the following steps: S1. Obtain the original acoustic emission signal inside the concrete structure, perform wavelet packet transform on the original acoustic emission signal, decompose it into a preset number of energy sub-bands, and calculate the normalized energy value of each energy sub-band to construct an energy spectral density vector; calculate the Shannon entropy value, which characterizes the signal complexity, based on the energy distribution probability in the energy spectral density vector; use the time difference of arrival (TDOA) positioning algorithm to solve the source three-dimensional spatial coordinates of the location where the original acoustic emission signal occurred, and generate a unique event identifier with unique identification function based on the original waveform data; The three-dimensional spatial coordinates of the earthquake source, the energy spectral density vector, the Shannon entropy value, and the unique identifier of the event are encapsulated to generate a structured acoustic emission event data object containing multi-dimensional feature information. S2. Receive structured acoustic emission event data objects, combine them with environmental data to perform feature correction and multi-dimensional verification, and generate a focused scan command to be verified, which includes the target's three-dimensional spatial coordinates and risk confidence. S3. Based on the focused scanning command to be verified, a parameterized adaptive focused scanning configuration package is generated through inverse mapping of physical parameters based on energy spectrum characteristics. Specifically, step S3 includes: extracting the main frequency band energy value from the structured acoustic emission event data object corresponding to the focused scanning command to be verified, and mapping it to the target detection center frequency through a preset physical inversion function; selecting the pulse width from a preset mapping table according to the risk confidence in the focused scanning command to be verified, and calculating the scanning aperture angle that can cover the error area according to the positioning error radius; encapsulating the target detection center frequency, pulse width, scanning aperture angle, and the event unique identifier in the corresponding data object to generate the focused scanning configuration package. S4. In response to the focused scanning configuration package, drive the phased array ultrasound diagnostic system to perform targeted imaging of the target area and extract the acoustic image features of the defects. S5. Compare the correlation between the structured acoustic emission event data object and the acoustic image features of the defect, perform closed-loop verification, and adaptively update the discrimination parameters used for risk determination in step S2.

2. The method for evaluating the quality of concrete engineering in urban underground utility tunnels according to claim 1, characterized in that, Step S2 includes the following steps: Acquire local concrete temperature data and overall structural strain data that are synchronized with the structured acoustic emission event data object in time, and perform weighted correction on the energy spectral density vector in the data object; The corrected energy spectral density vector and the Shannon entropy value in the data object are compared with the preset background noise model to calculate the abnormal deviation of the current event. The trend of the change in the slope of the rising edge of the signal of acoustic emission events that occur continuously within a preset time window is detected. When the abnormal deviation exceeds the first risk threshold and the slope of the rising edge of the signal shows an overall upward trend, it is judged as a high-risk event, and a focused scan command to be verified is generated.

3. The method for evaluating the quality of concrete engineering in urban underground utility tunnels according to claim 1, characterized in that, Step S4 includes the following steps: After verifying that the unique event identifier in the focused scan configuration package matches the pre-stored focused scan command to be verified, the emission delay time of each element of the two-dimensional array transducer is calculated according to the parameters in the configuration package. A two-dimensional array transducer is driven to emit a focused ultrasonic beam and receive echo signals to synthesize an acoustic tomographic image of the target area. Abnormal response regions are identified from acoustic tomographic images, and the equivalent diameter, maximum echo amplitude, and acoustic attenuation coefficient of these regions are calculated to constitute the acoustic image features of the defects.

4. The method for evaluating the quality of concrete engineering in urban underground integrated pipe gallery according to claim 1, characterized in that, Step S5 includes the following steps: Calculate the correlation coefficient between the maximum echo amplitude in the acoustic image features of the defect and the total energy in the structured acoustic emission event data object that triggered this scan; The correlation coefficient is compared with the preset confirmation threshold. If it is greater than the confirmation threshold, it is determined to be a real defect activity. If it is less than or equal to the confirmation threshold, it is determined to be environmental noise or a false signal. Based on the judgment result, the risk judgment threshold used to calculate the abnormal deviation degree in step S2 is adjusted using probability statistics update logic.

5. The method for evaluating the quality of concrete engineering in urban underground integrated pipe gallery according to claim 2, characterized in that, The weighted correction of the energy spectral density vector is specifically performed by calling a preset environmental impact compensation model, calculating the signal attenuation coefficient caused by environmental temperature changes and structural deformation, and using the signal attenuation coefficient to compensate the original energy spectral density vector in order to eliminate non-damaging signal drift introduced by environmental factors.

6. The method for evaluating the quality of concrete engineering in urban underground integrated pipe gallery according to claim 1, characterized in that, The physical inversion function is configured to establish a cube root mapping relationship between the normalized energy value of the main frequency band of the acoustic emission signal and the center frequency of the ultrasonic target detection, so that the calculated target detection center frequency matches the physical scale of the potential defect.

7. The method for evaluating the quality of concrete engineering in urban underground integrated pipe gallery according to claim 3, characterized in that, The acoustic attenuation coefficient is calculated by estimating the logarithmic attenuation rate of the echo signal at different depths below the abnormal response area. The depth range is selected to avoid near-field interference and bottom reflection areas.

8. The method for evaluating the quality of concrete engineering in urban underground integrated pipe gallery according to claim 4, characterized in that, Adjusting the risk assessment threshold specifically includes: If the activity is determined to be a genuine defect, the structured acoustic emission event data object will be stored in the high-risk defect fingerprint database as a positive sample to lower the risk judgment threshold. If the signal is determined to be environmental noise or a spurious signal, the structured acoustic emission event data object is stored in the noise database as a negative sample to increase the risk assessment threshold.

Images