An acoustic parameter adaptive inversion method based on seabed type acoustic classification

Abstract

The application discloses a kind of based on seabed type acoustic classification's acoustic parameter self-adapting inversion method, belong to acoustic parameter processing technical field, including the following steps: S1, obtain seabed reflection loss data;S2, judge whether seabed type is single-layer seabed, if yes then enter S3, otherwise enter S4;S3, according to seabed reflection loss data, classify single-layer seabed, and enter S5;S4, according to seabed reflection loss data, classify double-layer seabed, and enter S5;S5, according to seabed classification result and seabed reflection loss data, carry out seabed acoustic parameter inversion.The application is combined with the seabed type and angle-frequency domain seabed reflection characteristics obtained, self-adaptively constructs the cost function and inversion method with stronger correlation, so that the inversion process is optimized from full parameter space to focus on specific characteristics, and the inversion result is decoupled, which is more conducive to improving the precision, efficiency and robustness of seabed acoustic parameter inversion.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of acoustic parameter processing, and particularly relates to an acoustic parameter adaptive inversion method based on seabed type acoustic classification. BACKGROUND

[0002] Seabed acoustic parameters generally refer to the sound velocity, density, absorption coefficient and layer thickness of the seabed. Since the seabed is one of the important boundaries of the ocean acoustic waveguide, accurate seabed acoustic parameters are necessary boundary conditions for accurately solving the wave equation. Therefore, obtaining seabed acoustic parameters is a prerequisite for ensuring the high performance of a sonar system, and the acquisition of seabed acoustic parameters is a hot issue in the application research of underwater sonar.

[0003] There are three main methods for obtaining seabed parameters at present: a) through seabed sampling, measurement and acquisition in the laboratory; b) through in-situ measurement; and c) using received underwater acoustic signals, adopting a signal processing method, and inversely deducing seabed acoustic parameters by combining a physical model, i.e. an acoustic inversion method. The seabed parameters obtained by sampling and in-situ measurement are relatively accurate, but the cost is high and the efficiency is low, especially for deep sea environments. At the same time, the obtained seabed parameters are discrete sampling values, and it is difficult to accurately represent large-scale seabed parameters through a small amount of sampling data. More importantly, when the wavelength of the sound wave is much larger than the sampling scale, physical sampling may not be feasible for the characterization of low-frequency (1 kHz) seabed parameter characteristics. At low frequencies, acoustic inversion is the most effective means of obtaining seabed geophysical characteristics. The underwater acoustic signal is delayed, attenuated and distorted due to the seabed boundary, and also carries a large amount of information about the seabed. The equivalent seabed parameters obtained by the acoustic inversion method can reflect the average seabed acoustic characteristics on a large scale, and it is a convenient, economical and efficient approach.

[0004] The acoustic inversion method takes matching field inversion as the main framework, mainly uses the mapping relationship between the acoustic characteristic physical quantity and the seabed acoustic parameter to establish a cost function, and obtains the optimal solution of the cost function in the parameter range through a traversal method or a multi-dimensional optimization algorithm on the assumed seabed model and parameter boundary. The optimal solution is the equivalent seabed acoustic parameter to be inverted.

[0005] In the acoustic inversion method, the most typical acoustic characteristic physical quantity is sound pressure. However, since the sound pressure is a global acoustic characteristic physical quantity, a nonlinear mapping relationship is established between the wave equation and the seabed acoustic parameter, which also leads to the fact that the process of inverting the seabed acoustic parameter through the sound pressure is easily affected by the mismatch of the marine environmental parameters and the transmission parameters, and the inversion result has a multiple solution problem. It is necessary to analyze the uniqueness and uncertainty of the inversion result through posterior probability density analysis of multiple and parallel optimization results.

[0006] In the acoustic inversion method, the acoustic feature physical quantity most directly related to the seabed acoustic parameters is the seabed reflection loss, which has the characteristics of two dimensions of angle-frequency. In a typical seabed acoustic type, the seabed reflection loss is established by an analytical theory expression and the seabed acoustic parameters to establish a simple and unique mapping relationship, so that the inversion process is more efficient, and the multiple solution problem of the inversion result is effectively solved.

[0007] The main problem faced by the current method is that:

[0008] 1. The current acoustic inversion method is mainly based on the assumed seabed model. Before starting the inversion calculation, the existing inversion method must pre-select a physical model describing the seabed structure and the parameterization mode corresponding to the model. Since the real seabed stratification and acoustic parameter distribution are extremely complex, there is generally not enough prior information to establish an accurate seabed model. It will cost a huge economic cost to obtain the seabed prior information through means such as direct sampling measurement.

[0009] 2. The existing inversion method (such as matched field inversion) attempts to use a single, fixed model and inversion strategy to process all possible seabed types. This method ignores the essential differences in the reflection characteristics of sound waves and key inversion parameters of different seabed acoustic structures.

[0010] 3. The influence of the seabed acoustic parameters on the reflection coefficient is coupled and nonlinear. Changing different parameter combinations can produce very similar reflection loss curves (especially within a limited grazing angle or frequency range), resulting in serious non-uniqueness of the inversion problem.

[0011] These problems will lead to the following problems: the mismatch of the seabed model becomes the root cause of the systematic error of the inversion, the inversion method is inefficient and cannot adapt to the diversity of the seabed, the inversion result is non-unique, the uncertainty is high, and the physical meaning is unclear. SUMMARY

[0012] In order to solve the above problems, the present application proposes an acoustic parameter adaptive inversion method based on seabed type acoustic classification.

[0013] The technical scheme of the present application is: an acoustic parameter adaptive inversion method based on seabed type acoustic classification comprises the following steps:

[0014] S1, obtaining seabed reflection loss data;

[0015] S2, judging whether the seabed type is a single-layer seabed, if yes, entering S3, otherwise entering S4;

[0016] S3, classifying the single-layer seabed according to the seabed reflection loss data, and entering S5;

[0017] S4, classifying the double-layered seabed according to the seabed reflection loss data, and entering S5;

[0018] S5, performing seabed acoustic parameter inversion according to the seabed classification result and the seabed reflection loss data.

[0019] Further, in S2, it is judged whether there is an interference fringe in the grazing angle of 50° to 90° in the seabed reflection loss data, if yes, the seabed type is a double-layered seabed, otherwise, the seabed type is a single-layered seabed.

[0020] Further, S3 includes the following sub-steps:

[0021] S31, judging whether there is a critical angle in the seabed reflection loss data, if yes, the seabed type is a high-velocity single-layered seabed, otherwise, entering S32;

[0022] S32, judging whether there is a full transmission angle in the seabed reflection loss data, if yes, the seabed type is a low-velocity single-layered seabed, otherwise, ending the inversion.

[0023] Further, S4 includes the following sub-steps:

[0024] S41, if there is no subcritical angle in the seabed reflection loss data and there is a periodic abnormal increase with frequency in the grazing angle of 0° to 10°, the seabed type is a low-velocity double-layered homogeneous seabed; if there is a subcritical angle in the seabed reflection loss data and there is no periodic abnormal increase with frequency, entering S42;

[0025] The abnormal increase is specifically that the reflection loss in the adjacent 1° grazing angle range is increased by 10 dB;

[0026] S42, if there is a full transmission angle in the seabed reflection loss data in the grazing angle of 0° to the subcritical angle and changes with frequency, the seabed type is a low-velocity double-layered inhomogeneous seabed; if there is no full transmission angle in the seabed reflection loss data in the grazing angle of 0° to the subcritical angle, entering S43;

[0027] S43, if there is a subcritical angle in the seabed reflection loss data and changes with frequency, the seabed type is a high-velocity double-layered inhomogeneous seabed; if there is no subcritical angle in the seabed reflection loss data and changes with frequency, the seabed type is a high-velocity double-layered homogeneous seabed.

[0028] Further, in S5, the expression for performing the high-velocity single-layered seabed inversion is:

[0029] ;

[0030] wherein E1(α b ) is a first cost function, N is the number of grazing angles, M is the number of frequencies, is the seabed reflection loss data, is the theoretical bottom loss of the high-velocity single-layer bottom, a b is the bottom absorption coefficient, is the grazing angle of the bottom loss data, f k is the sound frequency of the bottom loss data.

[0031] Further, in S5, the expression for the low-velocity single-layer bottom inversion is:

[0032] ;

[0033] ;

[0034] wherein, is the second cost function, is the third cost function, is the full-transmission angle of the bottom loss data, is the theoretical full-transmission angle of the low-velocity single-layer bottom, is the first parameter cluster to be inverted, N is the number of grazing angles, and M is the number of frequencies, is the bottom loss data, is the theoretical bottom loss of the low-velocity single-layer bottom, a b is the bottom absorption coefficient, is the grazing angle of the bottom loss data, f k is the sound frequency of the bottom loss data, is the inversion vector combination.

[0035] Further, in S5, the expression for the low-velocity double-layer homogeneous bottom inversion is:

[0036] ;

[0037] ;

[0038] wherein, E4(c s ) is the fourth cost function, E5(a s , a b ) is the fifth cost function, N is the number of grazing angles, and M is the number of frequencies, is the bottom loss data, is the theoretical bottom loss of the low-velocity double-layer homogeneous bottom, a b is the bottom absorption coefficient, is the interference fringe coordinate of the low-velocity double-layer homogeneous bottom loss data, is the interference fringe coordinate of the theoretical bottom loss of the low-velocity double-layer homogeneous bottom, f is the frequency, c s is the sediment layer sound velocity, a sabsorption coefficient of sediment layer, grazing angle of bottom loss data, f k sound frequency of bottom loss data, grazing angle.

[0039] Further, in S5, the expression for the low-velocity double-layer inhomogeneous bottom inversion is:

[0040] ;

[0041] ;

[0042] ;

[0043] ;

[0044] wherein, sixth generation cost function, seventh generation cost function, E8(h s ) is the eighth generation cost function, E9(α b ) is the ninth generation cost function, total transmission angle of low-velocity double-layer inhomogeneous bottom loss data, theoretical total transmission angle of low-velocity double-layer inhomogeneous bottom, second to-be-inverted parameter cluster, g c sound velocity gradient of sediment layer, p s density of sediment layer, Z ws surface acoustic impedance of seawater-sediment layer, downward modified wave function of seawater-sediment layer surface, grazing angle, ω is sound frequency, ω t_exp transmission frequency extracted from measured data, N is the number of grazing angles, M is the number of frequencies, interference fringe coordinate of low-velocity double-layer inhomogeneous bottom loss data, f exp sound frequency of bottom loss data, theoretical bottom loss interference fringe of low-velocity double-layer inhomogeneous bottom, h s thickness of sediment layer, α b absorption coefficient of basement, bottom loss data, theoretical bottom loss of low-velocity double-layer inhomogeneous bottom inversion, grazing angle of bottom loss data, f k sound frequency of bottom loss data.

[0045] Further, in S5, the expression for the high-velocity double-layer inhomogeneous bottom inversion is:

[0046] ;

[0047] wherein, E 10 (α s ,α sb ) is the tenth cost function, N is the number of grazing angles, M is the number of frequencies, is the bottom loss data, is the theoretical bottom loss of high-velocity double-layer inhomogeneous bottom, is the grazing angle of the bottom loss data, f k is the acoustic frequency of the bottom loss data, α b is the bottom absorption coefficient, α s is the sediment absorption coefficient.

[0048] Further, in S5, the expression for high-velocity double-layer homogeneous bottom inversion is:

[0049] ;

[0050] ;

[0051] wherein, E 11 (c s ,h s ) is the eleventh cost function, E 12 (α s ,c b ,α b ) is the twelfth cost function, α b is the base absorption coefficient, c s is the sediment velocity, is the interference fringe coordinate of high-velocity double-layer homogeneous bottom loss data, is the theoretical bottom loss interference fringe of high-velocity double-layer homogeneous bottom, is the grazing angle, f is the frequency, h s is the sediment thickness, N is the number of grazing angles, M is the number of frequencies, is the bottom loss data, is the theoretical bottom loss of high-velocity double-layer homogeneous bottom, is the grazing angle of the bottom loss data, f k is the acoustic frequency of the bottom loss data, α s is the sediment absorption coefficient, c b is the base velocity.

[0052] The beneficial effects of the present application are: compared with the matching field based acoustic inversion method which performs inversion by assuming a seabed model, the present application obtains the seabed type by the characteristics embodied in the angle-frequency domain seabed reflection loss, and provides more prior information for the inversion process; compared with the matching field based acoustic inversion method which uses the same cost function and inversion method for all seabed models, the present application adaptively constructs the cost function and inversion method with stronger correlation by combining the obtained seabed type and angle-frequency domain seabed reflection characteristics, so that the inversion process is decoupled from the full parameter space optimization to focus on specific characteristics, and the precision, efficiency and robustness of the seabed acoustic parameter inversion are more improved. BRIEF DESCRIPTION OF DRAWINGS

[0053] Figure 1 A flowchart of the acoustic parameter adaptive inversion method based on seabed type acoustic classification;

[0054] Figure 2 A seabed reflection loss graph in the angle-frequency domain with a SWAMI experimental time window of about 2m;

[0055] Figure 3 A distribution graph of the optimal value;

[0056] Figure 4 A sediment layer sound velocity distribution graph;

[0057] Figure 5 A sediment layer absorption coefficient distribution graph;

[0058] Figure 6 An inversion result graph;

[0059] Figure 7 A thickness inversion result graph under different power terms;

[0060] Figure 8 A basement absorption coefficient inversion result graph; DETAILED DESCRIPTION

[0061] The embodiments of the present application will be further described below with reference to the accompanying drawings.

[0062] As shown in the drawings, the present application provides an acoustic parameter adaptive inversion method based on seabed type acoustic classification, comprising the following steps: Figure 1 S1, obtaining seabed reflection loss data;

[0063] S2, judging whether the seabed type is a single layer seabed, if yes, entering S3, otherwise entering S4;

[0064] S3, classifying the single layer seabed according to the seabed reflection loss data, and entering S5;

[0065] S4, obtaining the seabed type according to the seabed reflection loss data, and entering S5;

[0066] S4, classifying the double-layered seabed according to the seabed reflection loss data, and entering S5;

[0067] S5, performing seabed acoustic parameter inversion according to the seabed classification result and the seabed reflection loss data.

[0068] The present application can effectively obtain the prior information of the seabed from the acoustic characteristic physical quantity, and can fully utilize the different seabed reflection characteristics closely related to the seabed acoustic parameters to reduce the uncertainty of the inversion, so as to improve the precision of the seabed acoustic parameter inversion.

[0069] The typical seabed type acoustic classification can be divided into the following three types according to the layered condition: single-layered seabed (including three parameters, seabed sound velocity c b , seabed density p b , and seabed absorption coefficient a b ); double-layered homogeneous seabed (including seven parameters, sediment layer sound velocity c s , sediment layer density p s , sediment layer absorption coefficient a s , sediment layer thickness h s , base sound velocity c b , base density p b , and base absorption coefficient a b ); and double-layered non-homogeneous seabed (in addition to the seven parameters of the double-layered homogeneous seabed, the sediment layer sound velocity gradient g c ). According to the sound velocity ratio of the seabed and seawater, the seabed can be divided into high-speed seabed and low-speed seabed, so the typical seabed type acoustic classification has six types.

[0070] The angle-frequency domain seabed reflection loss data requires that the grazing angle ranges from 0° to 90°, and the frequency f ranges from 50 Hz to 1 kHz or above; and the seawater sound velocity c w at the seabed boundary.

[0071] In the embodiment of the present application, in S2, it is judged whether the seabed reflection loss data has interference fringes within the grazing angle of 50° to 90°, if yes, the seabed type is double-layered seabed, otherwise the seabed type is single-layered seabed.

[0072] In the embodiment of the present application, S3 includes the following sub-steps:

[0073] S31, judging whether the seabed reflection loss data has a critical angle, if yes, the seabed type is high-speed single-layered seabed, otherwise entering S32;

[0074] S32, judging whether the bottom reflection loss data exist a full transmission angle, if yes, the bottom type is a low-velocity single-layer bottom, otherwise, ending the inversion.

[0075] grazing angle satisfies When the sound wave occurs total reflection, the RL is close to zero, and with the increase of the grazing angle, the RL increases accordingly, is the critical angle.

[0076] The RL abnormally increases at the grazing angle satisfies With the increase of the grazing angle, the RL first increases and then decreases, is the full transmission angle.

[0077] In the embodiment of the present application, S4 comprises the following sub-steps:

[0078] S41, if the bottom reflection loss data do not exist a subcritical angle and the grazing angle is 0-10° and the periodic abnormal increase with frequency appears, the bottom type is a low-velocity double-layer homogeneous bottom; if the bottom reflection loss data exist a subcritical angle and do not appear the periodic abnormal increase with frequency, S42 is entered;

[0079] The abnormal increase specifically refers to that the reflection loss in the adjacent 1° grazing angle range is 10dB larger;

[0080] S42, if the bottom reflection loss data exist a full transmission angle within the grazing angle 0- the subcritical angle and vary with frequency, the bottom type is a low-velocity double-layer inhomogeneous bottom; if the bottom reflection loss data do not exist a full transmission angle within the grazing angle 0- the subcritical angle, S43 is entered;

[0081] S43, if the bottom reflection loss data exist a subcritical angle varying with frequency, the bottom type is a high-velocity double-layer inhomogeneous bottom; if the bottom reflection loss data do not exist a subcritical angle varying with frequency, the bottom type is a high-velocity double-layer homogeneous bottom.

[0082] The subcritical angle is generated by the sound velocity ratio of the seawater-sediment interface and only exists in the double-layer bottom, the grazing angle satisfies When the sound wave occurs total reflection at the seawater-sediment interface, the RL is close to 0dB. In the embodiment of the present application, in S5, the expression for performing the high-velocity single-layer bottom inversion is:

[0083]

[0084] ;

[0085] Wherein, E1(alpha b ​Let N be the first cost function, N be the number of grazing angles, and M be the number of frequencies. For seabed reflection loss data, For the theoretical seabed reflection loss of a single layer of seabed with high sound velocity, α b The absorption coefficient of the seabed. f is the grazing angle for seabed reflection loss data. k The acoustic frequency represents the data lost due to seabed reflection.

[0086] High-speed single-layer seabed inversion: Input seawater sound velocity c w and critical angle The seabed sound speed c is inverted based on the following formula. b :

[0087] ;

[0088] Substitute the speed of sound at the bottom of the sea c b The inversion results were used to invert the seabed density ρ using Hamilton's empirical formula. b :

[0089] ;

[0090] Where ρ is density.

[0091] By constructing the first cost function, the seabed absorption coefficient α is inverted. b The absorption coefficient α at the seabed b When it is the optimal value, E1(α) b Take the minimum value. The theoretical seabed reflection loss for hypersonic single-layer seabed is calculated using the BOUNCE model.

[0092] In this embodiment of the invention, in S5, the expression for inverting a single layer of seabed at low sound speed is:

[0093] ;

[0094] ;

[0095] in, For the second cost function, For the third cost function, The full transmission angle for seabed reflection loss data. The theoretical full transmission angle for a low-sound-velocity single-layer seabed. Let N be the number of grazing angles and M be the number of frequencies, representing the first set of parameters to be inverted. For seabed reflection loss data, For the theoretical seabed reflection loss of a single layer of seabed at low sound speeds, α b The absorption coefficient of the seabed. The grazing angle of the seafloor reflection loss data is f k The sound wave frequency of the seafloor reflection loss data is f The inversion vector combination is obtained.

[0096] Low sound speed single-layer seafloor inversion: full transmission angle The seafloor sound speed c b , the seafloor density p b , and the seafloor absorption coefficient a b have the following relationship:

[0097] ;

[0098] Wherein, c w is the seawater sound speed, p w is the seawater density, and i is the imaginary part symbol.

[0099] The numerical range of the seafloor absorption coefficient a b is set to [0, 1] dB / lambda, and the optimization step is 0.01 dB / lambda. The c b and p b combination satisfying the full transmission angle relationship of the above formula under each step a b is calculated by using the second cost function.

[0100] is the theoretical full transmission angle. When a is the optimal value, the second cost function takes the maximum value. The inversion parameter vector is a b , and the inversion result is the inversion vector combination a . Substituting a into the third cost function, the seafloor acoustic parameters are inverted. When a b is the optimal value, the third cost function takes the minimum value.

[0101] In the embodiment of the application, in S5, the expression for performing low sound speed double-layer uniform seafloor inversion is:

[0102] ;

[0103] ;

[0104] Wherein, E4(c s ) is the fourth cost function, E5(a s , a b ) is the fifth cost function, N is the number of grazing angles, M is the number of frequencies, is the seafloor reflection loss data, is the theoretical seafloor reflection loss of the low sound speed double-layer uniform seafloor, a b is the seafloor absorption coefficient, Interference fringe coordinate of bottom loss data for low-velocity two-layer homogeneous seafloor, Interference fringe coordinate of theoretical bottom loss for low-velocity two-layer homogeneous seafloor, f is frequency, c s is sediment velocity, α s is sediment absorption coefficient, is grazing angle of bottom loss data, f k is acoustic frequency of bottom loss data, is grazing angle.

[0105] Low-velocity two-layer homogeneous seafloor inversion: extract the starting frequency point freq0 and frequency interval Δfreq which appear periodic abnormal increase with frequency f in bottom loss data, and the sediment velocity c s and thickness h s The relationship is as follows:

[0106] ;

[0107] ;

[0108] ;

[0109] ;

[0110] Where, μ0 is the grazing angle of sediment, c w is seawater velocity, ρ b is the density of the basement, ρ s is the density of the sediment, c b is the basement velocity. The interference fringe formation condition of bottom loss data in the grazing angle range of 50° to 90° is as follows:

[0111] ;

[0112] Where, n is the order of the fringe. Extract the interference fringe coordinate of bottom loss data in the grazing angle range of 50° to 90°, and the sediment velocity is inversed by the fourth cost function. The fourth cost function takes the minimum value when c s is the optimal value. After the sediment velocity c s is inversed, the sediment thickness h s can be inversed by the interference fringe formation condition and the frequency interval Δfreq. The basement velocity c b is inversed by substituting the inversed sediment velocity c s , the sediment thickness h s , and the starting frequency point freq0 into the above formula, and the sediment density ρs and the substrate density p b .

[0113] The fifth generation cost function is constructed to inverse the sediment layer absorption coefficient a s and the substrate absorption coefficient a b , the sediment layer absorption coefficient a s and the substrate absorption coefficient a b The fifth generation cost function takes the minimum value when the optimal value is obtained.

[0114] In the embodiment of the present application, in S5, the expression for the low sound speed double-layer non-uniform seabed inversion is:

[0115] ;

[0116] ;

[0117] ;

[0118] ;

[0119] wherein, is the sixth generation cost function, is the seventh generation cost function, E8(h s ) is the eighth wherein, is the sixth generation cost function, is the seventh generation cost function, E8(h s ) is the eighth generation cost function, E9(a b ) is the ninth generation cost function, is the full transmission angle of the low sound speed double-layer non-uniform seabed reflection loss data, is the theoretical full transmission angle of the low sound speed double-layer non-uniform seabed, is the second to-be-inverted parameter cluster, g c is the sediment layer sound speed gradient, p s is the sediment layer density, Z ws is the seawater-sediment layer surface acoustic impedance, is the seawater-sediment layer surface downward correction wave function, is the grazing angle, and ω is the sound wave frequency, ω t_exp is the transmission frequency extracted from the measured data, N is the number of grazing angles, and M is the number of frequencies, is the interference fringe coordinate of the low sound speed double-layer non-uniform seabed reflection loss data, f exp is the sound wave frequency of the seabed reflection loss data, is the theoretical seabed reflection loss interference fringe of the low sound speed double-layer non-uniform seabed, h s is the sediment layer thickness, a b is the substrate absorption coefficient, for bottom loss data, for the theoretical bottom loss of low-velocity double-layer inhomogeneous bottom inversion, for the grazing angle of bottom loss data, f k for the sound frequency of bottom loss data.

[0120] Low-velocity double-layer inhomogeneous bottom inversion: full transmission angle of low-velocity double-layer inhomogeneous bottom The expression is:

[0121] ;

[0122] ;

[0123] where c w is the sound speed of seawater, ρ w is the density of seawater, ρ s is the density of sediment layer, c s (0) is the surface sound speed of sediment layer, δ s (0) is the attenuation term of sediment layer.

[0124] The sixth generation cost function is constructed based on the full transmission angle expression, and the sixth generation cost function takes the maximum value when is the optimal value. is the inversion parameter vector, and the inversion result is the inversion vector combination of ρ s , and α s is the absorption coefficient of sediment layer.

[0125] When the angle satisfies the full transmission angle, the maximum reflection loss in the bandwidth occurs at the transmission frequency, and the relationship expression of the full transmission angle, the transmission frequency and the related bottom acoustic parameters is as follows:

[0126] ;

[0127] where Z ws is the impedance ratio of seawater-sediment layer interface, f + (0) is the correction function of wave function on the surface of sediment layer, is the grazing angle, is the full transmission angle, e is the exponential, i is the imaginary part symbol, is the sound speed of sediment layer at 0 m, is the 2 / 3 order second type Hankel function, is the 1 / 3 order first type Hankel function, is the surface phase of sediment layer.

[0128] The seventh generation cost function is constructed, and the two-dimensional inversion is to be inverted parameter cluster and sediment layer sound speed gradient g cWhen g c and When the optimal value is reached, the seventh cost function reaches its maximum value. The interference condition for large grazing angle interference of seabed reflection loss in low-sound, double-layered, non-uniform seabed is:

[0129] ;

[0130] in, This represents the real part of the phase difference within the deposition layer. denoted as the phase at depth h of the sedimentary layer, where an odd number of m corresponds to a quarter-wave layer and an even number of m corresponds to a half-wave layer. For the phase within the sedimentary layer, the integral approximation expression is related to the power term n of the sound velocity profile of the sedimentary layer.

[0131] The deposition layer thickness h is obtained through one-dimensional inversion using the eighth cost function. s When h s When the optimal value is reached, the eighth cost function reaches its minimum value. This is because the deposition layer thickness h... s The inversion is also related to the power term n of the acoustic velocity profile of the sedimentary layer. Therefore, different power terms n can be substituted to perform the inversion, in order to determine the optimal power term n and the corresponding sedimentary layer thickness h. s .

[0132] The sound velocity c of the sedimentary layer at maximum depth s (h s (c) represents the base sound velocity. b The basis density ρ is inverted using Hamilton's empirical formula. b Matrix absorption coefficient α b Inversion is performed using the ninth cost function. The seabed absorption coefficient α... b When the value is optimal, the ninth cost function takes the minimum value.

[0133] In this embodiment of the invention, in S5, the expression for hypersonic double-layer non-uniform seabed inversion is:

[0134] ;

[0135] Among them, E 10 (α s ,α sb () represents the tenth cost function, where N is the number of grazing angles and M is the number of frequencies. For seabed reflection loss data, The theoretical seabed reflection loss for high-sound velocity double-layered non-uniform seabed. f is the grazing angle for seabed reflection loss data. k For the acoustic frequency of the seabed reflection loss data, α b α is the absorption coefficient of the seabed. s is the absorption coefficient of the sediment layer.

[0136] High-speed two-layer inhomogeneous seabed inversion: Subcritical angle at each frequency point The relationship with frequency f and effective depth Δh is as follows:

[0137] ;

[0138] Using the subcritical angle at each frequency point , the effective depth Δh of each frequency is calculated according to the above formula. The seabed is discretely layered according to the effective depth, the first layer is Δh1, corresponding to the highest frequency point; the second layer is Δh2-Δh1, and Δh2 is the effective depth of the second highest frequency point. By analogy, the thickness of the nth layer is Δh n -Δh n-1 . The subcritical angle is used in each layer z to invert the sediment layer sound speed c s (z):

[0139] ;

[0140] The sediment layer density ρ s (z) is inverted by the Hamilton empirical formula. Due to the influence of the sound speed gradient, the sediment layer sound speed c s (z) and the density ρ s (z) are related to the depth. The sediment layer sound speed c s (z max ) at the maximum depth is the basement sound speed c b , and the sediment layer density ρ s (z max ) at the maximum depth is the basement density ρ b . The sediment layer sound speed gradient g c is inverted by taking the first-order derivative of the sediment layer sound speed c s (z) at the depth z:

[0141] ;

[0142] The sediment layer absorption coefficient α s is approximately constant with depth, and the tenth generation cost function is constructed to invert the sediment layer absorption coefficient α s and the basement absorption coefficient α b . When the sediment layer absorption coefficient α s and the basement absorption coefficient α b are optimal values, the tenth generation cost function takes the minimum value.

[0143] In the embodiment of the present application, in S5, the expression for high-speed two-layer homogeneous seabed inversion is:

[0144] ;

[0145] ;

[0146] wherein E 11 (c s ,h s ) is the eleventh cost function, E 12 (α s ,c b ,α b ) is the twelfth cost function, α b is the sediment absorption coefficient, c s is the sediment sound speed, is the interference fringe coordinate of the high sound speed double-layer homogeneous seabed reflection loss data, is the theoretical seabed reflection loss interference fringe of the high sound speed double-layer homogeneous seabed, is the grazing angle, f is the frequency, h s is the sediment thickness, N is the number of grazing angles, M is the number of frequencies, is the seabed reflection loss data, is the theoretical seabed reflection loss of the high sound speed double-layer homogeneous seabed, is the grazing angle of the seabed reflection loss data, f k is the sound wave frequency of the seabed reflection loss data, α s is the sediment absorption coefficient, c b is the base sound speed.

[0147] High sound speed double-layer homogeneous seabed inversion: the sediment sound speed and thickness are obtained by matching the interference fringe coordinate two-dimensional inversion, and the eleventh cost function is obtained. When c s and h s are the optimal values, the eleventh cost function takes the minimum value.

[0148] The twelfth cost function is constructed based on the matching of the seabed reflection loss mode, and the sediment absorption coefficient α s , the base sound speed c b and the absorption coefficient α b are inverted, and the sediment density and the base density are inverted by the Hamilton empirical formula. The sediment absorption coefficient α s and the base sound speed c b , the absorption coefficient α b are the optimal values, and the twelfth cost function takes the minimum value. Since the twelfth cost function has three parameters to be solved, it is necessary to perform optimization calculation by using genetic algorithm and other optimization algorithms, and the corresponding optimization algorithm is relatively mature, and can be realized by using the optimization toolbox of MATLAB.

[0149] In view of the fact that the prior art fails to fully utilize the seabed prior information and the correlation of the inversion method, the method disclosed by the present application is implemented as follows:

[0150] Data is from the SWAMI experiment located in muddy terrain in New England. Grazing angle. The data covers a range of 0°–85° and a frequency range of 0–4 kHz, representing seabed reflection loss data at a thickness of approximately 2 meters. Based on SWAMI hydrographic measurements, the sound velocity at the seabed is taken as 1470 m / s, and the density as 1.0 g / cm³. 3 .

[0151] like Figure 2 As shown, the angle-frequency domain seabed reflection loss is approximately 2m in the SWAMI experiment ("×" represents the transmission frequency, the dotted line represents the coordinates of the extracted interference fringes, and the dashed line represents the theoretical interference fringes calculated from the inversion results).

[0152] Step 1: Seabed reflection loss data at grazing angle Interference fringes exist within a range of 50° to 90°, such as Figure 2 The black dotted line indicates a double-layered seabed; proceed to step 3.1.

[0153] Step 3.1: The seabed reflection loss data has a subcritical angle and is at the grazing angle. No frequency-dependent operation within the 0°-10° range A periodic abnormal increase occurred; the subcritical angle was recorded. If the location is correct, proceed to step 3.2;

[0154] Step 3.2: Seabed reflection loss data in the grazing angle range of 0° to subcritical angle Within, there exists a total transmission angle. Furthermore, as the frequency f changes, the seabed type is a low-sound, double-layered, non-uniform seabed. The total transmission angle at the highest frequency was recorded through image pixel and coordinate transformation. The frequency ω corresponding to the maximum reflection loss at the full transmission angle t ≈1871Hz (transmission frequency), proceed to step 4.4;

[0155] Step 4.4: The parameter inversion range for the low-sound-velocity double-layer non-uniform seabed is given in Table 1.

[0156] Table 1

[0157]

[0158] Combined with the full transmission angle In the density parameter space, obtain the optimal value of the cost function for each density, along with the corresponding sound velocity and absorption coefficient, to form a parameter cluster. The result is as follows Figures 3-5 As shown, the red "+" marks the results of the second inversion step.

[0159] Assuming the sound speed profile is linear (n=+1), the parameter set obtained through inversion is used. and transmission frequency ω t_exp The inversion results of sound speed gradient and density normalization are shown in Figure 6 , and the corresponding sound speed and absorption coefficient results are marked with red + signs in Figures 3-5 .

[0160] The results are shown in Figures 3-6 , and the inversion results obtained in Figure 6 Although not concentrated, the numerical difference between the optimal value and the suboptimal value is obvious, and the optimal inversion result can be clearly distinguished. On the other hand, it can be observed from Figures 3-5 that the sediment layer sound speed, density and absorption coefficient obtained in the second step do not make the cost function value of maximum, because the inversion result at this time is globally optimal, not a single cost function optimal.

[0161] The sediment layer thickness is obtained using the inversion results, and the inversion results are given in Figure 7 . Considering the assumption of the sound speed profile shape, the inversion results for different n are given in Table 2, and the thickness inversion results for different n are given in Figure 7 . It can be known from Figure 7 that the result for the power term n = +1 is the best.

[0162] Table 2

[0163]

[0164] When the power term n = +1, the basement sound speed c b is (m / s), the basement density ρ b is 1.3434 g / cm 3 , which is inversed by Hamilton empirical formula.

[0165] Substitute the above inversion results into the BOUNCE model to calculate the theoretical sea bottom reflection loss in the parameter range of the basement absorption coefficient, and the basement absorption coefficient is inversed, and the results are shown in Figure 8 .

[0166] The inversion results of the acoustic parameters of the low sound speed double-layer non-uniform sea bottom are: the sediment layer sound speed c s is 1446.48 m / s, the sediment layer density ρ s is 1.645 g / cm 3 , the sediment layer absorption coefficient α s is 0.0111 (dB / m / kHz), the sediment layer thickness h s is 2.36 m, the sediment layer sound speed gradient g c 1.6 / s, the basement sound speed c bis 1450.25 m / s, base density p b is 1.3434 g / cm 3 , base absorption coefficient a b is 0.18 dB / λ (0.1241 dB / m / kHz).

[0167] Those skilled in the art will appreciate that the embodiments described herein are presented for purposes of illustration and should not be construed as limiting the scope of the present application. Various other modifications and alterations of the present application will become apparent to those skilled in the art from the foregoing disclosure without departing from the scope of the present application.

Claims

1. A method of acoustic parameter adaptive inversion based on seabed type acoustic classification, characterized in that, Includes the following steps: S1. Obtain seabed reflection loss data; S2. Determine if the seabed type is a single-layer seabed. If yes, proceed to S3; otherwise, proceed to S4. S3. Based on the seabed reflection loss data, classify the single-layer seabed into hypersonic single-layer seabed and hyposonic single-layer seabed, and proceed to S5. S4. Based on the seabed reflection loss data, classify the double-layer seabed into low-sound speed double-layer homogeneous seabed, low-sound speed double-layer non-homogeneous seabed, high-sound speed double-layer non-homogeneous seabed, and high-sound speed double-layer homogeneous seabed, and proceed to S5. S5. Based on the seabed classification results and seabed reflection loss data, perform seabed acoustic parameter inversion; In S5, the expression for inverting a single layer of seabed at low sound speed is: ； ； wherein, is a second cost function, is a third cost function, is a full-transmission angle of the seafloor reflection loss data, is a theoretical full-transmission angle of a low-velocity single-layer seafloor, is a first set of parameters to be inverted, N is a number of grazing angles, and M is a number of frequencies, is a seafloor reflection loss data, is a theoretical seafloor reflection loss of a low-velocity single-layer seafloor, a b is a seafloor absorption coefficient, is a grazing angle of the seafloor reflection loss data, f k is a sound frequency of the seafloor reflection loss data, is an inversion vector combination; , c b is a base velocity, p b is a base density.

2. The method of claim 1, wherein, In step S2, it is determined whether there are interference fringes in the seabed reflection loss data within a grazing angle of 50° to 90°. If so, the seabed type is a double-layer seabed; otherwise, the seabed type is a single-layer seabed.

3. The method of claim 1, wherein, S3 includes the following sub-steps: S31. Determine if there is a critical angle in the seabed reflection loss data. If so, the seabed type is hypersonic single-layer seabed; otherwise, proceed to S32. S32. Determine whether the seabed reflection loss data has a full transmission angle. If so, the seabed type is a low-sound-velocity single-layer seabed; otherwise, end the inversion.

4. The method of claim 1, wherein, S4 includes the following sub-steps: S41. If the seabed reflection loss data does not have a subcritical angle and the grazing angle increases periodically with frequency from 0° to 10°, then the seabed type is a low-sound-velocity, two-layer homogeneous seabed; if the seabed reflection loss data has a subcritical angle and does not increase periodically with frequency, then proceed to S42. The abnormal increase specifically refers to a 10dB increase in reflection loss within an adjacent 1° grazing angle range; S42. If the seabed reflection loss data has a full transmission angle within the range of 0° grazing angle to subcritical angle and varies with frequency, then the seabed type is a low-sound-velocity double-layer non-uniform seabed; if the seabed reflection loss data does not have a full transmission angle within the range of 0° grazing angle to subcritical angle, then proceed to S43. S43. If the seabed reflection loss data shows that the subcritical angle varies with frequency, then the seabed type is a hypersonic double-layer non-uniform seabed; if the seabed reflection loss data does not show that the subcritical angle varies with frequency, then the seabed type is a hypersonic double-layer uniform seabed.

5. The method of claim 1, wherein, In S5, the expression for hypersonic single-layer seabed inversion is: ； wherein E1(α b ) is a first cost function, N is the number of grazing angles, M is the number of frequencies, is the bottom loss data, is the theoretical bottom loss of a high sound speed single layer bottom, α b is the bottom absorption coefficient, is the grazing angle of the bottom loss data, f k is the sound frequency of the bottom loss data.

6. The method of claim 1, wherein, In S5, the expression for low-sound speed double-layer uniform seabed inversion is: ； ； E4(c s ) is the fourth cost function, E5(a s , a b ) is the fifth cost function, N is the number of grazing angles, M is the number of frequencies, is the seabed reflection loss data, is the theoretical seabed reflection loss of the low-velocity double-layer homogeneous seabed, a b is the seabed absorption coefficient, is the interference fringe coordinate of the low-velocity double-layer homogeneous seabed reflection loss data, is the interference fringe coordinate of the theoretical seabed reflection loss of the low-velocity double-layer homogeneous seabed, f is the frequency, c s is the sediment layer sound speed, a s is the sediment layer absorption coefficient, is the grazing angle of the seabed reflection loss data, f k is the sound wave frequency of the seabed reflection loss data, is the grazing angle.

7. The method of claim 1, wherein, In S5, the expression for low-sound speed double-layer non-uniform seabed inversion is: ； ； ； ； in, The sixth cost function, For the seventh cost function, E8(h) s E9(α) is the eighth cost function. b ) is the ninth cost function, The full transmission angle for low-sound velocity double-layer non-uniform seabed reflection loss data. The theoretical full transmission angle for a low-sound-velocity, double-layered, non-uniform seabed. For the second set of parameters to be inverted, g c ρ represents the sound velocity gradient of the sedimentary layer. s Z represents the density of the sedimentary layer. ws The acoustic impedance of the seawater-sediment layer surface. This represents the downlinked corrected wavefunction of the seawater-sediment layer surface. Let ω be the grazing angle, and ω be the sound wave frequency. t_exp The transmission frequencies are extracted from the measured data, where N is the number of grazing angles and M is the number of frequencies. The interferometric fringe coordinates for low-velocity double-layer non-uniform seabed reflection loss data, f exp The acoustic frequency for seabed reflection loss data. For the theoretical seabed reflection loss interference fringes of a low-sound-velocity, double-layered, non-uniform seabed, h s α represents the thickness of the deposition layer. b The substrate absorption coefficient, For seabed reflection loss data, The theoretical seabed reflection loss is used for inversion of a low-sound-velocity, double-layered, non-uniform seabed. f is the grazing angle for seabed reflection loss data. k The acoustic frequency for seabed reflection loss data; c s (0) represents the sound velocity at the surface of the sediment layer, α s is the absorption coefficient of the sediment layer.

8. The adaptive inversion method for acoustic parameters based on seabed type acoustic classification according to claim 1, characterized in that, In S5, the expression for hypersonic double-layer non-uniform seabed inversion is: ； Among them, E 10 (α s ,α b () represents the tenth cost function, where N is the number of grazing angles and M is the number of frequencies. For seabed reflection loss data, The theoretical seabed reflection loss for high-sound velocity double-layered non-uniform seabed. f is the grazing angle for seabed reflection loss data. k α represents the acoustic frequency of the seabed reflection loss data. b α is the absorption coefficient of the seabed. s is the absorption coefficient of the sediment layer.

9. The adaptive inversion method for acoustic parameters based on seabed type acoustic classification according to claim 1, characterized in that, In S5, the expression for hypersonic double-layer uniform seabed inversion is: ； ； Among them, E 11 (c s ,h s ) is the eleventh cost function, E 12 (α s ,c b ,α b ) is the twelfth cost function, α b c is the substrate absorption coefficient. s The sound velocity of the sedimentary layer, The coordinates of the interferometric fringes for hypersonic double-layer uniform seabed reflection loss data. The theoretical seabed reflection loss interference fringes for a hypersonic double-layered homogeneous seabed. where f is the grazing angle, f is the frequency, and h is the frequency. s Where N is the thickness of the deposition layer, N is the number of grazing angles, and M is the number of frequencies. For seabed reflection loss data, The theoretical seabed reflection loss for a hypersonic double-layered homogeneous seabed. f is the grazing angle for seabed reflection loss data. k α represents the acoustic frequency of the seabed reflection loss data. s c is the absorption coefficient of the sedimentary layer. b The base velocity is 1.

Images