Road surface cavity water-containing disease detection method based on radar detection

Abstract

The invention relates to the technical field of denoising processing, in particular to a road cavity water-containing disease detection method based on radar detection, and the method comprises the steps: obtaining single-channel reflected wave data of each position on a road, obtaining a reflection sequence of each position, obtaining a front sequence and a rear sequence of each peak value, and obtaining a local smoothing coefficient of each peak value; the position sequence distribution condition of all peak values of the reflection sequence of each position is combined to obtain a reflection wave abnormal coefficient of each position, the hole confidence degree of each position is obtained according to the correlation degree between each position and the reflection sequence of the adjacent position, and the hole confidence degree of each position is calculated according to the reflection wave abnormal coefficient of each position and the hole confidence degree of each position. And the implicit noise degree of each position is obtained, and the filtering threshold value of each position is constructed by combining the signal length of the reflected wave collected at each position, so that the pavement cavity water-containing disease is detected. According to the invention, the detection precision of the pavement cavity water-containing disease can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of denoising processing, in particular to a road hollow water disease detection method based on radar detection. BACKGROUND

[0002] Due to factors such as groundwater infiltration and pipeline leakage, hollows and water diseases under the road surface become an important hidden danger affecting road traffic safety. If the hollows and water hidden under the road surface are not discovered and treated in time, serious accidents such as sudden road collapse may occur, which poses a threat to the safety of pedestrians and vehicles. Therefore, in order to effectively identify the structural characteristics and abnormal conditions underground, non-destructive detection is often needed through radar technology to prevent road collapse accidents.

[0003] In the prior art, ground penetrating radar equipment is usually used in combination with image recognition technology to detect hollow water diseases of the road surface. However, due to the influence of the detection environment on high-frequency electromagnetic waves during detection, the echo signals usually contain a large amount of noise, which needs to be suppressed to ensure the accuracy of the data. In addition, the prior art usually uses the same filtering parameters for noise removal for the collected same road surface data, but it does not fully consider that even the same road specifications have differences in underground soil quality and aboveground environment at different positions, so the noise interference received at different positions also has differences, which makes it difficult for the fixed parameter filtering method to adapt to all situations, and the noise suppression effect is not good. SUMMARY

[0004] In order to solve the above technical problems, the present application provides a road hollow water disease detection method based on radar detection to solve the existing problems.

[0005] The road hollow water disease detection method based on radar detection of the present application adopts the following technical scheme: One embodiment of the present application provides a road hollow water disease detection method based on radar detection, comprising the following steps: Obtain single-channel reflection wave data of each position on the road and obtain reflection sequences of each position; Statistically analyze the peak values in the reflection sequences of each position, construct a window with each peak value as the center to obtain the pre-sequence and post-sequence of each peak value, and obtain the local smoothing coefficient of each peak value according to the difference in the degree of dispersion between the pre-sequence and post-sequence of each peak value and the difference in the degree of change, and the correlation coefficient between the data in the window and the fitted data; Obtain the reflection wave anomaly coefficient of each position by analyzing the change and fluctuation of the data in the reflection sequence of each position, combining the position distribution of all peak values in the reflection sequence, and the average level of the local smoothing coefficient of all peak values; According to the correlation degree between each position and the reflection sequence of its adjacent position, and the difference between the local smooth coefficients of each peak in the reflection sequence of each position, the cavity confidence of each position is obtained; By the reflection wave anomaly coefficient of each position and the cavity confidence of each position, the noise degree of each position is obtained, and combined with the signal length of the reflection wave collected at each position, the filtering threshold of each position is constructed, and the reflection wave is filtered by using the denoising algorithm, and the neural network is combined to detect the pavement cavity water disease.

[0006] Preferably, the method for obtaining the pre-sequence and post-sequence of each peak comprises: For the window centered on each peak, the pre-sequence of each peak is constructed by all data points from the first data point in the window to each peak, and the post-sequence of each peak is constructed by all data points between each peak and the last data point in the window.

[0007] Preferably, the method for calculating the local smooth coefficient of each peak comprises: ; In the formula, is the local smooth coefficient of the u-th peak; is the absolute difference between the discrete degrees of the pre-sequence and the post-sequence of the u-th peak; is the absolute difference between the number of positive elements in the first difference sequence of the pre-sequence of the u-th peak and the number of negative elements in the first difference sequence of the post-sequence of the u-th peak; is the correlation between all data in the window centered on the u-th peak and its fitting data; is a constant to avoid the denominator being 0.

[0008] Preferably, the method for calculating the reflection wave anomaly coefficient of each position comprises: ; In the formula, is the reflection wave anomaly coefficient of the i-th position; is the mean value of the local smooth coefficients of all peaks in the reflection sequence of the i-th position; is the first ratio of the i-th position; is the difference between the mean value of the bit sequence of all peaks in the reflection sequence of the i-th position and the bit sequence of the central data point; is a constant to avoid the denominator being 0.

[0009] Preferably, the method for obtaining the first ratio of each position further comprises: counting the peaks and valleys in the reflection sequence of each position, and obtaining the first ratio of each position by the ratio of the sum of the number of peaks and valleys to the number of elements in the reflection sequence.

[0010] Preferably, the method for calculating the void confidence of each position is: ; In the formula, is the void confidence of the ith position; is the correlation between the reflection sequence of the ith position and the reflection sequence of its adjacent position; is the first difference value of the ith position; is a constant to avoid the denominator being 0.

[0011] Preferably, the method for obtaining the first difference value of each position is further: by calculating the local smoothing coefficient of each peak in the reflection sequence of each position, and combining the threshold method to divide the peaks into two parts, to calculate the absolute difference value between the mean values of the local smoothing coefficients of the two parts, to obtain the first difference value of each position.

[0012] Preferably, the method for calculating the noise level of each position is: ; In the formula, is the noise level of the ith position; is the reflection wave anomaly coefficient of the ith position; is the void confidence of the ith position; is a constant to avoid the denominator being 0.

[0013] Preferably, the method for calculating the filter threshold of each position is: ; In the formula, is the filter threshold of the ith position; is the noise level of the ith position; is the signal length of the reflection wave collected at the ith position; is the logarithmic function with base 10.

[0014] Preferably, the detection of the pavement void water disease in combination with the neural network further comprises: taking the radar detection profile obtained by filtering the signal as the input of the neural network to obtain the detection result of the pavement void water disease.

[0015] The present application has at least the following beneficial effects: The present application considers that the prior art fails to fully analyze the influence of noise interference difference at different positions of the same section, thereby causing the problem of poor noise suppression effect, therefore, the local smoothness coefficient is constructed to analyze the local smoothness degree of the reflected wave at each peak point, and the influence of noise interference is more accurately quantitatively evaluated; secondly, according to the construction of the reflected wave anomaly coefficient, the abnormal degree of the overall waveform of the reflected wave at different positions can be reflected, and sensitive detection of diseases and noise is realized; by calculating the correlation between the reflection sequence at each position and the reflection sequence at the adjacent position, the cavity confidence is constructed, thereby reflecting the spatial correlation characteristics of the reflected wave and the spatial continuity of the structural disease, and the accuracy of the cavity water disease identification is improved.

[0016] Further, the present application considers the noise interference degree at each position of the same section, and constructs the implicit noise degree, so as to set the most suitable threshold according to the noise interference at different positions, thereby realizing fine noise suppression at different positions and improving the accuracy of subsequent pavement cavity water disease detection. BRIEF DESCRIPTION OF DRAWINGS

[0017] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or the prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0018] Figure 1 The step flow chart of the pavement cavity water disease detection method based on radar detection provided by the present application is shown in the following figure: Figure 2 The step flow chart of the local smoothness coefficient acquisition method provided by the present application is shown in the following figure. DETAILED DESCRIPTION

[0019] In order to further illustrate the technical means and effects adopted by the present application to achieve the predetermined invention purpose, the specific implementation, structure, features and effects of the pavement cavity water disease detection method based on radar detection according to the present application are described in detail as follows by combining with the drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0020] Unless otherwise defined, such as the term "comprise", "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, such that a circuit structure, article or device that includes a list of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent in such article or device. Without more limitations, the element defined by the phrase "comprising a" does not exclude the presence of additional identical elements in the article or device comprising the element. In addition, the term "and / or" used herein includes any and all combinations of one or more related listed items. All technical and scientific terms used herein have the same meaning as understood by those skilled in the art of the technology to which the present application belongs.

[0021] The specific scheme of the radar detection-based road hollow water disease detection method provided by the present application will be specifically described below in combination with the drawings.

[0022] The radar detection-based road hollow water disease detection method provided by an embodiment of the present application, specifically, please refer to Figure 1 , comprising the following steps: Step one: acquiring single-channel reflection wave data of each position on the road, and obtaining reflection sequences of each position.

[0023] The ground penetrating radar detection device is installed to the detection vehicle. During the driving process of the detection vehicle, the transmitting antenna in the radar cavity transmits 1 channel of electromagnetic wave to the road according to the GPS positioning system every first distance, and the center frequency is 1 GHz. The distance between the radar receiving antenna and the transmitting antenna is a second distance. At the same time, the Ricker wavelet is selected as the radar pulse excitation source, and the frequency is also 1 GHz. The sampling time interval is set to a preset length, and the sampling time window length is set to a preset length. Wherein, the transmitting antenna and the receiving antenna are both at a third distance from the ground surface, and they are synchronously moved and record the single-channel reflection wave data of the ground penetrating radar. Preferably, in the present embodiment, the first distance is 0.5 m, the second distance is 3 cm, the preset length is , the preset length is , and the third distance is 0.043 m.

[0024] Through the ground penetrating radar, a certain asphalt road is detected, and single-channel reflection wave data is collected every first distance; the single-channel wave data collected at each position is equally divided and sampled, and a preset number of sampling points are uniformly extracted therefrom, which form the reflection sequence of the corresponding position in the original order. Wherein, the driving distance of the detection vehicle is set to a fourth distance, and the reflection sequence of each position during driving is constructed to detect whether the hollow water disease of the road has occurred on the road surface driven. Preferably, in the present embodiment, the sampling points of each single-channel wave are 2000, and the fourth distance is 50 m.

[0025] Step two: count the peaks in the reflection sequence of each position, and construct a window centered on each peak to obtain the pre-sequence and post-sequence of each peak. According to the difference in the degree of dispersion between the pre-sequence and the post-sequence of each peak, and the difference in the degree of change, combined with the correlation coefficient between the data in the window and the fitted data, the local smoothing coefficient of each peak is obtained.

[0026] Considering that in the same material and the same paving specification road, due to the daily use of the road and the change of environmental factors, the noise interference degree of different positions in the same road will also be different, therefore, it is necessary to evaluate the noise interference of the collected data of different positions and perform adaptive filtering, so as to improve the detection accuracy of the road cavity water disease.

[0027] In the case of good road conditions, the reflection wave of ground penetrating radar shows a smooth overall fluctuation curve. Even if a cavity water disease occurs, the reflection wave still maintains the overall smooth change trend, but because of the clutter interference, there are many peaks, and the local data on both sides of each peak still has the smooth feature. However, if noise interference occurs, it will show high-frequency, irregular spikes or local fluctuations, resulting in waveform distortion and loss of smoothness of the data.

[0028] Therefore, the reflection sequence of the i-th position is analyzed, all peaks in the reflection sequence of the i-th position are obtained by a peak detection algorithm, the possibility of the u-th peak being a noise peak point is detected, and a window of size t is constructed around the u-th peak. All data points between the first data point in the window and the u-th peak are constructed as a pre-sequence, and all data points between the u-th peak and the last data point in the window are constructed as a post-sequence. Preferably, the peak detection algorithm is not limited to extreme value detection, AMPD detection algorithm, etc. In this embodiment, the AMPD algorithm is used; t is an odd number between 7 and 15, and in this embodiment, t is 11.

[0029] Then, the absolute difference between the dispersion degree of the pre-sequence and the post-sequence of the u-th peak is calculated, which reflects whether the data fluctuation degree on both sides of the u-th peak is consistent, and whether the local data fluctuation of the u-th peak is stable. The smaller the value is, the more consistent the data fluctuation degree on both sides of the u-th peak is, and the more consistent the smooth data feature of the reflection wave is. Preferably, the calculation of the dispersion degree is not limited to variance, coefficient of variation, mean square deviation, and in this embodiment, the coefficient of variation is used.

[0030] Further, the first-order differential sequence of the pre-sequence of the u-th peak value is obtained, and the number of positive elements in the first-order differential sequence is counted, thereby reflecting the degree of continuous increase of the data on the left side of the u-th peak value. Meanwhile, the first-order differential sequence of the post-sequence of the u-th peak value is obtained, and the number of negative elements in the first-order differential sequence is counted, thereby reflecting the degree of continuous decrease of the data on the right side of the u-th peak value. Further, the absolute difference between the number of positive elements in the first-order differential sequence of the pre-sequence of the u-th peak value and the number of negative elements in the first-order differential sequence of the post-sequence of the u-th peak value is counted, thereby reflecting the smooth change effect of the data in the local range of the u-th peak value; the smaller the value, the more stable the data fluctuation in the local range of the u-th peak value.

[0031] Further, all data in the window with the u-th peak value as the center are curve-fitted, and the fitted data are sampled according to the number of data points in the window, and the correlation coefficient of the u-th peak value between the data in the window and the fitted data is calculated. The correlation coefficient is used to reflect the similarity between the data waveform represented by the local data of the u-th peak value and the fitted curve waveform; the larger the value, the more consistent the original waveform and the fitted waveform, and the smoother the waveform of the original data. Preferably, the correlation coefficient is not limited to the Pearson correlation coefficient and the Spearman correlation coefficient, and the Pearson correlation coefficient is used in the embodiment. The curve fitting is not limited to the least square fitting and the nonlinear fitting, and the least square fitting is used in the embodiment.

[0032] According to the difference and change degree difference between the discrete degrees of the pre-sequence and the post-sequence of the u-th peak value, and in combination with the correlation coefficient between the data in the window and the fitted data, the local smoothness coefficient of the u-th peak value is calculated In the embodiment, the specific calculation formula is as follows: ; In the formula, is the local smoothness coefficient of the u-th peak value; is the absolute difference between the discrete degrees of the pre-sequence and the post-sequence of the u-th peak value; is the absolute difference between the number of positive elements in the first-order differential sequence of the pre-sequence of the u-th peak value and the number of negative elements in the first-order differential sequence of the post-sequence of the u-th peak value; is the correlation between all window data and the fitted data in the window with the u-th peak value as the center; is a constant to avoid a denominator of 0, and the value range is (0, 0.01). In the embodiment, the value is 0.001, and the value has little effect on the calculation and can be ignored. The implementer can take the value by himself / herself.

[0033] Wherein, the local smoothness coefficient can reflect the smoothness of the data around the u-th peak value; the greater the value, the more stable the data fluctuation in the local range of the u-th peak value, the more consistent with the smooth change characteristics of the reflected wave, and the smaller the noise interference reflected. Wherein, the step flow chart of the method for obtaining the local smoothness coefficient provided in the embodiment is as shown in Figure 2

[0034] Step three: obtain the reflected wave anomaly coefficient of each position by the change and fluctuation of the data in the reflection sequence of each position, in combination with the bit sequence distribution of all peak values in the reflection sequence, and the average level of the local smoothness coefficients of all peak values.

[0035] Since the local smoothness coefficient can only reflect whether the local data around each peak value is smooth, it is difficult to fully reflect the overall waveform characteristics of the reflected wave. Therefore, it is necessary to combine the overall change trend of the reflected wave to make a more in-depth analysis of the i-th position.

[0036] In the case of good road conditions, the reflected wave collected by the ground penetrating radar has good smoothness, especially the waveform after the main reflected wave generated at the interface between the middle layer and the lower layer is very stable, with little fluctuation and few peak and valley values. When there is a cavity water disease, although the waveform is still smooth, there will be obvious multi-peak phenomenon, the number of peak and valley values will significantly increase and show a certain regularity. At the same time, if it is noise interference, the distribution of peak and valley values will be chaotic and the number of peak and valley values will also significantly increase. Therefore, the number and distribution of peak and valley values can be used to further reflect whether the reflected wave at the i-th position is seriously disturbed by noise or whether a cavity water disease occurs.

[0037] Through the above analysis, the sum of the peak and valley values in the reflection sequence of the i-th position is obtained by the extreme value detection algorithm, and the ratio of the sum to the number of elements in the reflection sequence is recorded as the first ratio of the i-th position. The first ratio can reflect the proportion of extreme values in the reflected wave at the i-th position, and the greater the value, the more the fluctuation times of the reflected wave collected at the i-th position. Wherein, the extreme value detection algorithm is a known technology, and the specific process will not be described again.

[0038] Further, the average of all peak bit sequence values in the reflection sequence of the i-th position is calculated, and the difference between the bit sequence average and the data point bit sequence in the reflection sequence is used to reflect whether the peak values are uniformly distributed in the reflection sequence, and the smaller the value, the more uniform the distribution of the peak values in the reflection sequence, thereby reflecting the greater the possibility that the reflected wave at the i-th position is disturbed by noise or that a cavity water disease occurs.

[0039] ​The reflection anomaly coefficient of the ith position is calculated according to the mean value of the local smoothing coefficients of all peaks in the reflection sequence of the ith position, in combination with the bit sequence distribution of all peaks in the reflection sequence of the ith position, and the first ratio of the ith position In the embodiment, the specific calculation formula is: ; In the formula, is the reflection anomaly coefficient of the ith position; is the mean value of the local smoothing coefficients of all peaks in the reflection sequence of the ith position; is the first ratio of the ith position; is the difference between the mean value of the bit sequence of all peaks in the reflection sequence of the ith position and the bit sequence of the central data point; is a constant to avoid a denominator of 0.

[0040] The reflection anomaly coefficient can reflect whether the reflection wave collected at the ith position conforms to the standard waveform characteristics under a good road condition; the greater the value, the more abnormal the reflection wave at the ith position, and the greater the possibility of a cavity water disease or the more serious the noise interference.

[0041] Step four: according to the correlation degree between the reflection sequence of each position and its adjacent position, and in combination with the difference between the local smoothing coefficients of each peak in the reflection sequence of each position, the cavity confidence of each position is obtained.

[0042] By combining the reflection waveform data of other positions, whether the ith position is a cavity water disease area can be further evaluated. When the same device is used to detect the same road, although the soil density and environmental conditions of each position may differ, the reflection waveforms between different points will still exhibit a certain similarity under normal circumstances. Even if a cavity water or other structural disease occurs, since such diseases usually have spatial continuity and aggregation characteristics, the waveform anomalies caused by them also exhibit a certain consistency in adjacent areas.

[0043] Through the above analysis, the ith position and the jth position adjacent thereto are taken as examples for analysis. The correlation degree between the reflection sequence of the ith position and the reflection sequence of the jth position is obtained through a correlation coefficient algorithm, which is used to reflect the consistency of the data changes of the ith position and its adjacent position. The greater the value, the more consistent the data change trend, thereby reflecting the greater the possibility that the ith position is a cavity water disease area. Preferably, the correlation coefficient is not limited to the Pearson correlation coefficient or the Spearman correlation coefficient, and the Pearson correlation coefficient is used in the embodiment.

[0044] Further, although the cavity water disease causes more peaks to appear, the local smoothness coefficients between all peaks are still relatively consistent due to the local smoothness of each peak. The local smoothness coefficients of each peak in the reflection sequence of the ith position are calculated and used as the input of the Otsu threshold method. The peaks are divided into two parts by the Otsu threshold method, and the absolute difference between the average local smoothness coefficients of the two parts is calculated, which is recorded as the first difference value of the ith position. The smaller the first difference value is, the smaller the difference in the local smoothness between the peaks in the reflection sequence is.

[0045] Therefore, according to the correlation between the reflection sequence of the ith position and the reflection sequence of its adjacent position, combined with the first difference value of the ith position, the cavity confidence of the ith position is calculated In this embodiment, the specific calculation formula is: ; In the formula, is the cavity confidence of the ith position; is the correlation between the reflection sequence of the ith position and the reflection sequence of its adjacent position; is the first difference value of the ith position; is a constant to avoid a denominator of 0.

[0046] The cavity confidence can reflect the possibility of the ith position being a cavity water disease area. The larger the value is, the more likely the ith position is a cavity disease water area. The smaller the value is, the more likely the ith position is not a cavity disease area or the degree of noise interference received by the ith position is greater.

[0047] Step five: obtain the noise degree of each position by the reflection wave anomaly coefficient of each position and the cavity confidence of each position, construct the filtering threshold of each position combined with the signal length of the reflection wave collected at each position, and use the denoising algorithm to filter the reflection wave and combine the neural network to detect the cavity water disease of the road surface.

[0048] In order to further analyze the degree of noise interference received by the ith position, the reflection wave anomaly coefficient of the ith position and the cavity confidence of the ith position are combined to construct the noise degree of the ith position In this embodiment, the specific calculation formula is: ; In the formula, is the noise degree of the ith position; is the reflection wave anomaly coefficient of the ith position; is the cavity confidence of the ith position; is a constant to avoid denominator zero.

[0049] wherein the implicit noise degree can reflect the abnormal degree of the reflection wave collected at the i-th position, the greater the value, the more the reflection waveform does not conform to the standard waveform characteristics, and the greater the possibility that it does not belong to the cavity disease characteristic waveform, and the more the noise data contained in the data collected at the i-th position, and the more serious the noise interference.

[0050] Since the setting of the wavelet threshold needs to be considered when denoising by the wavelet denoising algorithm, if the threshold is large, in addition to suppressing noise, it may cause the loss of radar echo signal characteristics; if the threshold is too small, it may not be able to completely remove the noise. Therefore, the embodiment optimizes the selection of the threshold according to the implicit noise degree of the data collected at each position. Taking the i-th position as an example, the filtering threshold of the i-th position is constructed by the implicit noise degree of the i-th position and the signal length of the reflection wave collected at the i-th position. In the embodiment, the specific calculation formula is: ; wherein, is the filtering threshold of the i-th position; is the implicit noise degree of the i-th position; is the signal length of the reflection wave collected at the i-th position; is the logarithmic function with base 10; in the embodiment, the signal length of the reflection wave is collected, not the data length of the reflection sequence.

[0051] the filtering threshold of the i-th position and the wavelet denoising algorithm is used to filter the reflection wave collected at the i-th position. Then the filtered signal is presented in the form of radar detection profile by the control system of the ground penetrating radar. Then the radar detection profile is used as the algorithm input of the neural network to detect and identify the cavity water-containing disease, so as to evaluate whether the i-th position of the road surface has the cavity water-containing disease. In the same way, all positions during driving are detected, so as to realize the disease detection of the cavity water-containing phenomenon of the road surface. The wavelet denoising algorithm is a known technology, and the specific process is not described again.

[0052] Preferably, in the embodiment, the training and use of the neural network are as follows: Collect 350 Z (take) the detection profile of the ground penetrating radar of the good road state, 150 radar detection profile of the road with the cavity water defect; train the CNN convolutional neural network through all the collected images, wherein the pooling layer of the CNN convolutional neural network adopts the maximum pooling, and adopts the Tanh function as the activation function, and obtains the trained neural network model. The radar detection profile filtered in the embodiment is used as the input of the CNN convolutional neural network, and the output of the neural network is the binary classification result of the radar detection profile as the cavity water defect, that is, the existence or nonexistence of the cavity water disease, so as to realize the disease detection and identification of the position. The training and use of the neural network are known technologies, and the specific process will not be described here.

[0053] It can be understood that the reference to "one embodiment" or "some embodiments" and the like described in the present application specification means that the specific features, structures or characteristics described in connection with the embodiment are included in one or more embodiments of the present application. Therefore, if "in one embodiment", "in some embodiments", "in other some embodiments", "in other some embodiments" and the like appear in the specification, it does not necessarily refer to the same embodiment, but means "one or more but not all embodiments", unless otherwise specifically emphasized. The terms "include", "contain", "have" and their variants mean "include but not limited to", unless otherwise specifically emphasized.

[0054] It should be noted that the above sequence of the embodiments of the present application is only for description, not representing the advantages and disadvantages of the embodiments. And the above description is made for the specific embodiments of the present application. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are possible or can be advantageous. At the same time, the size of the serial number of each step in the embodiment does not mean the execution order, and the execution order of each process should be determined by its function and inherent logic, and should not constitute any limitation on the implementation process of the embodiments in the specification.

[0055] The above embodiments are only used to illustrate the technical solutions of the present application, but not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for some technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.

Claims

1. A method for detecting water-containing cavities in road surfaces based on radar detection, characterized in that: The following steps are involved: Obtain single-channel reflection wave data at each location on the road and obtain the reflection sequence at each location; The peak values ​​in the reflection sequence at each position are counted, and a window is constructed with each peak as the center to obtain the pre-sequence and post-sequence of each peak. According to the difference in the degree of dispersion and the degree of change between the pre-sequence and post-sequence of each peak, the correlation coefficient between the data in the window and its fitted data is combined to obtain the local smoothing coefficient of each peak. The reflection wave anomaly coefficient at each location is obtained by combining the fluctuation of the data in the reflection sequence at each location, the position distribution of all peaks in the reflection sequence, and the average level of the local smoothing coefficients of all peaks; The hole confidence of each position is obtained based on the correlation between the reflection sequences of each position and its adjacent positions, and the difference between the local smoothing coefficients of each peak in the reflection sequence of each position. The implicit noise level at each location is obtained through the reflection wave anomaly coefficient and the cavity confidence at each location. Combined with the signal length of the reflection wave collected at each location, the filtering threshold at each location is constructed. The reflection wave is filtered using a denoising algorithm, and the pavement cavity water-containing disease is detected in combination with a neural network.

2. The method for detecting water-containing pavement voids based on radar detection according to claim 1, characterized in that: The method for obtaining the pre-sequence and post-sequence of each peak is as follows: For the window centered on each peak, all data points from the first data point to each peak in the window are used to construct the pre-sequence of each peak, and all data points from each peak to the last data point in the window are used to construct the post-sequence of each peak.

3. The method for detecting water-containing pavement voids based on radar detection according to claim 1, characterized in that: The calculation method of the local smoothing coefficient of each peak is: ； Where, is the local smoothing coefficient of the u-th peak; is the absolute difference between the discrete degrees of the preceding sequence and the following sequence of the u-th peak; is the absolute difference between the number of positive elements in the first-order difference sequence of the preceding sequence and the number of negative elements in the first-order difference sequence of the following sequence of the u-th peak; is the correlation between the data in all windows and their fitted data in the window centered at the u-th peak; This is to avoid the denominator being a constant of 0.

4. The method for detecting water-containing pavement voids based on radar detection according to claim 1, wherein: The calculation method of the reflection wave anomaly coefficient at each position is: ； Where, is the reflection wave anomaly coefficient at the i-th position; is the mean of all peak local smoothing coefficients in the reflection sequence at the i-th position; is the first ratio of the i-th position; is the difference between the position mean of all peaks in the reflection sequence at position i and the position of its central data point; This is to avoid the denominator being a constant of 0.

5. The method for detecting water-containing pavement voids based on radar detection according to claim 4, characterized in that: The method for obtaining the first ratio of each position further includes: counting the peaks and valleys in the reflection sequence of each position, and calculating the ratio of the sum of the number of peaks and valleys to the number of elements in the reflection sequence to obtain the first ratio of each position.

6. The method for detecting water-containing pavement voids based on radar detection according to claim 1, wherein: The calculation method of the hole confidence of each position is: ； Where, is the hole confidence of the i-th position; is the correlation between the reflection sequence of the i-th position and the reflection sequence of its adjacent positions; is the first difference at the i-th position; This is to avoid the denominator being a constant of 0.

7. The method for detecting water-containing pavement voids based on radar detection according to claim 6, characterized in that: The method for obtaining the first difference value of each position is further as follows: by calculating the local smoothing coefficient of each peak in the reflection sequence of each position, and combining the threshold method to divide the peak into two parts, so as to calculate the absolute difference between the means of the local smoothing coefficients of the two parts, the first difference value of each position is obtained.

8. The method for detecting water-containing pavement voids based on radar detection according to claim 1, wherein: The calculation method of the implicit noise level at each position is: Where, is the implicit noise level of the i-th position; is the reflection wave anomaly coefficient at the i-th position; is the hole confidence of the i-th position; This is to avoid the denominator being a constant of 0.

9. The method for detecting water-containing pavement voids based on radar detection according to claim 1, wherein: The calculation method of the filtering threshold at each position is: Where, is the filtering threshold at the i-th position; is the implicit noise level of the i-th position; is the signal length of the reflected wave collected at the i-th position; It is a logarithmic function with base 10.

10. The method for detecting water-containing pavement voids based on radar detection according to claim 1, wherein: The method of detecting water-containing cavity defects in road surface by combining with a neural network further includes: using a radar detection profile obtained from a filtered signal as an input of the neural network to obtain a detection result of water-containing cavity defects in road surface.

Images