Large-gap pavement structure depth detection method based on sound processing technology

Abstract

The invention provides a large-gap pavement structure depth detection method based on a sound processing technology, and the method comprises the steps: carrying out the preprocessing of a sound signal based on the sound processing technology, guaranteeing the accuracy of the signal, carrying out the calibration according to a plurality of reference values in a structure depth detection process, and taking the distance between a microphone and a knocking point as an impact factor. The sound characteristic parameters of the receiving equipment are uniformly evaluated in the same distance environment, errors caused by different distance influence factors are eliminated, and the result precision can be greatly improved. According to the position of the microphone and the sound path distance difference of sound transmission, the coordinate of the position to be detected can be accurately positioned through establishment of the data model, the workload of obtaining the information of the position to be detected is avoided, the road surface pollution caused by detection means such as a sanding method is avoided, and the detection and evaluation efficiency is greatly improved; the calculation result can provide a theoretical basis for pavement disease cause analysis, construction condition evaluation, project acceptance and the like, and the technical content is simple to operate and easy to implement.

Description

Technical Field

[0001] The invention relates to the field of sound or sound processing technology, and in particular to the application of sound processing technology in large-void pavement structure depth detection. The invention proposes a large-void pavement structure depth detection method based on sound processing technology. Technical Background

[0002] Large-void pavement is an asphalt pavement that can quickly drain rainwater from the road surface. It is characterized by more coarse aggregate and a large void ratio. Rainwater is quickly discharged from the road surface through the interconnected voids inside its structure, thereby eliminating water accumulation on the road surface. Therefore, it is also called a porous anti-skid layer or a multi-void pavement. Large-void pavement has more coarse aggregate. During the construction process, steel wheels are generally used for rolling. After rolling, larger voids are formed on the surface, which has better anti-skid properties and ensures its water permeability after construction. Therefore, the surface texture condition of large voids is an important indicator for evaluating its construction quality and anti-skid drainage function. The structural depth is usually used as an evaluation indicator for the surface texture condition.

[0003] The structural depth prediction method based on the full-scale asphalt pavement full life cycle test and feedback neural network (CN116956583A) is to build a full-scale pavement in an open-air environment with the same materials and structure as the intended pavement, load it on the full-scale pavement, and after a certain amount of axle load is accumulated, use a vehicle-mounted laser texture depth meter to measure the structural depth. Loading is continued until the pavement is loaded throughout its life cycle, and atmospheric parameters are collected. According to the atmospheric data, vehicle axle load and structural depth test data, a feedback neural network model is established to predict the structural depth value of the pavement after the equivalent axle load acts several times and the atmospheric temperature acts after it is opened to traffic.

[0004] A road surface structure depth detection method based on laser vision (CN116716778A) obtains the three-dimensional texture and structure depth of the road surface structure depth detection required, establishes a corresponding structure depth and texture structure relationship model based on a CNN network or a DNN network, divides the texture structure into several texture distribution areas, sets a weight coefficient for each area, obtains the texture three-dimensional model of the inspection area to be inspected through the device, and calculates the structure depth of the inspection area to be inspected by weighted average through the weight coefficient of the texture three-dimensional model in the texture distribution area and the structure depth of the area.

[0005] A method for detecting the structural depth of an asphalt pavement (CN117218089B) obtains images from multiple perspectives to obtain a depth map of the image, establishes a depth fitting plane, and calculates the structural depth of a point to be detected based on the number of image pixels at the point to be detected, the absolute depth value of the pixels, and the absolute structural depth of a texture reference surface.

[0006] It can be seen from the above literature that the existing structural depth detection methods all obtain the texture information of the road surface through equipment, establish a road texture relationship model that affects the structural depth, and estimate the detection value of the structural depth through the model. The existing literature all uses image processing technology to detect the structural depth.

[0007] A large number of literature studies have shown that different surface textures of objects will produce different sound signals when knocked. It is completely feasible to obtain road surface texture information through sound recognition technology and determine its surface structural depth through road surface texture information. In the existing literature, the detection of road surface structural depth requires the location information to be calibrated in advance, and then the detection of the points to be checked is carried out. In the calculation model, the results are calibrated with a single relationship, and the error of the calculation result is large. At the same time, the scope involved in the detection process of the structural depth of the points to be checked is unclear, and the detection results cannot be evaluated according to the methods in the specifications, which can easily cause distortion of the results. The present invention proposes a method based on sound signal processing technology to process the collected sound signals and establish a relationship model between the sound signal, the structural depth and the position information of the points to be checked. It has the advantages of being simple, fast, and ready for results. Therefore, it is urgent and necessary to propose a large-void road surface structural depth detection method based on sound processing technology. Summary of the invention

[0008] The present invention is based on sound processing technology. A unified method is adopted for collecting sound signals of the points to be inspected and the reference points, which ensures the consistency of the structural depth detection of the points to be inspected with the standard method at the source. The sound signals of the points to be inspected are processed, and the position information of the points to be inspected is obtained through sound signal recognition equipment. Multiple sound signal and structural depth relationship models are established, and parameters that affect the detection results are standardized to eliminate errors caused by factors of different scales. The method has the advantages of objectivity, high efficiency, and accuracy.

[0009] The technical problem to be solved by the present invention is achieved by adopting the following technical solutions:

[0010] The method for detecting the depth of large-void pavement structure based on sound processing technology includes the following steps:

[0011] Step 1: Select a large gap road surface to be detected and deploy a microphone sound signal collection device MIC1;

[0012] Step 2: Select several reference points, use a small hammer to knock on the reference point position of the large gap road surface, and use a microphone to collect the sound signal generated by the knocking, perform signal processing on the audio collected by MIC1, extract the corresponding characteristic parameters, and use the sand spreading method to determine the construction depth of the reference point. According to the reference point characteristic parameters and the construction depth, the audio characteristic parameters and the construction depth reference value are calculated;

[0013] Step 3: On the large-gap road surface where the structural depth needs to be detected, two microphone sound signal collection devices MIC2 and MIC3 are arranged at a certain distance. A small hammer hits the large-gap road surface, and the audio collected by MIC2 and MIC3 is processed to extract the corresponding characteristic parameters. The coordinate position of the detection point is calculated according to the sound path difference between MIC1 and MIC2 and MIC3;

[0014] Step 4: Establish a structural depth calculation model and calculate the structural depth of the large-void pavement detection point based on the model.

[0015] Further technology of the present invention:

[0016] Preferably, the audio is subjected to signal processing to extract corresponding characteristic parameters in the following steps:

[0017] (1) Preprocessing the collected audio signal, wherein the preprocessing includes windowing and short-time sound signal frequency conversion;

[0018] (2) Perform fast Fourier transform on the processed sound signal to obtain spectrum information;

[0019]

[0020] Where: x(n) is the preprocessed sound signal, M is the number of points, which is 256;

[0021] X(i)—Fourier transform to obtain spectrum signal;

[0022] (3) Calculate the logarithmic energy of each frame of spectral line energy after Mel filtering:

[0023]

[0024]

[0025] Where: f(m)—center frequency of the mth filter, Hz; i—frequency, Hz;

[0026] f(m-1)—m-1th filter center frequency, Hz;

[0027] f(m+1)—center frequency of the m+1th filter, Hz;

[0028] H m (i) – Mel filter transfer function;

[0029] S(m)——logarithmic energy of each filter output;

[0030] (4) Perform discrete cosine transform on S(m) to obtain MFCC feature parameters.

[0031] Preferably, the calculated audio feature parameters and the constructed depth reference value are performed according to the following formula:

[0032]

[0033] Where:

[0034] MFCC(a)—MFCC feature parameters after the hammer hits the reference point and is processed, db;

[0035] MFCC b —Baseline MFCC feature parameters, db; D b —reference distance, m;

[0036] TD b —Benchmark construction depth, mm; TD(a)—the construction depth of the benchmark point measured by the sand laying method, mm; D(a)—the distance from the sound-taking point to the striking position of the benchmark point, m; j—the number of benchmark points.

[0037] Preferably, the detection point coordinate position is calculated according to the acoustic path difference between MIC1 and MIC2 and MIC3 according to the following formula:

[0038] Y1(t)=S(t-τ1)+n1(t)

[0039] Y2(t)=S(t-τ2)+n2(t)

[0040] Y3(t)=S(t-τ3)+n3(t)

[0041] R 12 (τ)=E[Y1(t)Y2(t-τ)]

[0042] R 13 (τ)=E[Y1(t)Y3(t-τ)]

[0043] τ 12 = argmax τ R 12 (τ)

[0044] τ 13 = argmax τ R 13 (τ)

[0045]

[0046] Where: Y1(t)—microphone MIC1 signal model;

[0047] Y2(t)—microphone MIC2 signal model;

[0048] Y3(t)—microphone MIC3 signal model; t—time, s;

[0049] S(t-τ1)——pure signal model of microphone MIC1;

[0050] n1(t)——Noise signal model of microphone MIC1;

[0051] S(t-τ2)——pure signal model of microphone MIC2;

[0052] n2(t)—noise signal model of microphone MIC2;

[0053] S(t-τ3)——microphone MIC3 pure signal model;

[0054] n3(t)——Noise signal model of microphone MIC3;

[0055] c—speed of sound in air, m / s;

[0056] R 12 (τ)—Y1(t), Y2(t) correlation function;

[0057] R 13 (τ)—Y1(t), Y3(t) correlation function;

[0058] τ 12 —The sound path difference between MIC1 and MIC2, s; τ 13 —The sound path difference between MIC1 and MIC3, s;

[0059] cosθ 12 —Azimuth of the sound source relative to MIC1 and MIC2, °;

[0060] cosθ 13 —Azimuth of the sound source relative to MIC1 and MIC3, °; X j —Construct the horizontal coordinate of the depth detection point;

[0061] Y j —Vertical coordinate of the construction depth detection point; X1—horizontal coordinate of the microphone MIC1 point;

[0062] Y1—vertical coordinate of microphone MIC1 point; X2—horizontal coordinate of microphone MIC2 point;

[0063] Y2—vertical coordinate of microphone MIC2 point;

[0064] X3—microphone MIC3 point abscissa; Y3—microphone MIC3 point ordinate;

[0065] L 13 ——The distance between microphone MIC1 and microphone MIC3, m;

[0066] L12 ——The distance between microphone MIC1 and microphone MIC2, m.

[0067] Preferably, a structural depth calculation model is established according to the following formula, and the structural depth of the large gap road surface detection point is calculated according to the model:

[0068]

[0069] In the formula: TD j1 — Construction depth measured by microphone MIC1, mm;

[0070] TD j2 — Construction depth measured by microphone MIC2, mm;

[0071] TD j3 — Construction depth measured by microphone MIC3, mm;

[0072] MFCC1—microphone MIC1 characteristic parameters, db;

[0073] MFCC2—microphone MIC2 characteristic parameters, db;

[0074] MFCC3—microphone MIC3 characteristic parameters, db;

[0075] D1—the distance between microphone MIC1 and the tapping point, m;

[0076] D2—the distance between microphone MIC2 and the tapping point, m;

[0077] D3—the distance between microphone MIC3 and the tapping point, m;

[0078] Δd1—D1 and D b The absolute value of the difference between them, m;

[0079] Δd2—D2 and D b The absolute value of the difference between them, m;

[0080] Δd3—D3 and D b The absolute value of the difference between them, m;

[0081] TD j —Structural depth of the knock point test, mm.

[0082] Preferably, the reference points are arranged within a range of 7.5 to 10 m from the microphone MIC1, and the number of the reference points is not less than 3.

[0083] Preferably, the void ratio of the large-void pavement is not less than 18%.

[0084] The beneficial technical effects of the present invention are:

[0085] The large-void pavement structural depth detection method of the present invention is based on sound processing technology. Sound processing technology is adopted and sound signals are preprocessed to ensure the accuracy of the signals. In the process of structural depth detection, multiple reference values ​​are set for calibration, and signal characteristic parameters of multiple receiving devices are uniformly judged in the same distance environment. The distance between the microphone and the knocking point is used as an influencing factor to determine the structural depth detection value, which can greatly improve the accuracy of the results.

[0086] The large-void pavement structural depth detection method of the present invention is based on sound processing technology. The sound processing technology is used to establish a data model according to the microphone position and the sound path difference of sound propagation. The coordinates of the position to be detected can be accurately located, thereby avoiding the workload of obtaining the information of the position to be detected. The structural depth distribution of any position of the road surface can be determined in real time, which is convenient for timely improvement during the construction process.

[0087] The large-void pavement structure depth detection method of the present invention based on sound processing technology is a non-destructive, rapid and pollution-free detection means, which avoids the pollution of the road surface by detection means such as sand laying method, and adopts voice processing technology, thereby greatly improving the efficiency of detection and evaluation. The calculation results can provide a theoretical basis for the analysis of the causes of pavement diseases, the assessment of construction conditions, and project acceptance. The technical content is simple to operate and easy to implement. DETAILED DESCRIPTION

[0088] In order to make the technical means, creative features, objectives and effects achieved by the present invention easy to understand, the present invention is further explained below in conjunction with specific embodiments.

[0089] All the raw materials in the present invention are not particularly limited in their sources and can be purchased from the market or prepared according to conventional methods well known to those skilled in the art.

[0090] There is no particular limitation on the purity of all raw materials in the present invention. The present invention preferably uses analytically pure materials or materials of conventional purity used in the field of composite materials.

[0091] The mix ratio of large-void pavement is designed indoors, and the void ratio is controlled at 20%. The designed large-void pavement mixture is paved. After paving, the structural depth is tested. The main steps include the following:

[0092] The method for detecting the depth of a large-void pavement structure based on sound processing technology is characterized by comprising the following steps:

[0093] Step 1: Select the road surface with large gaps to be detected and deploy a microphone sound signal collection device MIC1.

[0094] Step 2: Select 4 reference points on the road surface, which are 7.5m, 8.0m, 8.5m, and 9.0m away from the acquisition device MIC1. Use a small hammer to hit the large gap road surface reference point, and use a microphone to collect the sound signal generated by the knocking. Perform signal processing on the audio collected by MIC1 according to the following method to extract the corresponding feature parameters:

[0095] The collected audio signal is preprocessed, which includes windowing processing and short-time sound signal frequency conversion.

[0096] Perform fast Fourier transform on the processed sound signal to obtain spectrum information.

[0097]

[0098] Where: x(n) is the preprocessed sound signal, M is the number of points, which is 256;

[0099] X(i)—Fourier transform to obtain spectrum signal;

[0100] Calculate the logarithmic energy of each frame spectral line energy after Mel filtering:

[0101]

[0102] Where: f(m)—center frequency of the mth filter, Hz; i—frequency, Hz;

[0103] f(m-1)—m-1th filter center frequency, Hz;

[0104] f(m+1)—center frequency of the m+1th filter, Hz;

[0105] H m (i) – Mel filter transfer function;

[0106] S(m)——logarithmic energy of each filter output;

[0107] S(m) is subjected to discrete cosine transform, and the MFCC feature parameters obtained are 72.2db, 72.3db, 70.7db, and 70.1db respectively.

[0108] The sand spreading method was used to determine the structural depths of the reference points, which were 1.12mm, 1.21mm, 1.05mm, and 0.97mm, respectively. Based on the characteristic parameters of the reference points and the structural depths, the audio characteristic parameters and structural depth reference values ​​were calculated according to the following formula:

[0109]

[0110]

[0111] Where:

[0112] MFCC(a)—MFCC feature parameters after the hammer hits the reference point and is processed, db;

[0113] MFCC b —Baseline MFCC feature parameters, db; D b —reference distance, m; TD b —reference construction depth, mm;

[0114] TD(a)—the structural depth of the reference point measured by the sand spreading method, mm; D(a)—the distance from the sound taking point to the striking position of the reference point, m; j—the number of reference points;

[0115] Step 3: On the large gap road surface where the structural depth needs to be detected, two microphone sound signal collection devices MIC2 and MIC3 are arranged at a certain distance. The coordinates of the three microphones are as follows:

[0116] Table 1 Microphone coordinates

[0117] Location Horizontal axis Vertical axis MIC1 842 605 MIC2 842 742 MIC3 842 788

[0118] A small hammer hits the road surface with large gaps, and the audio collected by MIC2 and MIC3 is processed to extract the corresponding feature parameters as follows:

[0119] Table 2 Microphone receiving signal characteristic parameters

[0120] Location Characteristic parameters MIC1 72.2db MIC2 68.3db MIC3 66.5db

[0121] According to the acoustic path difference between MIC1 and MIC2 and MIC3, the coordinate position of the detection point is calculated according to the following formula:

[0122] Y1(t)=S(t-τ1)+n1(t)

[0123] Y2(t)=S(t-τ2)+n2(t)

[0124] Y3(t)=S(t-τ3)+n3(t)

[0125] R 12 (τ)=E[Y1(t)Y2(t-τ)]

[0126] R 13 (τ)=E[Y1(t)Y3(t-τ)]

[0127] τ 12 = argmax τ R 12 (τ)

[0128] τ 13 = argmaxτ R 13 (τ)

[0129]

[0130] According to the above formula, the speed of sound is 340m / s, and the coordinates of the tapping point are calculated as (905, 602). The distance between the tapping point and the microphone is as follows:

[0131] parameter Distance(m) <![CDATA[D1]]> 6.3 <![CDATA[D2]]> 15.4 <![CDATA[D3]]> 19.6

[0132] Step 4: Establish a structural depth calculation model and calculate the structural depth of the large-void pavement detection point based on the model.

[0133]

[0134] According to the method specified in the "Highway Roadbed and Pavement Field Inspection Regulations", the structural depth of the inspection point was measured by manual sand laying method, and the result was TD=1.06mm, which was 1.6% different from the inspection result of the above inspection method, indicating that this inspection method has good applicability.

[0135] The above shows and describes the basic principles, main features and characteristics of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments, and the above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of the present invention to be protected is defined by the attached claims and their equivalents.

Claims

1. A method for detecting the depth of a large-void pavement structure based on sound processing technology, characterized in that: The following steps are involved: Step 1: Select a large gap road surface to be detected and deploy a microphone sound signal collection device MIC1; Step 2: Select several reference points, use a small hammer to knock on the reference point position of the large gap road surface, and use a microphone to collect the sound signal generated by the knocking, perform signal processing on the audio collected by MIC1, extract the corresponding characteristic parameters, and use the sand spreading method to determine the construction depth of the reference point. According to the reference point characteristic parameters and the construction depth, the audio characteristic parameters and the construction depth reference value are calculated; Step 3: On the large-gap road surface where the structural depth needs to be detected, two microphone sound signal collection devices MIC2 and MIC3 are arranged at a certain distance. A small hammer hits the large-gap road surface, and the audio collected by MIC2 and MIC3 is processed to extract the corresponding characteristic parameters. The coordinate position of the detection point is calculated according to the sound path difference between MIC1 and MIC2 and MIC3; Step 4: Establish a structural depth calculation model and calculate the structural depth of the large-void pavement detection point based on the model.

2. The method for detecting the depth of large-void pavement structure based on sound processing technology according to claim 1, characterized in that: Audio signal processing and extraction of corresponding feature parameters are carried out in the following steps: (1) Preprocessing the collected audio signal, wherein the preprocessing includes windowing and short-time sound signal frequency conversion; (2) Perform fast Fourier transform on the processed sound signal to obtain spectrum information; Where: x(n) is the preprocessed sound signal, M is the number of points, which is 256; X(i)—Fourier transform to obtain spectrum signal; (3) Calculate the logarithmic energy of each frame of spectral line energy after Mel filtering: Where: f(m)—center frequency of the mth filter, Hz; i—frequency, Hz; f(m-1)—m-1th filter center frequency, Hz; f(m+1)—center frequency of the m+1th filter, Hz; H m (i) – Mel filter transfer function; S(m)——logarithmic energy of each filter output; (4) Perform discrete cosine transform on S(m) to obtain MFCC feature parameters.

3. The method for detecting the depth of large-void pavement structure based on sound processing technology according to claim 2, characterized in that: The calculated audio feature parameters and constructed depth reference values ​​are performed according to the following formula: Where: MFCC(a)—MFCC feature parameters after the hammer hits the reference point and is processed, db; MFCC b —Baseline MFCC feature parameters, db; D b —reference distance, m; TD b —Benchmark construction depth, mm; TD(a)—the construction depth of the benchmark point measured by the sand laying method, mm; D(a)—the distance from the sound-taking point to the striking position of the benchmark point, m; j—the number of benchmark points.

4. The method for detecting the depth of large-void pavement structure based on sound processing technology according to claim 3 is characterized in that: According to the acoustic path difference between MIC1 and MIC2 and MIC3, the coordinate position of the detection point is calculated according to the following formula: Y1(t)=S(t-τ1)+n1(t) Y2(t)=S(t-τ2)+n2(t) Y3(t)=S(t-τ3)+n3(t) R 12 (τ)=E[Y1(t)Y2(t-τ)] R 13 (τ)=E[Y1(t)Y3(t-τ)] t 12 =argmax τ R 12 (t) t 13 =argmax τ R 13 (t) Where: Y1(t)—microphone MIC1 signal model; Y2(t)—microphone MIC2 signal model; Y3(t)—microphone MIC3 signal model; t—time, s; S(t-τ1)——pure signal model of microphone MIC1; n1(t)——Noise signal model of microphone MIC1; S(t-τ2)——pure signal model of microphone MIC2; n2(t)—noise signal model of microphone MIC2; S(t-τ3)——microphone MIC3 pure signal model; n3(t)——Noise signal model of microphone MIC3; c—speed of sound in air, m / s; R 12 (τ)—Y1(t), Y2(t) correlation function; R 13 (τ)—Y1(t), Y3(t) correlation function; τ 12 —The sound path difference between MIC1 and MIC2, s; τ 13 —The sound path difference between MIC1 and MIC3, s; cosθ 12 —Azimuth of the sound source relative to MIC1 and MIC2, °; cosθ 13 —Azimuth of the sound source relative to MIC1 and MIC3, °; X j —Construct the horizontal coordinate of the depth detection point; Y j —Construct the ordinate of the depth detection point; X1—horizontal coordinate of microphone MIC1 point; Y1—vertical coordinate of microphone MIC1 point; X2—horizontal coordinate of microphone MIC2 point; Y2—vertical coordinate of microphone MIC2 point; X3—microphone MIC3 point abscissa; Y3—microphone MIC3 point ordinate; L 13 ——The distance between microphone MIC1 and microphone MIC3, m; L 12 ——The distance between microphone MIC1 and microphone MIC2, m.

5. The method for detecting the depth of large-void pavement structure based on sound processing technology according to claim 4 is characterized in that: The structural depth calculation model is established according to the following formula, and the structural depth of the large gap road surface detection point is calculated based on the model: In the formula: TD j1 — Construction depth measured by microphone MIC1, mm; TD j2 — Construction depth measured by microphone MIC2, mm; TD j3 — Construction depth measured by microphone MIC3, mm; MFCC1—microphone MIC1 characteristic parameters, db; MFCC2—microphone MIC2 characteristic parameters, db; MFCC3—microphone MIC3 characteristic parameters, db; D1—the distance between microphone MIC1 and the tapping point, m; D2—the distance between microphone MIC2 and the tapping point, m; D3—the distance between microphone MIC3 and the tapping point, m; Δd1—D1 and D b The absolute value of the difference between them, m; Δd2—D2 and D b The absolute value of the difference between them, m; Δd3—D3 and D b The absolute value of the difference between them, m; TD j —Structural depth of the knock point test, mm.

6. The method for detecting the depth of large-void pavement structure based on sound processing technology according to claim 5 is characterized in that: The reference points are arranged at points within a range of 7.5 to 10 m from the microphone MIC1, and the number of the reference points is not less than 3.

7. The method for detecting the depth of large-void pavement structure based on sound processing technology according to claim 1 is characterized in that: The void ratio of the large-void pavement is not less than 18%.

Images