Ground penetrating radar intelligent identification method and system for crack detection

Abstract

The invention relates to a ground penetrating radar intelligent identification method and system for crack detection, and the method comprises the steps: constructing angle sequence data through extracting travel time and amplitude attributes of all azimuth angles; based on a double-branch separation phenomenon of a co-polarization main reflection branch, splitting and marking the azimuth angle, and screening a stable section structure; calculating the main reflection time difference of arrival between the azimuth angles, constructing a two-dimensional difference chart, detecting a main ridge chain, and converting an overlapped coordinate sequence of the main ridge chain and a stable section into a direction angle and a coverage length to form a crack trend indication quantity; the method comprises the following steps: firstly, calculating a crack trend indication quantity, screening a significant trend with a large coverage length from the crack trend indication quantity, calculating a comprehensive direction credibility score of each stable section in combination with direction consistency, and selecting the highest score as a main crack trend, so that a crack trend result with direction continuity and space coverage can be output in a complex underground structure environment. And the stability and accuracy of crack identification are improved.

Description

Technical Field

[0001] This invention relates to the field of crack detection technology, and specifically to a ground-penetrating radar intelligent identification method and system for crack detection. Background Technology

[0002] Ground-penetrating radar (GPR), as an important non-destructive testing (NDT) technology, has gradually formed a systematic technical system in fields such as geological exploration, road structure assessment, underground pipeline detection, and concrete structure defect detection. Existing crack detection methods mainly rely on the energy characteristics of co-polarized echo waveforms, time delay differences, or reflection intensity distribution based on single-directional scanning to characterize discontinuous structures within the medium. Some studies utilize combinations of two-dimensional radar profiles to form directional structure maps to improve crack orientation analysis capabilities; simultaneously, polarization difference analysis is used to identify scattering differences at medium interfaces to distinguish reflection types and scatterers. However, traditional methods often emphasize single-polarization or single-directional strategies, failing to fully utilize the variation characteristics of crack reflection at different azimuth angles, making it difficult to support the continuous orientation identification of complex crack structures. Furthermore, current crack detection methods based on energy thresholds or geometric fitting rely heavily on manually set parameters, limiting their analytical capabilities for multi-source reflection interference, reflection branching phenomena, and directional continuity judgment, resulting in identification accuracy being significantly affected by noise, scattering paths, and medium complexity.

[0003] Existing technologies in crack orientation analysis have several limitations. Firstly, traditional profile inspection methods cannot construct continuous structural segment information from multiple azimuth angles, making it difficult to identify reflection branching characteristics that change with azimuth angle. Secondly, most methods only calculate the difference in reflection amplitude, ignoring the cross-angle correlation of the arrival time difference of the main reflection between different azimuth angles, thus failing to establish a complete difference matrix reflecting the crack orientation distribution. Furthermore, existing crack orientation calculation methods largely rely on overall fitting or unified orientation estimation, lacking mechanisms for screening stable angle ranges and evaluating orientation reliability, making them prone to orientation shifts or misjudgments in the presence of local noise interference or multi-crack intersection structures. Moreover, when identifying orientations based on difference maps, existing studies rarely construct quantitative indicators from the perspective of orientation ridge structures and coverage length, resulting in identification results that fail to reflect the significance and consistency of crack orientation. Summary of the Invention

[0004] This invention proposes a ground-penetrating radar intelligent identification method and system for crack detection, in order to solve the technical problems in the background art.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: A ground-penetrating radar intelligent identification method for crack detection according to the present invention includes the following steps: S1. Collect the original waveform data of common polarization and cross polarization in azimuth angle, and extract the common polarization travel time attribute data, common polarization amplitude attribute data, cross polarization travel time attribute data and cross polarization amplitude attribute data; S2. Based on the common polarization travel time attribute data and the common polarization amplitude attribute data, the azimuth angle is marked with a split mark amount, and adjacent azimuth angles with consecutive split marks are grouped to construct a stable segment structure. S3. Calculate the time difference of arrival of the principal reflection between the two adjacent azimuth angles and construct the crack direction indicator; S4. Based on the crack orientation indicator, construct a significant orientation set and a consistent orientation candidate set, and calculate the comprehensive orientation confidence score of the stable segment structure. Mark the stable segment structure with the highest comprehensive orientation confidence score as the main crack orientation of the current sampling period.

[0006] Preferably, step S1 includes the following steps: S11. Connect the transmitting channel and receiving channel of the ground penetrating radar system to the transmitting antenna and receiving antenna respectively, and set the azimuth sequence; The azimuth sequence is as follows: ,in, Indicates the first Azimuth angle Indicates the total number of azimuth angles; S12. Based on the transmitting antenna and the receiving antenna, a preset sampling period is set, and the common polarization raw waveform data and cross polarization raw waveform data of each azimuth angle are collected periodically according to the azimuth angle sequence, and the common polarization basic dataset and cross polarization basic dataset are constructed respectively. S13. Based on the original waveform data in the common polarization basic dataset and the cross polarization basic dataset, extract the common polarization travel time attribute data, common polarization amplitude attribute data, cross polarization travel time attribute data and cross polarization amplitude attribute data for each azimuth angle according to the same processing rules. S14, respectively, the first Azimuth The corresponding co-polarization travel time attribute data, co-polarization amplitude attribute data, cross-polarization travel time attribute data, and cross-polarization amplitude attribute data are used to construct azimuth data units; S15. Based on the azimuth sequence, the azimuth data units of each azimuth are combined sequentially to construct the angle sequence data.

[0007] Preferably, step S2 includes the following steps: S21. Based on the common polarization travel time attribute data and common polarization amplitude attribute data in the azimuth data unit, identify whether there is a double branch separation structure in the main reflection branch of the common polarization waveform, and mark the identification result as the azimuth splitting mark quantity. S22. Obtain the number of split markers for each azimuth angle, and based on the azimuth angle sequence, group adjacent azimuth angles with consecutive split markers to construct a candidate angle segment structure. Obtain all candidate angle segment structures, and based on the number of split markers, common polarization travel time attribute data, and common polarization amplitude attribute data within the candidate angle segment structure, calculate the duration of the split and the split intensity statistics. S23. Among all candidate angle segment structures, based on the statistics of continuous angle length and splitting intensity, the candidate angle segment structures that meet the preset stability conditions are marked as stable segment structures.

[0008] Preferably, step S3 includes the following steps: S31. Based on adjacent azimuth angles, calculate the arrival time difference of the main reflection and construct the arrival time difference matrix of the main reflection between azimuth angles; S32. Construct a two-dimensional difference map from the azimuth index of the primary reflection arrival time difference matrix between azimuth angles, and normalize and interpolate the two-dimensional difference map. S33. Based on the grayscale distribution of the normalized and interpolated two-dimensional difference map, the maximum value chain is detected, the main ridge direction sequence is constructed, the gradient of the two-dimensional difference map is calculated, the gradient vector field is obtained, and the region with prominent gradient change is extracted as the boundary trend. S34. For each main ridge direction, determine whether the coordinate points in the main ridge direction belong to the strip region. If the number of coordinate points in the set of coordinate points belonging to the strip region in the main ridge direction is greater than or equal to the preset number threshold, then the main ridge direction is taken as the candidate direction of the stable segment structure. S35. Calculate the orientation angle and coverage length of the overlapping point sequence and construct the crack orientation indicator; The method for calculating the arrival time difference of the primary reflection is as follows:

[0009] in, Azimuth Commonly polarized travel time attribute data, Azimuth Copolarized travel time attribute data; The time difference matrix of primary reflection arrival between azimuth angles is as follows:

[0010] The strip-shaped region is:

[0011] in, and They represent the first The starting and ending angles of a stable segment structure; The formula for determining whether a coordinate point in the direction of the main ridge belongs to a strip-shaped region is as follows:

[0012] in, Indicates the first The main ridge direction The middle belongs to the banded region The set of coordinate points, Indicates the first The direction of the main ridge; The formula for calculating the orientation angle and coverage length and constructing the crack direction indicator is as follows:

[0013]

[0014]

[0015] in, Indicates the first The orientation angle of a stable segment structure Indicates the first The coverage length of each stable segment structure This indicates the crack orientation of the k-th stable section structure.

[0016] Preferably, step S4 includes the following steps: S41. Based on the crack orientation indicator, obtain the coverage length of all candidate orientations of the stable segment structure, calculate the average length, obtain all candidate orientations with a coverage length greater than or equal to the average length, and construct a set of significant orientations. S42. Based on the main ridge direction sequence, calculate the angle difference between each direction angle in the significant direction set and the main ridge direction; if the angle difference is less than the preset direction consistency threshold, record it in the consistent direction candidate set. S43. Based on the candidate set of consistent directions, calculate the comprehensive directional credibility score of the stable segment structure according to the coverage length and the contribution of directional consistency. S44. Obtain the comprehensive directional confidence score of all stable segment structures, and mark the stable segment structure with the highest comprehensive directional confidence score as the main crack orientation in the current sampling period; The formula for calculating the credibility score of the comprehensive direction is as follows:

[0017] in, Indicates the first The overall directional reliability score of the stable segment structure Indicates the angle difference. This represents the preset weighting factor. This indicates the contribution of directional consistency. The formula for marking the stable segment structure with the highest comprehensive directional confidence score as the main crack orientation in the current sampling period is as follows:

[0018]

[0019] in, This indicates the stable segment with the highest overall reliability score. This represents the direction angle of the stable segment with the highest overall directional reliability score.

[0020] A ground-penetrating radar intelligent identification system for crack detection, the system comprising: The data acquisition module is used to collect the original waveform data of common polarization and cross polarization in azimuth angle, and extract the common polarization travel time attribute data, common polarization amplitude attribute data, cross polarization travel time attribute data, and cross polarization amplitude attribute data. The module for constructing the marker quantity and the segment structure is used to mark the split marker quantity for azimuth angles based on the common polarization travel time attribute data and the common polarization amplitude attribute data; and to group adjacent azimuth angles with consecutive split markers to construct candidate angle segment structures and stable segment structures. The matrix construction and indicator construction module is used to calculate the arrival time difference of the main reflection between two adjacent azimuths and construct the main reflection arrival time difference matrix between azimuths; construct a two-dimensional difference map and the main ridge direction sequence; and construct the crack orientation indicator. The set construction and labeling module is used to construct a significant orientation set based on the crack orientation indicator; construct a candidate set of consistent orientations; calculate the comprehensive orientation confidence score of the stable segment structure; and label the stable segment structure with the highest comprehensive orientation confidence score as the main crack orientation of the current sampling period.

[0021] Preferably, the data acquisition module includes: The data acquisition unit is used to connect the transmitting channel and receiving channel of the ground penetrating radar system to the transmitting antenna and receiving antenna respectively, and set the azimuth sequence; based on the transmitting antenna and receiving antenna, a preset sampling period is set, and the common polarization raw waveform data and cross polarization raw waveform data of each azimuth angle are collected periodically according to the azimuth sequence, and common polarization basic dataset and cross polarization basic dataset are constructed respectively; based on the raw waveform data in the common polarization basic dataset and the cross polarization basic dataset, the common polarization travel time attribute data, common polarization amplitude attribute data, cross polarization travel time attribute data and cross polarization amplitude attribute data of each azimuth angle are extracted according to the same processing rules.

[0022] Preferably, the marker quantity construction and segment structure construction module includes: The marker construction unit is used to identify whether there is a double-branch separation structure in the main reflection branch of the common polarization waveform based on the common polarization travel time attribute data and common polarization amplitude attribute data in the azimuth data unit, and to mark the identification result as the azimuth split marker quantity; to obtain the split marker quantity of each azimuth, and to group adjacent azimuths with consecutive split markers based on the azimuth sequence to construct candidate angle segment structures, to obtain all candidate angle segment structures, and to calculate the duration of the split and the split intensity statistics based on the split marker quantity, common polarization travel time attribute data and common polarization amplitude attribute data in the candidate angle segment structures; The segment structure construction unit is used to mark the candidate angle segment structures that meet the preset stability conditions as stable segment structures among all candidate angle segment structures, based on the statistics of the continuous angle length and splitting intensity.

[0023] Preferably, the matrix construction and indicator construction module includes: The matrix construction unit is used to calculate the arrival time difference of the primary reflection for adjacent azimuth angles and construct the arrival time difference matrix of the primary reflection between azimuth angles; it constructs a two-dimensional difference map by using the azimuth index as the horizontal and vertical axes of the arrival time difference matrix of the primary reflection between azimuth angles, and performs normalization and interpolation on the two-dimensional difference map; based on the gray-level distribution of the normalized and interpolated two-dimensional difference map, it detects the maximum value chain and constructs the main ridge direction sequence; it calculates the gradient of the two-dimensional difference map to obtain the gradient vector field, and extracts the regions with prominent gradient changes as the boundary trend; The indicator construction unit is used to map the angular range of the stable segment structure into a strip region in a normalized and interpolated two-dimensional difference map based on the stable segment structure; for each main ridge direction, it determines whether the coordinate points in the main ridge direction belong to the strip region; if the number of coordinate points in the set of coordinate points belonging to the strip region in the main ridge direction is greater than or equal to a preset number threshold, then the main ridge direction is taken as a candidate orientation of the stable segment structure; and it calculates the orientation angle and coverage length for the overlapping point sequence and constructs a crack orientation indicator.

[0024] Preferably, the set construction and tagging module includes: The set of construction units is used to obtain the coverage length of all candidate orientations of the stable segment structure based on the crack orientation indicator, calculate the average length, obtain all candidate orientations with a coverage length greater than or equal to the average length, and construct a set of significant orientations. Based on the main ridge direction sequence, the angle difference between each direction angle in the significant direction set and the main ridge direction is calculated; if the angle difference is less than the preset direction consistency threshold, it is recorded in the consistent direction candidate set. The marking unit is used to calculate the comprehensive directional confidence score of the stable segment structure based on the candidate set of consistent directions, the coverage length, and the contribution of directional consistency; to obtain the comprehensive directional confidence score of all stable segment structures, and to mark the stable segment structure with the highest comprehensive directional confidence score as the main crack orientation of the current sampling period.

[0025] As can be seen from the above technical solution, the present invention provides an intelligent identification method for ground-penetrating radar (GPR) for crack detection. Compared with the prior art, the present invention has the following advantages: by collecting co-polarization and cross-polarization waveforms for each azimuth angle within the sampling period and extracting travel time and amplitude attributes, the data from different azimuth angles are constructed into an azimuth angle data sequence with a consistent structure, enabling continuous comparison of crack responses in the angular dimension; based on this, by marking the splitting phenomenon of the double-branch separation in the co-polarization main reflection branch, and aggregating the azimuth angles with continuous splitting marks into candidate angle segments, and then filtering out stable segment structures based on the duration of the angle and the splitting intensity, crack identification is expanded from single-point anomaly judgment to continuous structural feature judgment, which can eliminate occasional anomalies caused by noise interference; subsequently... A two-dimensional difference map is constructed using the arrival time difference of the primary reflections between different azimuth angles. The orientation angle and coverage length of each stable segment are calculated by the overlap relationship between the main ridge chain and the stable segment, forming a direction indicator that reflects the direction of crack extension, thus giving crack identification directional significance and spatial scale significance. Finally, significant directions with coverage lengths greater than the mean are selected from the candidate directions, and a comprehensive direction credibility score is calculated based on the consistency between coverage length and direction. The stable segment with the highest score is selected as the main crack direction, so that the final output has the characteristics of directional continuity, wide spatial coverage, and strong noise resistance, thereby obtaining accurate and stable crack direction identification results in complex underground structural environments. Attached Figure Description

[0026] Figure 1 This is a flowchart illustrating a ground-penetrating radar intelligent identification method for crack detection according to the present invention. Figure 2 This is a schematic diagram of the structure of a ground-penetrating radar intelligent identification system for crack detection according to the present invention. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.

[0028] like Figure 1 As shown in the first embodiment, a ground-penetrating radar intelligent identification method for crack detection is provided. The method includes the following steps: acquiring the original waveform data of common polarization and the original waveform data of cross polarization at the azimuth angle, and extracting the common polarization travel time attribute data, the common polarization amplitude attribute data, the cross polarization travel time attribute data, and the cross polarization amplitude attribute data.

[0029] Specifically, the transmitting and receiving channels of the ground-penetrating radar system are connected to the transmitting and receiving antennas respectively, and an azimuth sequence is set, denoted as... ,in, Let represent the i-th azimuth angle, and I represent the total number of azimuth angles; based on the transmitting and receiving antennas, a preset sampling period is used, and sampling is performed according to the azimuth angle sequence. The common-polarization and cross-polarization raw waveform data are collected periodically at each azimuth angle, and common-polarization and cross-polarization basic datasets are constructed respectively. (In common-polarization mode, the transmitting antenna sends probe electromagnetic waves to the underground medium, and the receiving antenna receives the common-polarization reflected waveform at the corresponding azimuth angle, and the common-polarization reflected waveform is recorded as the common-polarization basic dataset; at the same azimuth angle, the antenna polarization mode is adjusted to cross-polarization mode, the transmitting antenna sends probe electromagnetic waves, and the receiving antenna receives the cross-polarization reflected waveform, and the cross-polarization reflected waveform is recorded as the cross-polarization basic dataset.) Based on the original waveform data in the common polarization basic dataset and the cross polarization basic dataset, the common polarization travel time attribute data, common polarization amplitude attribute data, cross polarization travel time attribute data and cross polarization amplitude attribute data for each azimuth angle are extracted according to the same processing rules (the original waveform is truncated to a fixed length, the truncated waveform is denoised and filtered, the main reflection peak is located in the filtering result according to a unified threshold, and the travel time attribute and amplitude attribute are calculated based on the arrival position of the main reflection peak). The i-th azimuth angle is respectively The corresponding co-polarization travel time attribute data, co-polarization amplitude attribute data, cross-polarization travel time attribute data, and cross-polarization amplitude attribute data are denoted as follows: , , and And construct an azimuth data unit, denoted as Based on azimuth sequence The azimuth data units of each azimuth angle are combined sequentially to construct the angle sequence data.

[0030] In this invention, common-polarization and cross-polarization waveforms are acquired for each azimuth angle, and travel time and amplitude attributes are extracted to construct azimuth angle data units. By constructing data units of different polarization signals into a unified format, a serialized data framework that can be used for subsequent angular dimension comparisons is formed, enabling continuous retrieval and statistics along the azimuth direction in subsequent analysis. This provides a directly comparable data structure for subsequent identification of crack response characteristics at different azimuth angles, extending crack identification from one-dimensional retrieval in the time domain to multi-angle structural analysis, improving the significance of crack response differences, and enhancing the reliability of subsequent splitting marking and direction analysis.

[0031] Step S2: For the common polarization data, extract the continuous angular segments in which electromagnetic wave splitting occurs in each azimuth angle to form the length, intensity and distribution characteristics of the stable segments.

[0032] Based on azimuth data unit Common polarization travel time attribute data Co-polarization amplitude attribute data Identify the presence of a dual-branch separation structure in the main reflection branch of the common-polarization waveform, and mark the identification result with the azimuth angle. The splitting marker quantity (used to characterize whether electromagnetic wave splitting occurs at this azimuth angle, and for use in determining subsequent angle continuity segments); Specifically as follows: In azimuth data unit The original waveform is used to obtain the common-polarization travel time attribute data. Establish a time window for the center ,in, Half width of the window ( Depending on the antenna bandwidth and sampling interval, it can be taken as 2 to 5 corresponding waveform resolution units or determined based on experimental fitting.

[0033] In the time window Peak detection is performed within the area to identify the set of local maxima. Each peak corresponds to a time and amplitude (If only one peak is detected, it is considered that there is no splitting). The peak detection principle adopts a uniform threshold. minimum peak spacing ( This is the background noise factor. Sampling unit × n, where n is the sampling point multiplier used to set the minimum peak spacing threshold); If there are two or more peaks in P, and the second peak satisfies the following conditions: time difference Amplitude ratio Then it is determined that the azimuth waveform has a two-branch separation structure, wherein, The maximum peak amplitude in the window. This is the second largest peak, indicating Preset amplitude ratio threshold; The formula for calculating the continuous splitting metric is as follows: ; in, Indicates azimuth angle Continuous splitting measure This indicates the time difference between the main peak and the secondary peak. This represents the preset time normalization benchmark for the split measurement. This indicates the amplitude of the second largest peak detected within the window. This indicates the amplitude of the largest peak detected within the window; It should be noted that the time difference reflects the difference in electromagnetic wave reflection paths caused by the crack (wider cracks will result in a larger time difference between the main peak and the secondary peak); the amplitude ratio reflects the intensity of the split signal (the amplitude of the secondary peak of a real crack will not be too low, while the amplitude of a false split secondary peak caused by noise is close to zero). Traditional methods only determine "whether there are two branches," while this formula uses continuous numerical values. Describe the "realism" of the split, for example: a split caused by a crack. It could be 0.8, while noise-induced splitting... It may be as low as 0.1. This formula can effectively filter out "false splits at a single azimuth angle" (such as occasional interference from radar antennas), only filtering out splits at multiple consecutive azimuth angles. Only when all values ​​meet the threshold will they be identified as crack signals, thus significantly reducing the false positive rate. The formula for calculating the number of splitting markers is as follows: ; in, Indicates the number of splitting markers. This indicates the preset splitting criterion threshold; Obtain the splitting marker value for each azimuth angle and based on the azimuth angle sequence. Adjacent azimuth angles with consecutive splitting markers are grouped to construct candidate angle segment structures (including the start and end angles of the corresponding candidate angle segments). All candidate angle segment structures are obtained. Based on the number of splitting markers, co-polarization travel time attribute data, and co-polarization amplitude attribute data within the candidate angle segment structures, the duration of the splitting angle and the splitting intensity statistics are calculated, using the following formulas: ; ; ; ; ; in, This represents the duration of the angle of the k-th candidate angle segment. Indicates the azimuth index of the candidate angle segment structure. Indicates the starting azimuth index of the candidate angle segment structure. The sampling step size, This represents the average splitting degree of the k-th candidate angle segment. This represents the splitting stability of the k-th candidate angle segment. This represents the coefficient of variation for the k-th candidate angle segment. This represents the splitting intensity statistic for the k-th candidate angle segment. Indicates the preset consistency weight; It should be noted that the electromagnetic wave splitting of a real crack will occur at multiple consecutive azimuth angles. (larger), while noise splitting typically lasts only 1-2 azimuth angles ( (Smaller). By setting The minimum threshold (e.g., ≥5°) can directly exclude candidate segments caused by noise; The larger the value, the wider the crack extends within that angle range, and the more representative the subsequent calculation of the "crack direction" will be, avoiding misjudgments of the direction caused by "local small-scale splits".

[0034] Reflects the average splitting intensity of candidate segments, Reflecting the degree of fluctuation in splitting intensity, the actual splitting intensity of a crack fluctuates little. Small), High score; large fluctuations in splitting intensity due to noise ( big), Low score.

[0035] Among all candidate angle segment structures, based on the statistics of continuous angle length and splitting intensity, candidate angle segment structures that meet the preset stability conditions are marked as stable segment structures.

[0036] Step S3: Calculate the arrival time difference of the primary reflection between two adjacent azimuth angles and construct the arrival time difference matrix of the primary reflection between azimuth angles; construct a two-dimensional difference map and the main ridge direction sequence; construct the crack direction indicator.

[0037] For azimuth angle and azimuth Calculate the time difference of arrival of the primary reflection. And construct the time difference matrix of primary reflection arrival between azimuth angles, denoted as: ; The time difference matrix of primary reflection arrival between azimuth angles A two-dimensional difference map is constructed using azimuth index as the horizontal and vertical axes, denoted as For two-dimensional difference plots Perform normalization and interpolation (so that) (It has grayscale or pseudocolor representation in the angle continuity domain). Based on the grayscale distribution of the normalized and interpolated two-dimensional difference map, maxima chains are detected, and main ridge direction sequences are constructed; the two-dimensional difference map... Gradient calculations are performed to obtain the gradient vector field, and regions with prominent gradient changes are extracted as boundary trends (both the main ridge and the boundary are represented by discrete coordinate sequences, and the corresponding angle index mapping is retained; the direction of the main ridge reflects the maximum extension direction of the difference change, and the boundary trend reflects the geometric shape of the steep change zone). Based on the stable segment structure, the angular range of the stable segment structure is mapped into a banded region in the normalized and interpolated two-dimensional difference map. ,in, and These represent the starting and ending angles of the k-th stable segment structure, respectively; For each main ridge direction, determine whether the coordinates of the points along that main ridge direction belong to the strip region. The details are as follows: ; in, Indicates the direction of the nth main ridge The middle belongs to the banded region The set of coordinate points, Indicates the direction of the nth main ridge; If the main ridge direction The middle belongs to the banded region set of coordinate points If the number of coordinate points in the data is greater than or equal to a preset threshold, then the main ridge direction will be adjusted. As a candidate orientation for stable segment structure; The orientation angle and coverage length of the overlapping point sequence are calculated, and a crack orientation indicator is constructed, as follows: ; ; ; in, This represents the orientation angle of the k-th stable segment structure. This represents the coverage length of the k-th stable segment structure. This indicates the crack orientation of the k-th stable section structure.

[0038] Step S4: Based on the crack orientation indicator, construct a significant orientation set; construct a consistent orientation candidate set; calculate the comprehensive orientation confidence score of the stable segment structure, and mark the stable segment structure with the highest comprehensive orientation confidence score as the main crack orientation of the current sampling period.

[0039] Based on crack orientation indicator The coverage lengths of all candidate orientations for the stable segment structure are obtained, and the average length is calculated. All candidate orientations with a coverage length greater than or equal to the average length are then identified, and a set of significant orientations is constructed, denoted as […]. ; Based on the main ridge direction sequence, the set of significant orientations The angle difference between each direction angle and the main ridge direction is calculated. ,in, The direction angle is obtained by fitting a straight line (such as using the least squares method) to R(n), representing the geometric orientation of the main ridge direction. R(n) represents the point sequence corresponding to the nth main ridge direction extracted from the difference map. If the angle difference is less than a preset direction consistency threshold, it is recorded in the consistent direction candidate set, denoted as . ; Based on the candidate set of consistent directions Based on the contribution of coverage length and directional consistency, the comprehensive directional reliability score of the k-th stable segment structure is calculated using the following formula: ; in, This represents the overall directional reliability score of the k-th stable segment structure. Indicates the angle difference. This represents the preset weighting factor. This indicates the contribution of directional consistency. Obtain the comprehensive directional confidence score of all stable structural segments, and mark the stable structural segment with the highest comprehensive directional confidence score as the main crack orientation for the current sampling period, as follows: ; ; in, This indicates the stable segment with the highest overall directional confidence score (i.e., the direction of the main fracture). This represents the direction angle of the stable segment with the highest overall directional reliability score.

[0040] Will The angular range is mapped to the azimuth domain and labeled in a strip structure. Mapped to the difference map coordinate domain, in Draw directional line segments at corresponding positions to ensure that the direction shown on the diagram matches the mathematical definition. Represent other stable structural segments as gray-scale banded areas, so that... It is displayed in layers with the secondary sections without changing the meaning of their orientation angles.

[0041] The results obtained in this period , Record as status The angle deviation and coverage length change are compared during the next scan cycle for directional stability tracking.

[0042] like Figure 2 As shown in this second embodiment, a ground-penetrating radar intelligent identification system for crack detection is provided. The system includes: a data acquisition module, a marker quantity construction and segment structure construction module, a matrix construction and indicator quantity construction module, and a set construction and marking module. Data acquisition module: Collects raw waveform data of common polarization and cross polarization in azimuth angle, and extracts common polarization travel time attribute data, common polarization amplitude attribute data, cross polarization travel time attribute data, and cross polarization amplitude attribute data; The module for constructing marker quantities and segment structures: Based on the common polarization travel time attribute data and the common polarization amplitude attribute data, it marks the split marker quantities for azimuth angles; it groups adjacent azimuth angles that have consecutive split markers to construct candidate angle segment structures and stable segment structures; Matrix construction and indicator construction module: Calculates the time difference of arrival of the main reflection between two adjacent azimuths and constructs the matrix of the time difference of arrival of the main reflection between azimuths; constructs a two-dimensional difference map and the main ridge direction sequence; constructs the crack orientation indicator. Set construction and labeling module: Based on the crack orientation indicator, construct a significant orientation set; construct a candidate set of consistent orientations; calculate the comprehensive orientation confidence score of the stable segment structure, and label the stable segment structure with the highest comprehensive orientation confidence score as the main crack orientation of the current sampling period.

[0043] Specifically, the data acquisition module includes a data acquisition unit; Data acquisition unit: Connects the transmitting and receiving channels of the ground-penetrating radar system to the transmitting and receiving antennas respectively, and sets the azimuth sequence; based on the transmitting and receiving antennas, presets the sampling period and samples according to the azimuth sequence. The original waveform data of common polarization and cross polarization at each azimuth angle are collected periodically in sequence, and the common polarization basic dataset and cross polarization basic dataset are constructed respectively. Based on the original waveform data in the common polarization basic dataset and cross polarization basic dataset, the common polarization travel time attribute data, common polarization amplitude attribute data, cross polarization travel time attribute data and cross polarization amplitude attribute data of each azimuth angle are extracted according to the same processing rules.

[0044] Specifically, the marker construction and segment structure construction module includes a marker construction unit and a segment structure construction unit; The marker construction unit: Based on the common polarization travel time attribute data and common polarization amplitude attribute data in the azimuth data unit, it identifies whether there is a double branch separation structure in the main reflection branch of the common polarization waveform, and marks the identification result as the azimuth split marker quantity; it obtains the split marker quantity of each azimuth, and based on the azimuth sequence, it groups adjacent azimuths with consecutive split markers to construct candidate angle segment structures, obtains all candidate angle segment structures, and calculates the duration angle length and split intensity statistics of the split based on the split marker quantity, common polarization travel time attribute data and common polarization amplitude attribute data in the candidate angle segment structure. Segment structure construction unit: Among all candidate angle segment structures, based on the statistics of continuous angle length and splitting intensity, candidate angle segment structures that meet the preset stability conditions are marked as stable segment structures.

[0045] Specifically, the matrix construction and indicator construction module includes a matrix construction unit and an indicator construction unit; Matrix Construction Unit: For adjacent azimuth angles, calculate the arrival time difference of the primary reflection and construct the arrival time difference matrix of the primary reflection between azimuth angles; construct a two-dimensional difference map using the azimuth index as the horizontal and vertical axes of the arrival time difference matrix of the primary reflection between azimuth angles, and normalize and interpolate the two-dimensional difference map; based on the gray-level distribution of the normalized and interpolated two-dimensional difference map, detect the maximum value chain and construct the main ridge direction sequence; calculate the gradient of the two-dimensional difference map to obtain the gradient vector field, and extract the region with prominent gradient changes as the boundary trend; Indicator Construction Unit: Based on the stable segment structure, the angular range of the stable segment structure is mapped to a strip region in the normalized and interpolated two-dimensional difference map; for each main ridge direction, it is determined whether the coordinate points in the main ridge direction belong to the strip region; if the number of coordinate points in the set of coordinate points belonging to the strip region in the main ridge direction is greater than or equal to the preset number threshold, the main ridge direction is taken as a candidate orientation of the stable segment structure; the orientation angle and coverage length are calculated for the overlapping point sequence and crack orientation indicators are constructed.

[0046] Specifically, the set construction and tagging module includes set construction units and tagging units; Set construction unit: Based on the crack orientation indicator, obtain the coverage length of all candidate orientations of the stable segment structure, calculate the average length, obtain all candidate orientations with a coverage length greater than or equal to the average length, and construct a set of significant orientations; Based on the main ridge direction sequence, calculate the angle difference between each direction angle in the significant direction set and the main ridge direction; if the angle difference is less than the preset direction consistency threshold, record it in the consistent direction candidate set. Marking Unit: Based on the candidate set of consistent directions, and based on the coverage length and the contribution of directional consistency, calculate the comprehensive directional confidence score of the stable segment structure; obtain the comprehensive directional confidence score of all stable segment structures, and mark the stable segment structure with the highest comprehensive directional confidence score as the main crack orientation of the current sampling period.

[0047] In summary, this invention provides an intelligent identification method and system for ground-penetrating radar (GPR) for crack detection. Compared with existing technologies, this invention has the following advantages: By acquiring co-polarization and cross-polarization waveforms for each azimuth angle within the sampling period and extracting travel time and amplitude attributes, data from different azimuth angles are constructed into an azimuth angle data sequence with a consistent structure, enabling continuous comparison of crack responses in the angular dimension; based on this, by marking the splitting phenomenon of the double-branch separation in the co-polarization main reflection branch, and aggregating azimuth angles with continuous splitting marks into candidate angle segments, and then filtering out stable segment structures based on the duration of the angle and the splitting intensity, crack identification is expanded from single-point anomaly judgment to continuous structural feature judgment, which can eliminate occasional anomalies caused by noise interference; subsequently... This invention constructs a two-dimensional difference map using the time difference of arrival of the primary reflections at different azimuth angles. By calculating the azimuth angle and coverage length of each stable segment through the overlap relationship between the main ridge chain and the stable segment, a direction indicator reflecting the direction of crack propagation is formed, giving crack identification directional significance and spatial scale meaning. Finally, significant directions with coverage lengths greater than the mean are selected from the candidate directions, and a comprehensive direction reliability score is calculated based on the consistency between coverage length and direction. The stable segment with the highest score is selected as the main crack direction, resulting in a final output with continuous direction, wide spatial coverage, and strong noise resistance, thus obtaining accurate and stable crack direction identification results in complex underground structural environments. Therefore, this invention effectively overcomes the various shortcomings of existing technologies and has high industrial application value.

[0048] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the flow or function according to the embodiments of this application is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., a solid-state disk (SSD)).

[0049] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0050] The various embodiments in this specification are described in a related manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0051] 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.

Claims

1. A ground-penetrating radar intelligent identification method for crack detection according to the present invention, characterized in that, Includes the following steps: S1. Collect the original waveform data of common polarization and cross polarization in azimuth angle, and extract the common polarization travel time attribute data, common polarization amplitude attribute data, cross polarization travel time attribute data and cross polarization amplitude attribute data; S2. Based on the common polarization travel time attribute data and the common polarization amplitude attribute data, the azimuth angle is marked with a split mark amount, and adjacent azimuth angles with consecutive split marks are grouped to construct a stable segment structure. S3. Calculate the time difference of arrival of the principal reflection between the two adjacent azimuth angles and construct the crack direction indicator; S4. Based on the crack orientation indicator, construct a significant orientation set and a consistent orientation candidate set, and calculate the comprehensive orientation confidence score of the stable segment structure. Mark the stable segment structure with the highest comprehensive orientation confidence score as the main crack orientation of the current sampling period.

2. The intelligent identification method for ground-penetrating radar for crack detection according to claim 1, characterized in that: S1 includes the following steps: S11. Connect the transmitting channel and receiving channel of the ground penetrating radar system to the transmitting antenna and receiving antenna respectively, and set the azimuth sequence; The azimuth sequence is as follows: ,in, Indicates the first Azimuth angle Indicates the total number of azimuth angles; S12. Based on the transmitting antenna and the receiving antenna, a preset sampling period is set, and the common polarization raw waveform data and cross polarization raw waveform data of each azimuth angle are collected periodically according to the azimuth angle sequence, and the common polarization basic dataset and cross polarization basic dataset are constructed respectively. S13. Based on the original waveform data in the common polarization basic dataset and the cross polarization basic dataset, extract the common polarization travel time attribute data, common polarization amplitude attribute data, cross polarization travel time attribute data and cross polarization amplitude attribute data for each azimuth angle according to the same processing rules. S14, respectively, the first Azimuth The corresponding co-polarization travel time attribute data, co-polarization amplitude attribute data, cross-polarization travel time attribute data, and cross-polarization amplitude attribute data are used to construct azimuth data units; S15. Based on the azimuth sequence, the azimuth data units of each azimuth are combined sequentially to construct the angle sequence data.

3. The intelligent identification method for ground-penetrating radar for crack detection according to claim 2, characterized in that: S2 includes the following steps: S21. Based on the common polarization travel time attribute data and common polarization amplitude attribute data in the azimuth data unit, identify whether there is a double branch separation structure in the main reflection branch of the common polarization waveform, and mark the identification result as the azimuth splitting mark quantity. S22. Obtain the number of split markers for each azimuth angle, and based on the azimuth angle sequence, group adjacent azimuth angles with consecutive split markers to construct a candidate angle segment structure. Obtain all candidate angle segment structures, and based on the number of split markers, common polarization travel time attribute data, and common polarization amplitude attribute data within the candidate angle segment structure, calculate the duration of the split and the split intensity statistics. S23. Among all candidate angle segment structures, based on the statistics of continuous angle length and splitting intensity, the candidate angle segment structures that meet the preset stability conditions are marked as stable segment structures.

4. The intelligent identification method for ground-penetrating radar for crack detection according to claim 3, characterized in that: S3 includes the following steps: S31. Based on adjacent azimuth angles, calculate the arrival time difference of the main reflection and construct the arrival time difference matrix of the main reflection between azimuth angles; S32. Construct a two-dimensional difference map from the azimuth index of the primary reflection arrival time difference matrix between azimuth angles, and normalize and interpolate the two-dimensional difference map. S33. Based on the grayscale distribution of the normalized and interpolated two-dimensional difference map, the maximum value chain is detected, the main ridge direction sequence is constructed, the gradient of the two-dimensional difference map is calculated, the gradient vector field is obtained, and the region with prominent gradient change is extracted as the boundary trend. S34. For each main ridge direction, determine whether the coordinate points in the main ridge direction belong to the strip region. If the number of coordinate points in the set of coordinate points belonging to the strip region in the main ridge direction is greater than or equal to the preset number threshold, then the main ridge direction is taken as the candidate direction of the stable segment structure. S35. Calculate the orientation angle and coverage length of the overlapping point sequence and construct the crack orientation indicator; The method for calculating the arrival time difference of the primary reflection is as follows: in, Azimuth Commonly polarized travel time attribute data, Azimuth Copolarized travel time attribute data; The time difference matrix of primary reflection arrival between azimuth angles is as follows: The strip-shaped region is: in, and They represent the first The starting and ending angles of a stable segment structure; The formula for determining whether a coordinate point in the direction of the main ridge belongs to a strip-shaped region is as follows: in, Indicates the first The main ridge direction The middle belongs to the banded region The set of coordinate points, Indicates the first The direction of the main ridge; The formula for calculating the orientation angle and coverage length and constructing the crack direction indicator is as follows: in, Indicates the first The orientation angle of a stable segment structure Indicates the first The coverage length of each stable segment structure This indicates the crack orientation of the k-th stable section structure.

5. The intelligent identification method for ground-penetrating radar for crack detection according to claim 4, characterized in that: S4 includes the following steps: S41. Based on the crack orientation indicator, obtain the coverage length of all candidate orientations of the stable segment structure, calculate the average length, obtain all candidate orientations with a coverage length greater than or equal to the average length, and construct a set of significant orientations. S42. Based on the main ridge direction sequence, calculate the angle difference between each direction angle in the significant direction set and the main ridge direction; if the angle difference is less than the preset direction consistency threshold, record it in the consistent direction candidate set. S43. Based on the candidate set of consistent directions, calculate the comprehensive directional credibility score of the stable segment structure according to the coverage length and the contribution of directional consistency. S44. Obtain the comprehensive directional confidence score of all stable segment structures, and mark the stable segment structure with the highest comprehensive directional confidence score as the main crack orientation in the current sampling period; The formula for calculating the credibility score of the comprehensive direction is as follows: in, Indicates the first The overall directional reliability score of the stable segment structure Indicates the angle difference. This represents the preset weighting factor. This indicates the contribution of directional consistency. The formula for marking the stable segment structure with the highest comprehensive directional confidence score as the main crack orientation in the current sampling period is as follows: in, This indicates the stable segment with the highest overall reliability score. This represents the direction angle of the stable segment with the highest overall directional reliability score.

6. A ground-penetrating radar intelligent identification system for crack detection, comprising executing a ground-penetrating radar intelligent identification method for crack detection as described in any one of claims 1-5, characterized in that, The system includes: The data acquisition module is used to collect the original waveform data of common polarization and cross polarization in azimuth angle, and extract the common polarization travel time attribute data, common polarization amplitude attribute data, cross polarization travel time attribute data, and cross polarization amplitude attribute data. The module for constructing the marker quantity and the segment structure is used to mark the split marker quantity for azimuth angles based on the common polarization travel time attribute data and the common polarization amplitude attribute data; and to group adjacent azimuth angles with consecutive split markers to construct candidate angle segment structures and stable segment structures. The matrix construction and indicator construction module is used to calculate the arrival time difference of the main reflection between two adjacent azimuths and construct the main reflection arrival time difference matrix between azimuths; construct a two-dimensional difference map and the main ridge direction sequence; and construct the crack orientation indicator. The set construction and labeling module is used to construct a significant orientation set based on the crack orientation indicator; construct a candidate set of consistent orientations; calculate the comprehensive orientation confidence score of the stable segment structure; and label the stable segment structure with the highest comprehensive orientation confidence score as the main crack orientation of the current sampling period.

7. A data analysis and early warning system for concrete defect detection according to claim 6, characterized in that: The data acquisition module includes: The data acquisition unit is used to connect the transmitting channel and receiving channel of the ground penetrating radar system to the transmitting antenna and receiving antenna respectively, and set the azimuth sequence; based on the transmitting antenna and receiving antenna, a preset sampling period is set, and the common polarization raw waveform data and cross polarization raw waveform data of each azimuth angle are collected periodically according to the azimuth sequence, and common polarization basic dataset and cross polarization basic dataset are constructed respectively; based on the raw waveform data in the common polarization basic dataset and the cross polarization basic dataset, the common polarization travel time attribute data, common polarization amplitude attribute data, cross polarization travel time attribute data and cross polarization amplitude attribute data of each azimuth angle are extracted according to the same processing rules.

8. The intelligent identification method for ground-penetrating radar for crack detection according to claim 7, characterized in that: The marker construction and segment structure construction module includes: The marker construction unit is used to identify whether there is a double-branch separation structure in the main reflection branch of the common polarization waveform based on the common polarization travel time attribute data and common polarization amplitude attribute data in the azimuth data unit, and to mark the identification result as the azimuth split marker quantity; to obtain the split marker quantity of each azimuth, and to group adjacent azimuths with consecutive split markers based on the azimuth sequence to construct candidate angle segment structures, to obtain all candidate angle segment structures, and to calculate the duration of the split and the split intensity statistics based on the split marker quantity, common polarization travel time attribute data and common polarization amplitude attribute data in the candidate angle segment structures; The segment structure construction unit is used to mark the candidate angle segment structures that meet the preset stability conditions as stable segment structures among all candidate angle segment structures, based on the statistics of the continuous angle length and splitting intensity.

9. The intelligent identification method for ground-penetrating radar for crack detection according to claim 8, characterized in that: The matrix construction and indicator construction module includes: The matrix construction unit is used to calculate the arrival time difference of the primary reflection for adjacent azimuth angles and construct the arrival time difference matrix of the primary reflection between azimuth angles; it constructs a two-dimensional difference map by using the azimuth index as the horizontal and vertical axes of the arrival time difference matrix of the primary reflection between azimuth angles, and performs normalization and interpolation on the two-dimensional difference map; based on the gray-level distribution of the normalized and interpolated two-dimensional difference map, it detects the maximum value chain and constructs the main ridge direction sequence; it calculates the gradient of the two-dimensional difference map to obtain the gradient vector field, and extracts the regions with prominent gradient changes as the boundary trend; The indicator construction unit is used to map the angular range of the stable segment structure into a strip region in a normalized and interpolated two-dimensional difference map based on the stable segment structure; for each main ridge direction, it determines whether the coordinate points in the main ridge direction belong to the strip region; if the number of coordinate points in the set of coordinate points belonging to the strip region in the main ridge direction is greater than or equal to a preset number threshold, then the main ridge direction is taken as a candidate orientation of the stable segment structure; and it calculates the orientation angle and coverage length for the overlapping point sequence and constructs a crack orientation indicator.

10. The intelligent identification method for ground-penetrating radar for crack detection according to claim 9, characterized in that: The set construction and tagging module includes: The set of construction units is used to obtain the coverage length of all candidate orientations of the stable segment structure based on the crack orientation indicator, calculate the average length, obtain all candidate orientations with a coverage length greater than or equal to the average length, and construct a set of significant orientations. Based on the main ridge direction sequence, the angle difference between each direction angle in the significant direction set and the main ridge direction is calculated; if the angle difference is less than the preset direction consistency threshold, it is recorded in the consistent direction candidate set. The marking unit is used to calculate the comprehensive directional confidence score of the stable segment structure based on the candidate set of consistent directions, the coverage length, and the contribution of directional consistency; to obtain the comprehensive directional confidence score of all stable segment structures, and to mark the stable segment structure with the highest comprehensive directional confidence score as the main crack orientation of the current sampling period.

Images