Radar signal time domain peak value-based void disease three-dimensional reconstruction method

Abstract

The invention discloses a void disease three-dimensional reconstruction method based on a radar signal time domain peak value, and belongs to the field of pavement disease nondestructive testing. Aiming at the problems that the existing ground penetrating radar (GPR) depends on manual interpretation, the resolution of a grey-scale map is low, and the three-dimensional space form of the void disease cannot be obtained, the method comprises the following steps: firstly, processing GPR data and utilizing time domain signal characteristics: identifying zero crossing points corresponding to the first peak value and the last peak value of a single-channel A-scan signal, and accurately positioning the starting depth and the ending depth of a void interface; based on the transverse continuity of the B-scan image, the relevance of adjacent multiple channels of signals is analyzed, the transverse start-stop channel number and the vertical depth range of the void area are determined, and a horizontal plane positioning bounding box is generated; and finally, fusing coordinate information of a multi-channel B-scan positioning frame, and constructing a void disease geometric model containing three-dimensional sizes of a long (X-axis), a wide (Y-axis) and a deep (Z-axis) through space conversion and superposition. According to the method, the depth is directly positioned from the time domain peak value, complex frequency domain calculation is avoided, and the noise immunity is high; horizontal boundary automatic extraction and disease three-dimensional visualization are realized, and accurate geometric parameters are provided for structural safety assessment; the method can be extensively applied to other pavement disease detection.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of road surface disease detection, and particularly relates to a method for processing two-dimensional ground penetrating radar signals by using time-frequency domain transformation to enhance signal features of disease areas, increase atlas resolution, realize three-dimensional visualization and reconstruction of road surface void disease areas. BACKGROUND

[0002] Cement pavement plays an important role in the transportation network in China, and is widely used in the pavement structure of trunk roads (provincial roads and county roads) due to its high bearing capacity, good stability and low maintenance cost. Local gaps between the pavement slab and the base are caused by uneven construction, uneven roadbed settlement, and the influence of vehicle load and temperature stress. When the caulking material between the slabs falls off, rainwater can easily enter the void area, and the range of the water-containing void area will further expand under alternating load, aggravating the development of the void area disease and greatly affecting the bearing capacity of the road, leading to settlement and misalignment of the cement slabs, and eventually causing broken slabs, which seriously affects traffic. It is urgent to establish a method for determining the geometric size of the disease area to assess the bearing capacity of the pavement slab, maintain the pavement before the slabs break, prolong the service life of the pavement, and provide a scientific basis for precise pavement maintenance.

[0003] The void disease detection methods mainly include pulse response method, acoustic vibration method, vibration sensing and ground penetrating radar (GPR) method. Among them, GPR is the most effective non-destructive testing technology for road void detection, and has been widely used in pavement disease and structure layer thickness detection. A large amount of data is generated during GPR pavement detection, and currently it is necessary to rely on experienced professionals to judge the disease area through GPR grayscale images. The grayscale image is a graph composed of GPR signals (A-scan) along the line direction, which has the problem of low resolution due to interference from clutter, leading to missed and misjudged road diseases by manual judgment, so that the disease area cannot be maintained in time and deteriorates rapidly, which seriously affects the service life of the pavement.

[0004] The A-Scan signal of GPR is the minimum information unit of GPR, which contains information such as depth, medium characteristics and thickness size. If the GPR signal can be used for disease identification, the disease boundary position can be accurately located. GPR signal is a transient non-stationary signal. If the traditional Fourier transform (FFT) is used for time-frequency domain conversion, the distribution of different frequency components in the time axis cannot be revealed due to the limitation of the characteristics of the FFT basis function. Continuous wavelet transform (CWT) is a time-frequency localization analysis method with fixed window size and variable shape. It captures local and global features of signals through different scales and is commonly used for signal time-frequency analysis or accurate signal transient positioning, especially for signals with sudden changes in instantaneous frequency. Therefore, continuous wavelet transform can increase the resolution of GPR signal and provide a new idea and method for positioning disease area and determining the geometric characteristics of disease area.

[0005] The patent with patent number CN109444176A provides a method for detecting the depth of concrete void under steel shell. The patent considers the influence of water content on the void depth by establishing a void depth calibration curve, but does not involve enhancing the resolution of disease area and highlighting the disease characteristics. Patents with patent numbers CN116468683A and CN113326865A use deep learning to identify patterns to obtain the size of the disease body and achieve three-dimensional reconstruction of the disease. Patents with patent numbers CN115690307A and CN104766365A convert the coordinates of the disease area to obtain the three-dimensional body of the disease. The above patents do not directly use GPR two-dimensional signal for three-dimensional visualization of disease or through the method of improving disease resolution to determine the disease area.

[0006] Currently, there is a lack of method for quickly determining the disease area by improving the resolution of GPR two-dimensional signal. Therefore, the applicant starts from signal processing and uses continuous wavelet transform to convert GPR time domain signal to time-frequency domain signal, i.e. each A-scan signal becomes a time-frequency domain graph to enhance signal characteristics. On this basis, a three-dimensional reconstruction method for road void disease area is proposed to realize three-dimensional feature display of void area and provide geometric parameters of disease area for road structure safety evaluation. SUMMARY

[0007] In order to solve the problem that the existing ground penetrating radar cannot determine the specific spatial form information of void disease, a three-dimensional reconstruction method for road void disease area is proposed.

[0008] The present application is realized by the following steps.

[0009] Step 1. Data acquisition: use ground penetrating radar (GPR) to scan the cement pavement with void disease and obtain the original GPR echo data.

[0010] Step 2. Data processing, referring to the patent method of patent application CN115542278A, the original GPR data collected in step 1 is processed to obtain a set of processed radar data images.

[0011] Step 3. Single-channel signal peak identification and reference determination, identify the time domain peak feature in the single-channel A-scan signal, locate the zero-crossing points corresponding to the first peak and the last peak, and determine the starting depth and termination depth of the void interface in the channel signal to form a single-channel void interface position interval reference.

[0012] Step 4. Two-dimensional positioning of the void area, based on the results of step 3, determine the starting channel number, termination channel number, starting depth and termination depth of the void area on the B-scan image, and obtain the position coordinate information of the void area in the two-dimensional profile.

[0013] Step 5. Void area plane contour framing, using the lateral continuity feature of the B-scan image, by analyzing the correlation of the void interface positions of adjacent multiple A-scan signals, generate the positioning boundary box of the void area in the horizontal plane.

[0014] Step 6. Three-dimensional model construction, superimpose and fuse the starting point coordinates and termination point coordinates of the void area positioning frame obtained in the B-scan image of different measurement channels (or measurement lines) along the depth direction (Z axis) to form a C-scan slice structure; finally, through spatial coordinate conversion, construct a three-dimensional geometric model of the void disease, and generate a spatial entity containing three-dimensional size parameters of length (X axis), width (Y axis) and depth (Z axis).

[0015] The beneficial effects of the present application are:

[0016] (1) By identifying the time domain peak feature (zero-crossing points corresponding to the first and last peaks) of the single-channel A-scan signal, the depth reference of the void interface is directly extracted. The present application avoids the computational complexity of traditional frequency domain transformation or full waveform inversion, and can still accurately realize depth positioning in a strong noise environment, providing a high reliability spatial reference for subsequent three-dimensional reconstruction.

[0017] (2) Based on the lateral continuity feature of the B-scan image, the void area plane positioning box is generated by correlation analysis of adjacent multiple A-scan signals. The horizontal boundary of the void area is efficiently and automatically extracted, improving the efficiency.

[0018] (3) The present application realizes the graphical recognition of the geometric size of the disease area by converting the GPR signal from two-dimensional to three-dimensional. This method provides a basis for analyzing the structural mechanics performance of the void area, and further evaluates the structural safety of the disease area. The proposed technical solution can also be applied to the three-dimensional visualization and three-dimensional reconstruction of other pavement diseases. BRIEF DESCRIPTION OF DRAWINGS

[0019] Figure 1 The three-dimensional reconstruction overall flow chart for the void disease;

[0020] Figure 2 The A-scan signal time domain waveform diagram without disease;

[0021] Figure 3 The A-scan signal time domain waveform diagram for the void disease;

[0022] Figure 4 The A-scan signal peak positioning principle diagram for the void area;

[0023] Figure 5 The B-scan two-dimensional void image fusion principle diagram

[0024] Figure 6 The B-scan void point positioning result diagram for the void area;

[0025] Figure 7 The B-scan void disease rectangular positioning diagram for the void area;

[0026] Figure 8 The B-scan positioning result of each channel;

[0027] Figure 9 The three-dimensional disease body superposition process principle diagram;

[0028] Figure 10 The three-dimensional reconstruction result diagram for the void disease;

[0029] Figure 11 The GPR field experiment diagram DETAILED DESCRIPTION

[0030] The specific content of the present application is further explained in detail below in combination with the drawings and examples. The implementation range of the present application is not limited to this. Any equivalent transformation based on the technical scheme of the present application falls within the protection scope of the present application.

[0031] The present application uses a simple and efficient identification method to locate the void disease of the road surface and perform three-dimensional reconstruction, which is convenient for the detection personnel to check the void state. The six steps can be realized, and the overall flow chart is shown in Figure 1 .

[0032] Step 1. Obtain the GPR signal through field test.

[0033] This step selects the typical road surface in the urban area of Guiyang City, Guizhou Province to carry out GPR field experiment. As shown in Figure 11As shown, the experiment adopts a vehicle-mounted mobile 3DGPR dynamic detection mode. The Raptor-45 type 3DGPR system is fixed on a special bearing platform at the rear of the measuring vehicle, and the engineering personnel in the vehicle use the pulse radar hawk data software to receive and store the underground three-dimensional scanning data in real time. The original radar signal data S[m, n] is obtained, m is the sampling point number of each A-scan in the GPR data, and n is the A-scan channel number contained in the GPR data, which is used for subsequent processing and identification.

[0034] Step 2. Original GPR data preprocessing. The original GPR data S[m, n] collected in step 1 is processed, including static correction, Z-score standardization and background removal data processing process, to obtain the sample set X[m, n], wherein the non-hollow disease signal is as shown in Figure 2 The hollow disease signal is as shown in Figure 3 .

[0035] Step 3. Single-channel signal peak value identification and reference determination.

[0036] (1) Extract the peak value characteristics of each signal of the field test, find the signal point with amplitude greater than 30000, perform symbol processing on the signal for analyzing the change trend of the signal, calculate the symbol value of each sampling point using the sign function, and correct the zero value point: when the signal value is zero, inherit the symbol value of the previous non-zero point as formula (1). If at least two zero-crossing points are not found, skip the signal. signed[k] = signed[k-1] (1)

[0037] (2) Process each peak value extracted. Find the last zero-crossing point before the first peak pair (the earliest alternating positive and negative peak value combination), and then calculate the accurate zero-crossing point time through linear interpolation to determine the depth d min of the top boundary of the hollow disease. Similarly, determine the last peak pair (the latest alternating positive and negative peak value combination), take the middle value of the sampling points of the two to determine the depth d max of the bottom boundary. If the peak value of the current signal has valid zero-crossing points before and after it, mark the column as an effective signal, and mark the two zero-crossing points with red solid dots as Figure 4 .

[0038] (3) After completing all signal processing, perform continuity verification: construct a neighborhood window with a size of 5 centered on each valid point as formula 2. If the number of valid points in the window is less than 5, mark the point as invalid. (W i = [max(1, i-2), min(c, i+2)] (2)

[0039] (4) The time delay parameter of the first peak pair is used to lock the top boundary of the void, and the time difference of the last peak pair is used to derive the bottom boundary of the void, so as to accurately locate the depth information of the single-channel void disease A-scan signal.

[0040] Step 4. As shown in Figure 5 , the minimum channel number and the maximum channel number corresponding to the A-scan signal determined as the void disease are extracted by transverse channel number analysis and marked as Figure 6 , as the transverse starting position and transverse ending position of the void area in the B-scan two-dimensional profile.

[0041] Step 5. The actual pavement material has natural unevenness and the void shape often presents irregular characteristics, so that the void sampling point position determined by each A-scan signal is not at the same horizontal depth. Therefore, it is necessary to traverse each A-scan data, select the minimum depth value of the top reflection interface and the maximum depth value of the bottom reflection interface obtained by all A-scan positioning, as the vertical starting depth and ending depth of the void area. And combined with the transverse range, the left top point coordinates and the right bottom point coordinates of the void area are constructed, so as to determine the two-dimensional area information of the void disease in each channel B-scan image as shown in Figure 7 , where c is the channel number. This type of radar has 8 channels, and steps 4 and 5 are repeated to obtain the positioning boundary box of the B-scan image of each channel as shown in Figure 8 .

[0042] Step 6. The spatial positioning of the void disease in the C-scan image needs to aggregate the B-scan coordinate information of each channel to realize: first, traverse the positioning box (c, x min , y min , x max , y max ) of the void area determined by each channel, and take the minimum transverse starting coordinate and the maximum transverse ending coordinate as the starting coordinate X min and the ending coordinate X max of the C-scan image in the x-axis direction, and determine the width range of the void area according to the starting channel number 1 and the ending channel number 8; as shown in Figure 9 , the starting depth Y min and the ending depth Y max of the C-scan image of the void disease are determined through the vertical starting and ending coordinates (the minimum starting depth y min and the maximum ending depth y max ), and the C-scan image coordinates (1, 8, X min , X max , Y min , Y max), to realize the spatial mapping of the void disease projection at a specific C-scan image depth level as shown in Figure 10

[0043] The above is only to illustrate the technical idea of the present application, and cannot limit the protection scope of the present application. Any modification made according to the technical idea of the present application on the basis of the technical scheme falls within the protection scope of the claims of the present application.​

Claims

1. A method for three-dimensional reconstruction of void defects based on radar signal time-domain peak values, characterized in that, Includes the following steps: Step 1: Data Acquisition. Use ground-penetrating radar (GPR) to scan the cement pavement with voids and obtain raw GPR echo data. Step 2: Data processing. Using the method of the applicant's patent with publication number CN115542278A, the raw GPR data collected in Step 1 is processed to obtain the processed radar data image set. Step 3: Single-channel signal peak identification and benchmark determination. Identify the time-domain peak characteristics in the single-channel A-scan signal, locate the zero-crossing points corresponding to the first and last peaks, thereby determining the starting and ending depths of the de-free interface in the signal channel, and forming a benchmark for the single-channel de-free interface position interval. Step 4: Two-dimensional localization of the voided region. Based on the results of Step 3, determine the starting trace number, ending trace number, starting depth, and ending depth of the voided region on the B-scan image, and obtain the position coordinate information representing the voided region on the two-dimensional profile. Step 5: Define the planar outline of the de-scan region. Utilize the lateral continuity feature of the B-scan image and analyze the positional correlation of the de-scan interfaces of adjacent A-scan signals to generate the positioning bounding box of the de-scan region on the horizontal plane. Step Six: 3D Model Construction. The starting and ending coordinates of the vacuolation area positioning boxes obtained from B-scan images of different measurement channels (or survey lines) are superimposed and fused in three dimensions along the depth direction (Z-axis) to form a C-scan slice structure. Finally, through spatial coordinate transformation, a three-dimensional geometric model of the vacuolation disease is constructed, generating a spatial entity containing three-dimensional dimensional parameters of length (X-axis), width (Y-axis), and depth (Z-axis).

2. The single-channel A-scan peak localization method as described in claim 1, characterized in that, In step three, the transition point between the positive and negative peaks at the beginning and end of the peak function is used to accurately locate the breakout position.

3. The three-dimensional reconstruction method for pavement void defect areas as described in claim 1, characterized in that, The frame blending method used in step six when overlaying the block diagrams is to achieve the best spatial selection effect.

Images