Tunnel lining defect determination method based on detection process self-consistency constraints

Abstract

The application provides a tunnel lining defect judgment method based on detection process self-consistency constraint, and the method comprises the following steps: collecting reference detection response data and disturbance detection response data on the same detection path respectively, and generating corresponding response sequences after unified pretreatment; constructing a self-consistency measurement sequence by position alignment correction, in combination with a weighted center value, a disturbance response discrete degree and spatial continuity; and adaptively generating a judgment threshold based on the sequence, so as to realize identification of an abnormal position and aggregated output of a defect section. The scheme does not need to rely on historical samples or manual calibration, and the robustness and automation level of defect judgment are improved.

Description

Technical Field

[0001] This invention belongs to the field of tunnel engineering inspection, and particularly relates to a method for determining tunnel lining defects based on self-consistency constraints of the inspection process. Background Technology

[0002] Tunnel lining, as a key component in the tunnel structural system directly bearing the requirements of surrounding rock pressure, waterproofing, protection, and durability, is prone to defects such as voids, cavities, cracks, and interlayer separation. These defects are typically well-hidden, have inconspicuous early physical responses, and evolve slowly. If not identified in time during the operational period, they can easily expand under long-term loads, environmental erosion, and traffic dynamics, leading to leakage, lining deterioration, and even structural safety risks. Therefore, the engineering field has long employed non-destructive testing methods such as ultrasonic testing, ground-penetrating radar testing, and impact-echo testing. These methods apply excitation to the lining surface and collect response signals, using echo amplitude, arrival time, or spectral characteristics to determine the internal structural state. However, in actual tunnel engineering environments, the test results are not only affected by the structural state but also highly sensitive to testing conditions. Factors such as the coupling state between the testing device and the lining surface, lining surface roughness, moisture content changes, testing path orientation, and equipment parameter settings can all significantly alter the test response characteristics without introducing actual structural defects, leading to misjudgments or missed diagnoses. In existing technologies, some methods rely on fixed empirical thresholds or manual interpretation to distinguish between normal and abnormal states, making it difficult to adapt to complex and ever-changing field conditions. Other methods attempt to introduce multiple tests or multi-parameter analysis, but these often remain at the level of comparing the test results themselves, lacking systematic constraints on whether the testing process remains stable and consistent under different conditions. Furthermore, judgment methods based on sample training are limited in tunnel engineering by difficulties in obtaining defect samples, uneven distribution of defect types, and significant differences in engineering conditions, hindering reliable generalization. Overall, existing technologies focus more on the anomalies of single or few test results, and less on analyzing whether the testing process itself can stably reflect the structural state under multiple changes in testing conditions, making it difficult to effectively guarantee the reliability and stability of test conclusions under complex field conditions. Summary of the Invention

[0003] The purpose of this invention is to design a tunnel lining defect determination method based on the self-consistency constraint of the detection process. This method can transform the self-consistency metric of the detection process into the determination result of tunnel lining defects, avoid dependence on external defect samples or fixed empirical thresholds, and thus achieve robust identification of tunnel lining defects in complex engineering environments.

[0004] To achieve the above objectives, this invention provides a method for determining tunnel lining defects based on self-consistency constraints in the detection process, the method comprising:

[0005] Reference detection response data are collected along the same detection path of the tunnel lining, and the reference detection response data are subjected to location discretization, local response segment extraction, scale unification and convergence processing to generate a reference detection response sequence.

[0006] While keeping the detection path, detection method, and data format unchanged, perturbation detection is performed on the detection path to obtain perturbation detection response data; the same processing procedure as the baseline detection response data is performed on the perturbation detection response data to generate a perturbation detection response sequence.

[0007] The disturbance detection response sequence is aligned and corrected with the reference detection response sequence.

[0008] Based on the corrected disturbance detection response sequence, a weighted center value is calculated at each detection location, and a self-consistency measurement sequence for the detection process is constructed by combining the dispersion of the disturbance response with the spatial continuity along the path.

[0009] Based on the self-consistency measurement sequence of the detection process, a self-reference judgment threshold is constructed within the entire detection path. Anomalies are judged at each detection position according to the self-reference judgment threshold, and continuous abnormal positions are aggregated into defect segments for output.

[0010] Furthermore, the reference detection response data and disturbance detection response data are either ultrasonic detection time-series response or ground-penetrating radar echo response sequence.

[0011] Furthermore, the local response segment interception refers to intercepting continuous sampled values ​​within a fixed length range before and after each detection position, and this fixed length is set before the detection operation begins and remains consistent throughout the entire detection process.

[0012] Furthermore, the scale uniformity refers to dividing each sampled value within a local response segment by the maximum absolute value of the sampled values ​​within that local response segment to eliminate overall amplitude differences.

[0013] Furthermore, the disturbance detection is achieved by switching the excitation frequency band, center frequency band, or acquisition gating window of the detection device, and the disturbance detection is performed three times.

[0014] Furthermore, the position alignment correction is performed by calculating the discrete cross-correlation function between the disturbance detection response sequence and the reference detection response sequence, introducing an offset magnitude penalty term, estimating the overall position offset, and then indexing the disturbance detection response sequence.

[0015] Furthermore, the weight of the weighted center value is determined by the overall position offset of the corresponding disturbance detection response sequence through an exponential decay function; the larger the offset, the smaller the weight.

[0016] Furthermore, the self-consistency measure of the detection process is composed of a weighted sum of a discrete term of the disturbance response and a spatial continuity term, wherein the discrete term of the disturbance response characterizes the degree of response fluctuation at the same detection location under different disturbance conditions, and the spatial continuity term characterizes the second-order difference of the weighted central values ​​of adjacent detection locations.

[0017] Furthermore, the self-reference judgment threshold is calculated based on the median value of the self-consistency measurement sequence of the detection process and the median value of its absolute deviation, and multiplied by a preset adjustment coefficient.

[0018] Furthermore, the output of the defective section must satisfy the requirement that the number of consecutive abnormal positions is not less than the preset minimum section length, and abnormal positions that do not reach the minimum section length are recorded separately as suspicious positions.

[0019] The beneficial technical effects of the present invention are at least as follows:

[0020] To address the aforementioned problems, this invention provides a tunnel lining defect determination method based on self-consistency constraints in the detection process. Using the same detection path as the basic object, it constructs a benchmark detection response with aligned positions and introduces multiple perturbation detections while maintaining consistency between the detection path and data format. This allows the detection process to form comparable response sequences under different detection conditions. Based on this, by analyzing the dispersion of the detection response at the same location under multiple perturbation conditions and its spatial continuity along the path, a self-consistency metric for the detection process is constructed, thereby characterizing whether the detection process still possesses the ability to stably represent the structural state at that location. When the lining structure is intact, the detection process exhibits strong consistency in the response at the same location under different perturbation conditions and maintains a continuous spatial distribution. However, when structural anomalies such as voids or cavities exist, changes in local structural characteristics amplify the impact of changes in detection conditions on the response, significantly degrading the consistency of the detection process and forming identifiable abnormal sections within the path. This invention further transforms the self-consistency measurement of the detection process into the judgment result of tunnel lining defects through the in-path self-reference judgment method, avoiding dependence on external defect samples or fixed empirical thresholds, thereby achieving robust identification of tunnel lining defects in complex engineering environments and improving the reliability and engineering applicability of the detection conclusions. Attached Figure Description

[0021] The present invention will be further described with reference to the accompanying drawings, but the embodiments in the drawings do not constitute any limitation on the present invention. For those skilled in the art, other drawings can be obtained based on the following drawings without creative effort.

[0022] Figure 1 This is a flowchart of the tunnel lining defect determination method based on the self-consistency constraint of the detection process according to the present invention. Detailed Implementation

[0023] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0024] In one or more embodiments, such as Figure 1 As shown, a method for determining tunnel lining defects based on self-consistency constraints in the detection process is disclosed. The method includes the following:

[0025] S1: Collect benchmark detection response data on the same detection path of the tunnel lining, and perform position discretization, local response segment extraction, scale unification and convergence processing on the benchmark detection response data to generate a benchmark detection response sequence;

[0026] Specifically, in tunnel lining inspection, the inspection device moves continuously along a predetermined inspection path on the lining surface and collects inspection response data according to a fixed sampling rhythm. The inspection response data is of a single type, such as a time-series response continuously acquired over time in an ultrasonic inspection scenario, or an echo response sequence continuously acquired along the inspection path in a radar inspection scenario.

[0027] Because the detection device moves continuously along the path, the original detection response typically appears as a continuous sequence that varies with the path position. If the response of a single sampling point is directly used as a representation of a specific location, it is easily affected by instantaneous changes in coupling state, differences in local surface conditions, or occasional noise, thus weakening the representativeness of the response at that location to the true structural state. Therefore, this step first discretizes a series of detection positions along the detection path at fixed spatial intervals. For each detection position, a continuous segment of samples adjacent to that position is extracted from the original detection response sequence to form a local detection response segment corresponding to that position. The length of this local detection response segment is determined before detection begins and remains consistent throughout the entire detection process to ensure uniformity in the processing methods across different locations.

[0028] After obtaining the local detection response segment, to ensure comparability of numerical scales across different detection locations, this step performs scale unification processing on the sampled values ​​within the local detection response segment. Specifically, for the first... Each detection location contains a corresponding local detection response segment. A series of consecutive sampled values, denoted as Each of them All data originates directly from the raw detection responses continuously output by the detection equipment near that location. First, a scaling factor is determined within this local detection response segment. This scaling factor is taken as the largest absolute value of the sampled values ​​within that local detection response segment, denoted as [missing value]. Subsequently, each sampled value within the local detection response segment is divided by the scaling factor to eliminate the impact of the overall amplitude difference in the response at that location on subsequent analysis.

[0029] After scaling, to obtain a single numerical value that represents the structural characteristics of the detection location, this step performs convergence processing on the local detection response segments after scaling. This convergence processing involves averaging all sampled values ​​within the local detection response segment, thereby preserving the overall characteristics of the local response while suppressing random fluctuations at individual sampling points. The corresponding calculation process is expressed as follows:

[0030] ;

[0031] in, Indicates the first step on the detection path The baseline detection response value corresponding to each detection location; Indicates the first The first sample collected near the detection location Each detection response sample value; This represents the largest absolute value of the sampled value within the local detection response segment corresponding to the detection location; This indicates the number of sampling points constituting the local detection response segment. Through the above calculations, all sampled values ​​within the local detection response segment are aggregated into a scalar form of the baseline detection response value on a uniform scale.

[0032] To facilitate understanding of the above calculation process, consider the following calculation example: Assume that the local detection response segment at a certain detection location is obtained from continuous acquisition. Composed of 1 sample value, respectively Then the largest absolute value of the sampled values ​​within this local detection response segment is... After scaling each sample value, we obtain... The average of the above results is then taken to obtain the baseline detection response value for that detection location. If, under another detection condition, the overall detection response at that location exhibits a proportionally amplified form of the above value, the baseline detection response value obtained after the same processing remains unchanged, thus ensuring that the baseline response primarily reflects structural characteristics and does not change with overall amplitude variations caused by detection conditions. Following this method, the corresponding baseline detection response values ​​are calculated for all detection locations along the detection path, and arranged according to the spatial order of the detection path to form a complete baseline detection response sequence, denoted as... ,in This represents the number of discrete detection locations along the detection path. This benchmark detection response sequence spatially and continuously reflects the response distribution characteristics of the tunnel lining structure along the detection path. It also exhibits numerical stability and comparability, providing a unified and clear reference basis for subsequently introducing disturbance detection along the same path and conducting self-consistency analysis of the detection process.

[0033] S2: While keeping the detection path, detection method, and data format unchanged, perform disturbance detection on the detection path to obtain disturbance detection response data; perform the same processing procedure as the baseline detection response data on the disturbance detection response data to generate a disturbance detection response sequence;

[0034] Specifically, in tunnel lining inspection, this step uses the baseline inspection response sequence obtained in the previous stage. Using a reference object, multiple disturbance detections are performed along the same detection path to construct a set of response sequences that reflect changes in the detection process. Here, the disturbance detection does not change the detection path or data type, but rather maintains consistency in the detection path, detection method, and output data format, achieved solely through preset level switching of the excitation or acquisition configuration of the detection device. It plays a dual role throughout the process: on the one hand, it is used to lock the discrete set and order of detection locations, so that multiple detections are mapped to the same location. On the one hand, it serves as a position; on the other hand, it serves as a reference for subsequent alignment and correction, enabling the results of each disturbance detection to be compared under the same position indexing system.

[0035] In practice, the detection device repeatedly scans along a detection path that is completely consistent with the benchmark detection. Disturbances are generated through pre-set settings within the device; for example, switching between different excitation frequency bands or configurations in ultrasonic testing scenarios, or switching between different center frequency bands or acquisition gating windows in radar testing scenarios. These settings are configured before the detection operation begins; on-site, only the settings need to be selected sequentially and the entire path scanned. During each scan, the detection device continuously acquires raw detection response data, forming a response sequence that continuously changes with the advancement position, and its data type is completely consistent with that of the benchmark detection.

[0036] After each perturbation scan, the continuous raw response obtained from that scan needs to be converted into a sequence with respect to the original response obtained from that scan. Consistent position-level response sequences This conversion process follows the basic idea of ​​forming the reference response in the previous stage, that is, discretely indexing the detection positions along the detection path according to the same position, and extracting a local response segment of fixed length near each detection position. Let the first... Secondary disturbance detection at location The nearby captured local response segment contains A series of consecutive sampled values, denoted as These sampled values ​​are all directly derived from the raw output of the detection device during this scan. Subsequently, the local response segment is first subjected to scale unification processing, and then converged processing to obtain the value at position [number missing]. Response value under secondary disturbance detection The calculation method is as follows:

[0037] ;

[0038] in, This represents the largest absolute value of the sampled values ​​within the local response segment, used to standardize the overall amplitude of the local response at that location. This indicates the number of sampling points contained in a local response segment. Its value is set before the start of this testing operation and remains consistent throughout the entire process. Since the scale term is composed of sampled values ​​within the same local response segment, its result is a dimensionless quantity, and after averaging... With reference sequence Maintaining consistency in numerical attributes ensures direct comparison of responses under different detection conditions. Arranged in order of detection path. to You can get the first one. The response sequence corresponding to the secondary disturbance detection .

[0039] When conducting multiple scans at the tunnel site, a common phenomenon is that there are overall deviations in the starting point or advancement rhythm between different scans, leading to... and A global shift occurs in the position index. Without processing, this global shift could easily be mistaken for a difference in structural response during subsequent comparisons. Therefore, this step utilizes... and The sequence matching relationship between them estimates an overall positional offset. And based on this, Correction is performed. The offset estimation is based on the discrete cross-correlation concept. It calculates the alignment score of two sequences under different candidate offsets and adds a penalty term for the offset magnitude to make the estimation result more consistent with the reality that the on-site scanning offset is usually small. The calculation form is as follows:

[0040] ;

[0041] in, This represents the overall offset of the candidate. This indicates the maximum allowable offset range for the search, which is determined before the detection operation begins based on an empirical range of path marking accuracy or equipment advancement error. This is a penalty coefficient used to suppress excessively large candidate offsets, ensuring that the final selected candidate is... This is more in line with engineering practice. The first term in the formula is the sequence alignment score, reflected in the candidate offset. The similarity between the next two sequences is calculated, with the second term being a regularization penalty term, used to prioritize solutions with smaller offsets when alignment scores are close. Because... and All are dimensionless quantities, and their product summation and penalty term are numerically comparable, ultimately yielding... This represents the overall translation amount at the location index level. According to this... right After index correction, the disturbance response sequence and the reference sequence can be aligned in the same indexing system.

[0042] To illustrate the feasibility of the above calculation process, specific numerical values ​​can be used as examples. For instance, at a certain detection location... , No. The local response segment captured by the secondary disturbance detection is taken as follows: Each sample value is... The maximum absolute value of this local response segment is then... The sampled values ​​were obtained after scaling. We can obtain the average of them. For example, in overall offset estimation, if , Search range Penalty coefficient The alignment score and penalty term can be calculated separately for different candidate offsets, and the final result is obtained. Thus Overall forward correction of a position index, so that its main response peak is consistent with Alignment.

[0043] Through the above process, multiple disturbance scans performed along the same detection path are converted into a set of... Perturbation detection response sequence that is strictly corresponding to the location index This also yields the overall offset set corresponding to each perturbation scan. These results provide direct and usable sequence input for subsequent steps to analyze the consistency of responses under different detection conditions at the same detection location.

[0044] S3: Perform position alignment correction between the disturbance detection response sequence and the reference detection response sequence; based on the corrected disturbance detection response sequence, calculate the weighted center value at each detection position, and combine the dispersion of the disturbance response with the spatial continuity along the path to construct a self-consistency measurement sequence for the detection process;

[0045] Specifically, this step involves multiple perturbation detection response sequences output from the previous stage. and the overall offset set As input, a self-consistency constraint for the detection process is constructed under the same detection path and the same location indexing system. For tunnel lining, two types of observable phenomena will appear during repeated scanning: first, whether the response at the same location under different disturbance conditions fluctuates stably around a certain central value; second, whether the response distribution along the detection path maintains spatial continuity. Based on these two phenomena, this step incorporates both the degree of dispersion within the disturbance and spatial continuity into the consistency metric, and includes the overall offset set... Converting to sequence-level confidence weights allows perturbation sequences with higher alignment quality to contribute more to the consensus assessment.

[0046] The construction of a consistency measure begins with determining the "weighted central value". For the first... There are detection locations, and the disturbance detection response set is as follows: ,in From sequence In mathematics, the weighted average is a classic form of estimating the central value, derived from the estimation of the mean in the sense of least squares; based on this, the entire sequence is shifted. Convert to weights The idea originates from the common practice in regularization and robust estimation that "the greater the deviation, the lower the confidence level." The squared penalty corresponds to the common quadratic energy form, and the exponential mapping corresponds to the common form of converting energy into confidence level. Weights are defined accordingly. And obtain the location Weighted central value :

[0047] ;

[0048] in, For position The weighted central value; For the first Secondary disturbance detection at location The response value comes from ; For the first Weights of the perturbation sequence; From ; This is the weight attenuation coefficient, set before operation, and taken as a positive value. The preferred range is 0.1 to 2, and 0.5 can be used initially (increase when path control accuracy is high, decrease when path deviation is large). Because... and Calculated from the dimensionless response value. Since it is generated by an exponential function and is a pure number, the numerical properties of both sides of the expression are consistent, which is reasonable.

[0049] In obtaining Post-construction location-level detection process self-consistency measurement This metric comprises two terms: the first is an intra-perturbation discrete term, derived from the weighted variance form in statistics, used to describe the degree of dispersion around the central value at the same location under multiple perturbations; the second is a spatial continuity term, derived from the second-order difference (discrete curvature) penalty in discrete signal processing, used to describe the degree of local bending along the tunnel lining detection path, making local abrupt changes more prominent in the index. The two terms are added together to form a self-consistency metric.

[0050] ;

[0051] in, For position A measure of self-consistency; and The result is derived from the calculation in the previous equation; This is a balancing coefficient, set before operation, and taken as a positive value. The preferred range is 0.1 to 1, and it can be initially set to 0.5 (increased when spatial continuity along the path is emphasized, and decreased when the dispersion of multiple disturbance responses is emphasized). It is used to adjust the influence of the disturbance discrete term and the spatial continuity term. The logical relationship of this formula is: first calculate... , then calculate Finally Substituting back into the original expression yields These are two steps within the same deductive chain. Both items are composed of the squared differences of dimensionless response values, combined through weighted summation and addition. Therefore... It is a dimensionless quantity with consistent numerical properties. For the endpoints of the path, the curvature term can be calculated by substituting adjacent positions on one side, making the calculation process executable throughout the entire path.

[0052] The output of this step is the self-consistency measurement sequence of the detection process. The sequence consists of the input... and The above two equations are derived position by position. The calculation process consists of basic operators such as weighted average, weighted dispersion and discrete second-order difference, which is easy to implement in the detection equipment or post-processing software, and provides direct quantitative input for the next stage of defect judgment based on self-consistency constraints.

[0053] S4: Based on the self-consistency measurement sequence of the detection process, construct a self-reference judgment threshold within the entire detection path, and make anomaly judgments for each detection position according to the self-reference judgment threshold, and aggregate continuous abnormal positions into defect segments for output.

[0054] Specifically, this step involves using the self-consistency measurement sequence of the detection process obtained in the previous stage. Using the same detection path and location indexing system as input, the tunnel lining defect judgment and result output are completed. The tunnel lining exhibits a continuous distribution characteristic along the path. Under normal structural conditions, the self-consistency metric obtained from multiple disturbance detections typically forms a relatively concentrated level within the path range. However, when voids, cavities, or interface anomalies exist, the consistency of the detection process at local locations under multiple disturbances will significantly degrade, thus affecting the overall performance. The sequence shows a significant rise. Based on this engineering phenomenon, this step adopts an in-path self-referencing determination method to transform the self-consistency constraint of the detection process into location-level and segment-level defect determination results.

[0055] In practical implementation, an adaptive decision threshold is first constructed across the entire detection path. This threshold is derived from the concept of quantiles in robust statistics, where the median characterizes the typical consistency level within the path, and the median of the absolute deviation characterizes the general fluctuation scale within the path. By combining these two, the stability of the threshold can be maintained even with a small number of outliers, and the decision criterion can adaptively change with different detection tasks. An adjustment coefficient is then introduced based on this. The value is positive, preferably ranging from 1 to 5. It can be initially set to 3 (to be appropriately decreased when increased defect identification sensitivity is needed, and appropriately increased when the influence of occasional anomalies on the judgment result is needed). This value is used to adjust the judgment sensitivity based on engineering experience, thereby forming a self-reference threshold within the path. The calculation method is as follows:

[0056] ;

[0057] in, Represents a sequence The median value reflects the typical level of consistency in the detection process within the path; express The median value, relative to the absolute deviation of its median value, is used to characterize the general fluctuation range within the path; This is an adjustment coefficient, set before the detection operation begins, used to control the amplification of the threshold relative to the typical fluctuation range. Because... As a dimensionless consistency measure, the terms in this formula must maintain consistency in their numerical properties, resulting in a threshold. It is also a dimensionless quantity, which conforms to the conventional numerical judgment logic.

[0058] After obtaining the path self-reference threshold Then, for each location on the detection path Perform position-by-position determination. When Greater than When the consistency of the detection process at this location under multiple disturbance conditions is significantly lower than the typical level within the path, this location is marked as an anomaly; when Less than or equal to When anomalies occur, the location is marked as a normal location. Considering that tunnel lining structure anomalies usually have a certain spatial extension rather than appearing completely isolated, this step further performs spatial aggregation processing on the anomaly locations: scanning the anomaly markers along the detection path, merging consecutively occurring anomaly locations into candidate anomaly segments, and calculating the length of each segment; when the length of a candidate segment is not less than the preset minimum segment length, the segment is output as a tunnel lining defect segment; if only a single or extremely short anomaly location appears, it is recorded as a suspicious location for subsequent manual review. This aggregation method directly utilizes the engineering characteristic of the continuous distribution of the lining structure along the path, making the final output results more consistent with the usage habits of on-site inspection and maintenance decisions.

[0059] This invention also provides a tunnel lining defect determination device based on self-consistency constraints of the detection process, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the steps described in the above embodiments of the tunnel lining defect determination method based on self-consistency constraints of the detection process. Figure 1 The steps S1 to S4 described above; or, when the processor executes the computer program, it implements the functions of each module in the above system embodiments.

[0060] For example, the computer program may be divided into one or more modules, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules may be a series of computer program instruction segments capable of performing specific functions, which describe the execution process of the computer program in the tunnel lining defect determination device based on self-consistency constraints of the detection process.

[0061] The tunnel lining defect determination device based on self-consistency constraints of the detection process can be a computing device such as a desktop computer, laptop, handheld computer, or cloud server. The device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that the device may also include input / output devices, network access devices, buses, etc.

[0062] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the tunnel lining defect determination device based on self-consistency constraints of the detection process, connecting all parts of the device via various interfaces and lines.

[0063] The memory can be used to store the computer program and / or modules. The processor implements various functions of the tunnel lining defect judgment device based on the self-consistency constraint of the detection process by running or executing the computer program and / or modules stored in the memory and calling the data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function, etc.; the data storage area may store data created according to the operation of the air conditioning controller, etc. In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital card (SD card), flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0064] The module integrated into the tunnel lining defect determination device based on the self-consistency constraint of the detection process, if implemented as a software functional unit and sold or used as an independent product, can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0065] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0066] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for determining tunnel lining defects based on self-consistency constraints in the detection process, characterized in that, The method includes: Benchmark detection response data are collected along the same detection path of the tunnel lining. The benchmark detection response data is then processed by location discretization, local response segment extraction, scale unification, and convergence to generate a benchmark detection response sequence. The benchmark detection response data and disturbance detection response data are either ultrasonic detection time-series response or ground-penetrating radar echo response sequence. The convergence processing involves averaging all sampled values ​​within a local detection response segment. While keeping the detection path, detection method, and data format unchanged, disturbance detection is performed on the detection path to obtain disturbance detection response data; the disturbance detection response data is processed in the same way as the reference detection response data to generate a disturbance detection response sequence; the disturbance detection is achieved by switching the excitation frequency band, center frequency band, or acquisition gating window of the detection device, and the disturbance detection is performed three times; The disturbance detection response sequence is aligned and corrected with the reference detection response sequence. Based on the corrected disturbance detection response sequence, a weighted center value is calculated at each detection location. Combining the dispersion of the disturbance response with the spatial continuity along the path, a self-consistency metric sequence for the detection process is constructed. This self-consistency metric is composed of a weighted sum of a discrete term and a spatial continuity term. The discrete term represents the degree of response fluctuation at the same detection location under different disturbance conditions, and the spatial continuity term represents the second-order difference between the weighted center values ​​of adjacent detection locations. The self-consistency metric for the detection process is calculated as follows: ; For position The self-consistency measure of the detection process; This is the balance coefficient; For position The weighted central value; For the first Secondary disturbance detection at location The response value; For the first Weights of the perturbation sequence; From the global offset set ; This is the weight decay coefficient; Based on the self-consistency measurement sequence of the detection process, a self-reference judgment threshold is constructed within the entire detection path. Anomalies are judged at each detection position according to the self-reference judgment threshold, and continuous abnormal positions are aggregated into defect segments for output.

2. The method for determining tunnel lining defects based on self-consistency constraints in the detection process according to claim 1, characterized in that, The local response segment interception refers to intercepting continuous sampled values ​​within a fixed length range before and after each detection position, and this fixed length is set before the detection operation begins and remains consistent throughout the entire detection process.

3. The method for determining tunnel lining defects based on self-consistency constraints in the detection process according to claim 1, characterized in that, The scaling uniformity refers to dividing each sampled value within a local response segment by the maximum absolute value of the sampled values ​​within that local response segment, in order to eliminate overall amplitude differences.

4. The method for determining tunnel lining defects based on self-consistency constraints in the detection process according to claim 1, characterized in that, The position alignment correction is performed by calculating the discrete cross-correlation function between the disturbance detection response sequence and the reference detection response sequence, introducing an offset magnitude penalty term, estimating the overall position offset, and then indexing the disturbance detection response sequence.

5. The method for determining tunnel lining defects based on self-consistency constraints in the detection process according to claim 1, characterized in that, The weight of the weighted center value is determined by the overall position offset of the corresponding disturbance detection response sequence through an exponential decay function; the larger the offset, the smaller the weight.

6. The method for determining tunnel lining defects based on self-consistency constraints in the detection process according to claim 1, characterized in that, The self-reference judgment threshold is calculated based on the median value of the self-consistency measurement sequence of the detection process and the median value of its absolute deviation, and multiplied by a preset adjustment coefficient.

7. The method for determining tunnel lining defects based on self-consistency constraints of the detection process according to claim 1, characterized in that, The output of the defective section must satisfy the requirement that the number of consecutive abnormal positions is not less than the preset minimum section length, and abnormal positions that do not reach the minimum section length are recorded separately as suspicious positions.

Images