Tunnel nondestructive testing method based on seismic wave data and storage medium

Abstract

The invention discloses a tunnel nondestructive testing method based on seismic wave data and a storage medium, and the method comprises the steps: S1, receiving an actual measurement signal from a site, and synchronously inputting a signal processing wavelet; s2, introducing an iterative time-frequency analysis method to carry out iterative time-frequency analysis on the actually measured signals and the input signal processing wavelets; s3, inputting signal processing wavelets of different dominant frequencies, and respectively carrying out iterative time-frequency analysis; s4, selecting the dominant frequency of the signal processing wavelet with the least iteration times from the output time-frequency analysis result to output the optimal solution spectrum of the time-frequency analysis result; and S5, identifying the optimal solution spectrum of the time-frequency analysis result to obtain a tunnel nondestructive testing result. According to the method, on the basis of linear time-frequency analysis, L1 norm constraint is introduced, and the time degree-of-freedom analysis method is realized from the angle of inversion, so that the influence of multiple solutions is reduced, and the resolution and focusing performance of nondestructive testing are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of tunnel nondestructive testing, and in particular relates to a tunnel nondestructive testing method and a storage medium based on seismic wave data. Background Art

[0002] Tunnels are important infrastructure in my country. During the construction process, they often encounter various unfavorable geological areas, which lead to geological disasters such as landslides, water inrush, rock bursts, and gas explosions. Most tunnels have overcome the above disasters during the construction process. However, even if the above geological disasters do not occur, the quality of the tunnel lining may also have quality problems in the process of various unfavorable geological treatments. With the changes in geological conditions, the tunnel lining may also suffer from internal damage, leakage and other problems;

[0003] Nondestructive testing of tunnel lining quality is becoming increasingly important. It can detect defective linings in time to prevent them from expanding, and can also retest the damaged parts to determine the root cause of the damage and propose accurate treatment methods for the damage.

[0004] As a traditional geophysical exploration method, seismic exploration is of great significance in geological exploration due to its large detection depth and high resolution. The seismic attribute information obtained from seismic data can reflect a lot of hidden information. For example, the time-frequency analysis method is an important seismic attribute extraction method, which can simultaneously describe the energy density and intensity of earthquakes at different times and frequencies, and is of great significance for geological interpretation.

[0005] However, traditional time-frequency analysis methods have problems such as low resolution and insufficient focus. Summary of the invention

[0006] The purpose of the present invention is to overcome the defects of the prior art and provide a tunnel nondestructive testing method and storage medium based on seismic wave data. Based on linear time-frequency analysis, the method implements a time degree of freedom analysis method from the perspective of inversion, introduces L1 norm constraints to reduce the influence of multiple solutions, and improves the resolution and focusing of nondestructive testing.

[0007] The object of the present invention is achieved through the following technical solutions:

[0008] A tunnel nondestructive testing method based on seismic wave data comprises the following steps:

[0009] Step S1, receiving a measured signal from the field, and synchronously inputting a signal processing wavelet.

[0010] Step S2, introducing an iterative time-frequency analysis method to perform iterative time-frequency analysis on the measured signal and the input signal processing wavelet;

[0011] Step S3, inputting signal processing wavelets of different main frequencies, and performing iterative time-frequency analysis respectively;

[0012] If the main frequency of the signal processing wavelet meets the iteration accuracy within the preset number of iterations, the time-frequency analysis result is output;

[0013] If the main frequency of the signal processing wavelet does not meet the iteration accuracy within the preset number of iterations, the main frequency of the signal processing wavelet is reselected and iterative time-frequency analysis is performed until all the signal processing wavelets of the main frequency are input;

[0014] Step S4, selecting the main frequency of the signal processing wavelet with the least number of iterations from the output time-frequency analysis results, so as to output the optimal solution spectrum of the time-frequency analysis results;

[0015] Step S5: Identify the optimal solution spectrum of the time-frequency analysis result to obtain the tunnel nondestructive testing result.

[0016] In one embodiment, the signal processing wavelet is a Ricker wavelet matrix, and the main frequency of the Ricker wavelet matrix is ​​selected within the range of 0 to 400 Hz;

[0017] In one embodiment, in step S3, a Ricker wavelet matrix with a main frequency of 200 Hz is selected as the initial main frequency of the signal processing wavelet, and then the main frequency of the Ricker wavelet matrix is ​​gradually adjusted toward both ends;

[0018] In one embodiment, the main frequency interval of the input Ricker wavelet matrix is ​​1 Hz.

[0019] In one embodiment, in step S2, an L1 norm constraint is introduced to reduce the influence of multiple solutions.

[0020] In one embodiment, in step S2, introducing the iterative time-frequency analysis method includes:

[0021] Establish the linear inversion equation:

[0022] Dm=d+n;

[0023] Wherein, D is an orthogonal function, and its value range is within the range of an m×n matrix, and the value range of m is within the range of an n×1 matrix, n is the random noise in the measured signal, and m is the result of time-frequency analysis;

[0024] When D is irreversible, the time-frequency decomposition result m is expressed as:

[0025] m=(D H D) -1 D H d;

[0026] Among them, D H is the conjugate transpose or adjoint matrix of D, DH D is a reversible square matrix;

[0027] The time-frequency decomposition result m is equivalent to the least squares problem, that is, the cost function J LS Minimum, J LS The expression is as follows:

[0028]

[0029] L1 norm regularization is introduced to reduce the influence of noise in the solution. The regularization equation is as follows:

[0030]

[0031] Among them, ||m|| 1 is the sum of the absolute values ​​of all m, where m represents L 1 Constraints on parameters.

[0032] In one embodiment, L 1 The parameter solution is as follows:

[0033]

[0034] Where w i The frequency is f i The wavelet matrix, r i is the reflection coefficient, s is the seismic signal, nt is the number of time samples, and N is the number of frequency samples;

[0035] Replace the matrix D in the above formula with the operator A, and then the above formula can be written as:

[0036] Ax = d;

[0037] Among them, A is a matrix with fewer row elements than columns; d is the measured signal; x is the time-frequency analysis result to be calculated, that is, m;

[0038] Solve Ax = d, the expression is:

[0039]

[0040] If there is noise, the above equation becomes:

[0041] σ is the noise;

[0042] When σ=0, it can be rewritten as:

[0043] Where T starts iteration from 0;

[0044] Rewrite as φ(T)=||r T || 2 , where the residual rT =e-Ax T ;

[0045] Among them, x T for The optimal solution of φ(τ)=||r τ || 2 The solution is that the iterative method of r value is as follows:

[0046] r k+1 =r k +Δr k ;

[0047]

[0048] In the formula, φ′(T k )=-λ T =-||A T y T ||;

[0049] in,

[0050] In one embodiment, in step S3, the iteration accuracy is:

[0051]

[0052] in, and Represent the estimated values ​​of r, y and λ respectively;

[0053] In one embodiment, in step S3, the preset number of iterations is 1000 times;

[0054] The present invention also provides a storage medium storing a computer program, which, when executed by a processor, executes the above-mentioned tunnel nondestructive detection method based on seismic wave data.

[0055] The beneficial effects of the present invention are:

[0056] On the basis of linear time-frequency analysis, L1 norm constraint is introduced, and the time degree of freedom analysis method is realized from the perspective of inversion to reduce the influence of multiple solutions and improve the resolution and focusing of nondestructive testing. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] The present invention will be described in more detail below based on embodiments and with reference to the accompanying drawings, wherein:

[0058] Figure 1 Shows a flow chart of the detection method of the present invention;

[0059] Figure 2The schematic diagram of the inversion spectrum decomposition of the present invention is shown;

[0060] In the drawings, like reference numerals are used for like parts. The drawings are not necessarily to scale. DETAILED DESCRIPTION

[0061] The present invention will be further described below in conjunction with the accompanying drawings.

[0062] The present invention provides a tunnel nondestructive testing method based on seismic wave data, such as Figure 1 As shown, the following steps are included:

[0063] Step S1, receiving a measured signal from the field, and synchronously inputting a signal processing wavelet, the signal processing wavelet is a Ricker wavelet matrix, and the main frequency of the Ricker wavelet matrix is ​​selected within the range of 0 to 400 Hz.

[0064] Step S2, introducing an iterative time-frequency analysis method to perform iterative time-frequency analysis on the measured signal and the input signal processing wavelet, introducing an L1 norm constraint to reduce the influence of multiple solutions, including:

[0065] Establish the linear inversion equation:

[0066] Dm=d+n;

[0067] Wherein, D is an orthogonal function, and its value range is within the range of an m×n matrix, and the value range of m is within the range of an n×1 matrix, n is the random noise in the measured signal, and m is the result of time-frequency analysis;

[0068] When D is reversible, the time-frequency decomposition result m is expressed as:

[0069] m=D -1 d;

[0070] When D is irreversible, the time-frequency decomposition result m is expressed as:

[0071] m=(D H D) -1 D H d;

[0072] Among them, D H is the conjugate transpose or adjoint matrix of D, D H D is a reversible square matrix;

[0073] m=(D H D) -1 D H d is equivalent to the least squares problem, that is, the cost function J LS Minimum, J LS The expression is as follows:

[0074]

[0075] Since the problem to be solved in the actual detection process is often that the number of equations is not equal to the number of unknowns, a regularization method is introduced to reduce the influence of noise in the solution. The regularization equation is:

[0076]

[0077] Among them, ||m|| 1 is the sum of the absolute values ​​of all m, where m represents L 1 The parameter constraint term; λ is the regularization parameter, which plays a role in adjusting the regularization constraint term and the data error L 1 If the noise is known, then the noise satisfies χ 2 Distribution, if the noise is unknown and the noise influence is large, λ uses a large value to reduce the noise influence; if the noise is unknown and the noise influence is small, λ uses a small value to reduce the error between the measured signal and the orthogonal signal;

[0078] Among them, noise is mainly divided into system noise, environmental noise and random formation noise. Generally, known noise is system noise and environmental noise, including inherent noise of equipment, noise from water flow, animals, construction activities, etc. This kind of noise is generally regular and belongs to known noise, while unknown noise is other noise sources besides the foreseeable noise sources, such as formation noise;

[0079] The impact of noise is determined by observing the amplitude spectrum of the measured signal. When the amplitude values ​​of different frequency components differ by more than 2 times, and the waveforms of each frequency in the amplitude spectrum show irregular fluctuations, it is judged that the noise has a large impact.

[0080] When the amplitude values ​​of different frequency components do not differ by more than 2 times or the waveforms of each frequency in the amplitude spectrum show regular fluctuations, it is judged that the noise influence is small;

[0081] The value range of λ is a value greater than 0. As long as it is greater than 0, it can be 3-5 when the sound impact is large, and 1 when the noise impact is small.

[0082] Among them, for L 1 Parameter regularization solution, the solution method is as follows:

[0083] By folding the wavelet function of each frequency and the reflection coefficient, Dm=d+n is written as:

[0084]

[0085] Where w i The frequency is f i The wavelet matrix, r i is the reflection coefficient, s is the seismic signal, nt is the number of time samples, and N is the number of frequency samples;

[0086] Replace the matrix D in the above formula with the operator A, and then the above formula can be written as:

[0087] Ax = d;

[0088] Wherein, A is a matrix whose row number is less than the column number; d is the measured signal; x is the time-frequency analysis result to be calculated, i.e., m. In this embodiment, Ricker wave is used as the continuous wavelet transform operator A to reduce the computational memory of large-scale problems;

[0089] Solve Ax = d, the expression is:

[0090]

[0091] If there is noise, the above equation becomes:

[0092] σ is the noise;

[0093] When σ=0, and The same, rewritten as:

[0094] Where T starts iteration from 0;

[0095] Rewrite as φ(T)=||r T || 2 , where the residual r T =d-Ax T ;

[0096] Among them, x T for The optimal solution of φ(T)=||r T || 2 The solution is that the iterative method of r value is as follows:

[0097] r k+1 =r k +Δr k ;

[0098]

[0099] In the formula, φ′(T k )=-λ T =-||A T y T ||;

[0100] in,

[0101] Step S3, inputting signal processing wavelets of different main frequencies, and performing iterative time-frequency analysis respectively;

[0102] If the main frequency of the signal processing wavelet meets the iteration accuracy within the preset number of iterations, the time-frequency analysis result is output;

[0103] If the main frequency of the signal processing wavelet does not meet the iteration accuracy within the preset number of iterations, the main frequency of the signal processing wavelet is reselected and iterative time-frequency analysis is performed;

[0104] Among them, the Ricker wavelet matrix with a main frequency of 200Hz is selected as the initial main frequency of the signal processing wavelet, and then the main frequency of the Ricker wavelet matrix is ​​gradually adjusted to both ends, with a main frequency interval of 1Hz, until all the signal processing wavelets of the main frequency are input;

[0105] The iteration accuracy is:

[0106]

[0107] in, and Represent the estimated values ​​of r, y and λ respectively;

[0108] The preset number of iterations is 1000;

[0109] For example, the Ricker wavelet matrix with a main frequency of 200Hz is first selected as the initial main frequency of the signal processing wavelet for iterative time-frequency analysis. If the iteration accuracy is met within 1000 iterations, the time-frequency analysis result is output, and then the 199Hz and 201Hz Ricker wavelet matrices are selected as the signal processing wavelets for iterative time-frequency analysis respectively, until all the main frequencies of the Ricker wavelet matrices in the range of 0 to 400Hz are analyzed;

[0110] Step S4, selecting the main frequency of the signal processing wavelet with the least number of iterations from the output time-frequency analysis results, so as to output the optimal solution spectrum of the time-frequency analysis results;

[0111] Step S5, identifying the optimal solution spectrum of the time-frequency analysis result to obtain the tunnel nondestructive testing result;

[0112] It should be noted that this method implements the time degree of freedom analysis method from the perspective of inversion on the basis of linear time-frequency analysis, and introduces L1 norm constraints to reduce the influence of multiple solutions and improve the resolution and focusing of nondestructive testing.

[0113] In the description of the present invention, it is necessary to understand that the terms "upper", "lower", "bottom", "top", "front", "back", "inside", "outside", "left", "right", etc. indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention.

[0114] Although the present invention is described herein with reference to specific embodiments, it should be understood that these embodiments are merely examples of the principles and applications of the present invention. It should therefore be understood that many modifications may be made to the exemplary embodiments and that other arrangements may be devised without departing from the spirit and scope of the present invention as defined by the appended claims. It should be understood that the various dependent claims and features described herein may be combined in a manner different from that described in the original claims. It should also be understood that features described in conjunction with individual embodiments may be used in other described embodiments.

Claims

1. A tunnel nondestructive testing method based on seismic wave data, characterized in that: The steps include: Step S1, receiving a measured signal from the field, and synchronously inputting a signal processing wavelet; Step S2, introducing an iterative time-frequency analysis method to perform iterative time-frequency analysis on the measured signal and the input signal processing wavelet; Step S3, inputting signal processing wavelets of different main frequencies, and performing iterative time-frequency analysis respectively; If the main frequency of the signal processing wavelet meets the iteration accuracy within the preset number of iterations, the time-frequency analysis result is output; If the main frequency of the signal processing wavelet does not meet the iteration accuracy within the preset number of iterations, the main frequency of the signal processing wavelet is reselected and iterative time-frequency analysis is performed until all the signal processing wavelets of the main frequency are input; Step S4, selecting the main frequency of the signal processing wavelet with the least number of iterations from the output time-frequency analysis results, so as to output the optimal solution spectrum of the time-frequency analysis results; Step S5: Identify the optimal solution spectrum of the time-frequency analysis result to obtain the tunnel nondestructive testing result.

2. The tunnel nondestructive testing method based on seismic wave data according to claim 1 is characterized in that: The signal processing wavelet is a Ricker wavelet matrix, and the main frequency of the Ricker wavelet matrix is ​​selected within the range of 0 to 400 Hz.

3. The tunnel nondestructive testing method based on seismic wave data according to claim 2 is characterized in that: In step S3, a Ricker wavelet matrix with a main frequency of 200 Hz is selected as the initial main frequency of the signal processing wavelet, and then the main frequency of the Ricker wavelet matrix is ​​gradually adjusted toward both ends.

4. The tunnel nondestructive testing method based on seismic wave data according to claim 3 is characterized in that: The main frequency interval of the input Ricker wavelet matrix is ​​1 Hz.

5. The tunnel nondestructive testing method based on seismic wave data according to claim 1 is characterized in that: In step S2, an L1 norm constraint is introduced to reduce the influence of multiple solutions.

6. The tunnel nondestructive testing method based on seismic wave data according to claim 5 is characterized in that: In step S2, introducing the iterative time-frequency analysis method includes: Establish the linear inversion equation: Dm=d+n; Wherein, D is an orthogonal function, and its value range is within the range of an m×n matrix, and the value range of m is within the range of an n×1 matrix, n is the random noise in the measured signal, and m is the result of time-frequency analysis; When D is irreversible, the time-frequency decomposition result m is expressed as: m=(D H D) -1 D H d; Among them, D H is the conjugate transpose or adjoint matrix of D, D H D is a reversible square matrix; The time-frequency decomposition result m is equivalent to the least squares problem, that is, the cost function J LS Minimum, J LS The expression is as follows: L1 norm regularization is introduced to reduce the influence of noise in the solution. The regularization equation is as follows: Among them, ||m||1 is the sum of the absolute values ​​of all m, and m represents the constraint item of L1 parameters.

7. The tunnel nondestructive testing method based on seismic wave data according to claim 6 is characterized in that: Solve for the L1 parameter as follows: Where w i The frequency is f i The wavelet matrix, r i is the reflection coefficient, s is the seismic signal, nt is the number of time samples, and N is the number of frequency samples; Replace the matrix D in the above formula with the operator A, and then the above formula can be written as: Ax = d; Among them, A is a matrix with fewer row elements than columns; d is the measured signal; x is the time-frequency analysis result to be calculated, that is, m; Solve Ax = d, the expression is: s.t.Ax=d; If there is noise, the above equation becomes: st||Ax-d||2≤σ, σ is noise; When σ=0, it can be rewritten as: st||x|| l ≤T, where T starts iterating from 0; Rewrite as φ(T)=||r T ||2, where the residual r T =d-Ax T ; Among them, x T for The optimal solution of st||x||1≤T, and φ(T)=||r T ||2 Solved, the way to iterate the r value is as follows: r k+1 =r k +Δr k ; where φ′(T k ) = -λ T = -||A T y T ||; in, 8. The tunnel nondestructive testing method based on seismic wave data according to claim 7 is characterized in that: In step S3, the iteration accuracy is: in, and denote the estimated values ​​of r, y and λ respectively.

9. The tunnel nondestructive testing method based on seismic wave data according to claim 1, characterized in that: In step S3, the preset number of iterations is 1000.

10. A storage medium, characterized in that: It stores a computer program, which, when run by a processor, executes the tunnel non-destructive testing method based on seismic wave data as described in any one of claims 1 to 9.

Images