Acoustic tomography arrival peak fuzzy matching extraction method based on acoustic ray intrinsic structure

Abstract

The invention discloses a sound tomography arrival peak fuzzy matching extraction method based on a sound ray intrinsic structure, and relates to the technical field of data analysis, and the method comprises the following steps: S1, carrying out the matched filtering processing of a sound signal, and obtaining a matched filtering normalization result; s2, determining a plurality of sound ray intrinsic structures; s3, performing ray cluster division, and determining the sound ray sequence number with the maximum sound ray intrinsic structure amplitude in each ray cluster and the theoretical maximum peak arrival time; and S4, completing the extraction of the arrival peak of the sound signal according to the matched filtering normalization result, the sound ray sequence number with the maximum sound ray intrinsic structure amplitude in each ray cluster and the theoretical maximum peak arrival time. According to the method, the background field sound ray intrinsic structure and the sound tomography measured data are subjected to fuzzy matching, each sound ray arrival peak can be stably extracted under the conditions that the multipath effect is relatively strong and the number of sound rays is relatively large, and the sound ray arrival peak with relatively low signal-to-noise ratio is effectively extracted.

Description

Technical Field

[0001] The invention relates to the technical field of data analysis, and in particular to an acoustic tomography arrival peak fuzzy matching extraction method based on sound line intrinsic structure. Background Art

[0002] Sound waves are the only known form of energy radiation that can propagate underwater over long distances, so the speed of sound is an important parameter for studying the ocean. The sound velocity profile refers to the distribution of sound velocity at depth. The acquisition of sound velocity profiles has great application value. The traditional method of acquiring sound velocity profiles is time-consuming and labor-intensive. The acoustic tomography technology can quickly obtain the temporal and spatial variation characteristics of the sound velocity profile. The acoustic tomography method inverts the characteristics of the ocean that the sound wave passes through by measuring the relevant parameters of the sound line propagation signal, such as the propagation time (most typical), and obtains an estimate of the ocean dynamics state and its changes in a large area of ​​sea. Acoustic tomography data is a series of time series sound signals with a total observation time, and because the sound source is intermittent, it is necessary to extract the arrival time of the sound line in segments from the entire observation time data. The peak arrival delay of the sound line is the key input for the acoustic tomography method to invert the seawater sound velocity field, so the arrival peak extraction is one of the key technologies in the acoustic tomography method.

[0003] Currently, commonly used methods for extracting the arrival peak of acoustic tomography time delay include extraction methods based on the maximum arrival peak and multi-error peak extraction methods based on segmented recognition. The inversion effect of the acoustic tomography method depends on the number of sound lines identified on the transmission path. When the number of arrival peaks is less, the number of corresponding sound lines is less, which reduces the inversion effect of the acoustic tomography. Most of the current methods are only applicable to rivers, coastal areas and shallow seas, and cannot effectively cope with the situation of deep-sea acoustic tomography. In particular, when the multi-path effect is strong, due to the sharp increase in the number of sound lines, the current arrival peak extraction method is difficult to distinguish the arrival peak delay corresponding to each sound line, and cannot effectively distinguish and identify the arrival sound lines, and cannot effectively identify the sound lines with low signal-to-noise ratio after multiple reflections. In order to solve the above problems, the present invention proposes an acoustic tomography arrival peak fuzzy matching extraction method based on the eigenstructure of the sound line. Summary of the invention

[0004] In order to solve the above problems, the present invention proposes an acoustic tomography arrival peak fuzzy matching extraction method based on the eigenstructure of sound rays.

[0005] The technical solution of the present invention is: a method for extracting fuzzy matching of arrival peaks of acoustic tomography based on the intrinsic structure of sound rays comprises the following steps:

[0006] S1, using an acoustic tomography instrument to receive an acoustic signal, and performing matched filtering processing on the acoustic signal to obtain a matched filtering normalization result;

[0007] S2, determining a number of sound line eigenstructures;

[0008] S3, dividing the ray clusters according to the eigenstructures of several sound rays, and determining the number of the sound ray with the largest amplitude of the eigenstructure of the sound ray in each ray cluster and the theoretical maximum peak arrival time;

[0009] S4. Extract the arrival peak of the acoustic signal according to the normalized result of matched filtering, the number of the sound line with the largest sound line intrinsic structure amplitude in each ray cluster, and the theoretical maximum peak arrival time.

[0010] Furthermore, S1 includes the following sub-steps:

[0011] S11, using an acoustic tomograph to receive an acoustic signal and calculate the signal amplitude of a matched filter;

[0012] S12, normalizing the signal amplitude of the matched filter to obtain a matched filter normalization result.

[0013] Furthermore, in S11, the signal amplitude A of the matched filter C The calculation formula for [t] is:

[0014]

[0015] In the formula, C i [t] represents the amplitude of the orthogonal component of the acoustic signal, C q [t] represents the amplitude of the in-phase component of the acoustic signal, and t represents time.

[0016] Furthermore, in S12, the matched filtering normalization result The expression is:

[0017]

[0018] In the formula, A C [t] represents the signal amplitude of the matched filter, max(·) represents the maximum value operation, min(·) represents the minimum value operation, and t represents time.

[0019] Furthermore, S2 includes the following sub-steps:

[0020] S21. Set several parameters, input the parameters into the BELLHOP acoustic toolbox, and obtain several sound line eigenstructures;

[0021] S22. Sort several sound line eigenstructures in ascending order according to theoretical arrival time.

[0022] Furthermore, S3 includes the following sub-steps:

[0023] S31, determining a time window according to a plurality of sound line eigenstructures arranged in ascending order according to theoretical arrival times;

[0024] S32, dividing the eigenstructures of a plurality of sound rays in the time domain according to time windows to obtain a plurality of ray clusters;

[0025] S33, accumulating the number of eigenstructures of each ray cluster as the sequence number of the eigenstructure of the sound ray in each ray cluster;

[0026] S34, taking the time span between two sound ray eigenstructures with the largest theoretical arrival time difference in each ray cluster as the time span of each ray cluster;

[0027] S35. According to the time span of each ray cluster, obtain the sound ray number of the sound ray with the largest intrinsic structure amplitude and the theoretical maximum peak arrival time in each ray cluster.

[0028] Further, S4 includes the following sub-steps:

[0029] S41, determining the actual maximum peak arrival time of the first ray cluster according to the sound ray number corresponding to the maximum value of the peak in the matched filtering normalization result, and calculating the maximum disturbance time window according to the theoretical maximum peak arrival time in the remaining ray clusters;

[0030] S42, based on the maximum disturbance window, determining the sound line number corresponding to the maximum value of the peak in the matched filtering normalization result;

[0031] S43, defining the sound line number corresponding to the maximum value of the peak value in the matched filtering normalization result as l s1 , the sound line number corresponding to the maximum value of the peak in the matched filtering normalization result is defined as l f1 , will l s1 and l f1 Sort the results of matched filtering normalization by their maximum peak values ​​from large to small;

[0032] S44. Define the number of the sound line with the largest amplitude greater than the sound line eigenstructure in the first ray cluster as R s1 , define the number of the sound line with the largest amplitude smaller than the sound line eigenstructure in the first ray cluster as R f1 , R s1 and R f1 Sort by the amplitude of the sound line eigenstructure from large to small;

[0033] S45, will l s1 and l f1 Substitute into the discretized time series and get R s1 and R f1 The actual arrival time corresponding to the sound ray completes the identification of the sound ray arrival peak in the first ray cluster;

[0034] S46, repeat S43-S45 until the identification of the arrival peaks of the sound rays in all ray clusters is completed.

[0035] Furthermore, in S41, the calculation formula of the maximum disturbance time window is:

[0036]

[0037] In the formula, represents the actual maximum peak arrival time of the first ray cluster, represents the theoretical maximum peak arrival time within the remaining ray clusters.

[0038] The beneficial effects of the present invention are:

[0039] (1) The present invention proposes a fuzzy matching extraction method for acoustic tomography arrival peaks based on the eigenstructure of sound rays. The eigenstructure of sound rays is used as a physical constraint to adaptively divide ray clusters and adaptively extract and identify the arrival peaks corresponding to each sound ray within the time span of the cluster.

[0040] (2) The present invention first calculates the eigenstructure of sound rays under the background field, divides the arriving sound rays into ray clusters in the time domain, and then performs fuzzy matching through the eigenstructure of sound rays and the related data of acoustic tomography matched filtering. Then, the arrival peak of the sound rays in each ray cluster is located through continuous iteration, and finally the arrival time of each eigenwave in the measured data at each observation time is given;

[0041] (3) The present invention performs fuzzy matching between the intrinsic structure of background sound lines and the measured data of acoustic tomography, and can stably extract the arrival peaks of each sound line even when the multipath effect is strong and the number of sound lines is large, and effectively extract the arrival peaks of sound lines with weak signal-to-noise ratio. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 It is a flow chart of the fuzzy matching extraction method of the acoustic tomography arrival peak based on the eigenstructure of the sound line;

[0043] Figure 2 This is the result of matched filtering of acoustic tomography data;

[0044] Figure 3 This is the background sound velocity profile of the experimental sea area;

[0045] Figure 4 It is the topography and bottom parameter map of the experimental sea area;

[0046] Figure 5 This is the sound line distribution diagram on the experimental propagation line;

[0047] Figure 6 This is the intrinsic structure diagram on the experimental propagation line;

[0048] Figure 7This is the maximum peak arrival extraction result diagram of ray cluster 1;

[0049] Figure 8 This is the result diagram of the peak arrival delay extraction of ray cluster 1;

[0050] Fig. 9 This is the peak arrival delay extraction result diagram of ray cluster 2;

[0051] Fig.10 This is the peak arrival delay extraction result diagram of ray cluster 3;

[0052] Fig.11 This is the result diagram of the peak arrival delay extraction of ray cluster 4;

[0053] Fig.12 This is the result diagram of the peak arrival delay extraction of ray cluster 5. DETAILED DESCRIPTION

[0054] The embodiments of the present invention will be further described below in conjunction with the accompanying drawings.

[0055] like Figure 1 As shown, the present invention provides an acoustic tomography arrival peak fuzzy matching extraction method based on the eigenstructure of sound rays, comprising the following steps:

[0056] S1, using an acoustic tomography instrument to receive an acoustic signal, and performing matched filtering processing on the acoustic signal to obtain a matched filtering normalization result;

[0057] S2, determining a number of sound line eigenstructures;

[0058] S3, dividing the ray clusters according to the eigenstructures of several sound rays, and determining the number of the sound ray with the largest amplitude of the eigenstructure of the sound ray in each ray cluster and the theoretical maximum peak arrival time;

[0059] S4. Extract the arrival peak of the acoustic signal according to the normalized result of matched filtering, the number of the sound line with the largest sound line intrinsic structure amplitude in each ray cluster, and the theoretical maximum peak arrival time.

[0060] In this embodiment of the present invention, S1 includes the following sub-steps:

[0061] S11, using an acoustic tomograph to receive an acoustic signal and calculate the signal amplitude of a matched filter;

[0062] S12, normalizing the signal amplitude of the matched filter to obtain a matched filter normalization result.

[0063] In the embodiment of the present invention, in S11, the signal amplitude A of the matched filter C The calculation formula for [t] is:

[0064]

[0065] In the formula, C i [t] represents the amplitude of the orthogonal component of the acoustic signal, C q [t] represents the amplitude of the in-phase component of the acoustic signal, and t represents time.

[0066] In the embodiment of the present invention, in S12, the matched filtering normalization result The expression is:

[0067]

[0068] In the formula, A C [t] represents the signal amplitude of the matched filter, max(·) represents the maximum value operation, min(·) represents the minimum value operation, and t represents time.

[0069] In S1, the acoustic signal R(t) received by the acoustic tomography instrument is processed by matched filtering, and the result of the matched filtering is used as one of the inputs for extracting the arrival peak of the acoustic tomography. The process of matched filtering is as follows:

[0070] Assume that the sound source signal S(t a ) is the carrier modulated by the P sequence, and its waveform is as follows:

[0071] S(t a )=P(t a )sin(ωt a );

[0072] Among them, t a is time, P(t a ) is the waveform pattern of the P sequence, and ω is the angular frequency of the carrier.

[0073] Then the sound signal R(t a ) is expressed as:

[0074] R(t a )=αP(t a -t 0 )sin{ω(t a -t 0 )}+n(t a );

[0075] Where α is the sound source signal S(t a ) attenuation rate of underwater propagation, t 0 is the sound source signal S(t a ) is the propagation time from the sound source to the acoustic tomography receiver, n(t a ) is the environmental noise.

[0076] The acoustic signal R(t a ) is discretized using the sampling frequency:

[0077] t[l]=t a (l / f s ), l∈Z;

[0078] R[t]=αP[tt 0 ]sin{ω[tt 0 ]}+n[t];

[0079] In the formula, f s is the sampling frequency, t[l] is the discretized time series, l is the sequence number, and R[t] is the discretized sound signal.

[0080] For R[t] = αP[tt 0 ]sin{ω[tt 0 ]}+n[t] is demodulated and separated into orthogonal signals R i [t] and the same phase signal R q [t] Two components:

[0081]

[0082] By combining the above two formulas with R(t a )=αP(t a -t 0 )sin{ω(t a -t 0 )}+n(t a ) for a period of cross-correlation, the above two formulas become:

[0083]

[0084] In the formula, C i [t] and C q [t] are the quadrature and in-phase component amplitudes of the acoustic signal R(t), τ k represents the time window of translation, the subscript k represents the number of translations, and t 0 It is the propagation time taken by the sound source signal S(t) from the sound source to the acoustic tomograph.

[0085] Matched filtered signal amplitude A C [t] is:

[0086]

[0087] A C [t] Perform amplitude normalization processing to obtain the matched filtering normalization result.

[0088] In this embodiment of the present invention, S2 includes the following sub-steps:

[0089] S21. Set several parameters, input the parameters into the BELLHOP acoustic toolbox, and obtain several sound line eigenstructures;

[0090] S22. Sort several sound line eigenstructures in ascending order according to theoretical arrival time.

[0091] In S2, the sound velocity profile, topography, bottom and other environmental parameters on the propagation survey line are set (the set environmental parameters are called background field), the sound source frequency, sound source depth, receiving depth and propagation distance and other parameters are set, and input into the BELLHOP acoustic toolbox to calculate the sound line eigenstructure under the background field. The sound line eigenstructure includes the amplitude of the sound line, the theoretical arrival time, the outgoing grazing angle, the receiving grazing angle, the number of sea surface reflections and the number of seabed reflections. The BELLHOP calculation results show that there are M sound line eigenstructures, M is a positive integer, and the sound line eigenstructures are sorted from small to large according to the theoretical arrival time.

[0092] In this embodiment of the present invention, S3 includes the following sub-steps:

[0093] S31, determining a time window according to a plurality of sound line eigenstructures arranged in ascending order according to theoretical arrival times;

[0094] S32, dividing the eigenstructures of a plurality of sound rays in the time domain according to time windows to obtain a plurality of ray clusters;

[0095] S33, accumulating the number of eigenstructures of each ray cluster as the sequence number of the eigenstructure of the sound ray in each ray cluster;

[0096] S34, taking the time span between two sound ray eigenstructures with the largest theoretical arrival time difference in each ray cluster as the time span of each ray cluster;

[0097] S35. According to the time span of each ray cluster, obtain the sound ray number of the sound ray with the largest intrinsic structure amplitude and the theoretical maximum peak arrival time in each ray cluster.

[0098] In S31, define T w The time window used for ray cluster division is determined by the maximum theoretical arrival time difference of M sound rays, and is generally in the range of 0.1 to 1 s.

[0099] In S32, the line intrinsic structure is T in the time domain. w Divide. w The sound eigenstructure in the inner part is defined as a ray cluster. Assuming that there are N T w There are sound ray eigenstructures in it, that is, N ray clusters are obtained. N ray clusters contain a total of M sound ray eigenstructures.

[0100] S33, based on the division results of the N ray clusters, the number of the sound ray eigenstructures in each ray cluster is accumulated to calculate the sound ray eigenstructure serial number R in each ray cluster. n ; Calculate the time span of the two sound eigenstructures with the largest theoretical arrival time difference in each ray cluster, and obtain the time span T of each ray cluster n , subscript n is the number of the ray cluster, n = 1, 2, ..., N, the eigenstructure number of the sound line in the ray cluster R n and time span T n As one of the inputs of the tomography arrival peak.

[0101] In S34, Matlab's max function can calculate the maximum value and the sequence number of the maximum value in the array. Matlab's max function can be used to locate the sequence number of the sound line with the largest amplitude of the sound line intrinsic structure in each ray cluster. The corresponding sound ray arrival time is the theoretical maximum peak arrival time in each ray cluster. The superscript n is the serial number of the ray cluster. and It is also used as one of the inputs of the tomography arrival peak.

[0102] In this embodiment of the present invention, S4 includes the following sub-steps:

[0103] S41, determining the actual maximum peak arrival time of the first ray cluster according to the sound ray number corresponding to the maximum value of the peak in the matched filtering normalization result, and calculating the maximum disturbance time window according to the theoretical maximum peak arrival time in the remaining ray clusters;

[0104] S42, based on the maximum disturbance window, determining the sound line number corresponding to the maximum value of the peak in the matched filtering normalization result;

[0105] S43, defining the sound line number corresponding to the maximum value of the peak value in the matched filtering normalization result as l s1 , the sound line number corresponding to the maximum value of the peak in the matched filtering normalization result is defined as l f1 , will l s1 and l f1 Sort the results of matched filtering normalization by their maximum peak values ​​from large to small;

[0106] S44. Define the number of the sound line with the largest amplitude greater than the sound line eigenstructure in the first ray cluster as R s1 , define the number of the sound line with the largest amplitude smaller than the sound line eigenstructure in the first ray cluster as R f1 , R s1 and R f1Sort by the amplitude of the sound line eigenstructure from large to small;

[0107] S45, will l s1 and l f1 Substitute into the discretized time series and get R s1 and R f1 The actual arrival time corresponding to the sound ray completes the identification of the sound ray arrival peak in the first ray cluster;

[0108] S46, repeat S43-S45 until the identification of the arrival peaks of the sound rays in all ray clusters is completed.

[0109] l s1 Indicates that the sound line number corresponding to the maximum value of the peak in the matched filtering normalization result is greater than the sound line number corresponding to the maximum value of the peak, l f1 Indicates that the sound line number corresponding to the maximum value of the peak in the matched filtering normalization result is smaller than the sound line number corresponding to the maximum value of the peak, R s1 Indicates the number of the sound line with the largest amplitude in the first ray cluster, R f1 Indicates the number of the sound line with the largest amplitude smaller than the sound line eigenstructure in the first ray cluster.

[0110] In the embodiment of the present invention, in S41, the calculation formula of the maximum disturbance time window is:

[0111]

[0112] In the formula, represents the actual maximum peak arrival time of the first ray cluster, represents the theoretical maximum peak arrival time within the remaining ray clusters.

[0113] In the embodiment of the present invention, S4 may specifically be:

[0114] In S41, the matched filtering normalization result calculated by S1 is obtained by using the max function of Matlab. Find the sequence number corresponding to the maximum value of the peak Will Substituting into the discretized time series t[l], we get the actual maximum peak arrival time of the first ray cluster: (For the first ray cluster (n=1), its maximum peak is always The largest one, so You only need the sequence number of the maximum peak value to get For other ray clusters (n ≥ 2), we need to use the Given in The search range of the actual maximum peak arrival time in . and The maximum disturbance time window T max , which is the arrival time of the whole sound line in the sound signal R(t a The maximum disturbance time window (the disturbance of the sound line with the total time length) in time t is used as the time window for extracting all subsequent arrival peaks.

[0115] In S42, Matlab's findpeaks function can give the maximum value of a one-dimensional array and the sequence number of the maximum value in the original array. Using Matlab's findpeaks function and a one-dimensional peak detection method, find the time span T 1 Inside The sequence number corresponding to the peak maximum value l 1 .

[0116] In step S43, the peak maximum value sequence number obtained in step S41 is As the center, the peak maximum value number l 1 Greater than The serial number is divided into l s1 , the peak maximum number l 1 Less than The serial number is divided into l f1 , will l s1 and l f1 According to the corresponding The peak maximum values ​​are sorted from large to small.

[0117] In step S44, the first ray cluster obtained in step S34 has the largest sound ray sequence number of the sound ray eigenstructure amplitude. As the center, R 1 Medium or larger The sound line number is divided into R s1 , R 1 Smaller than The sound line number is divided into R f1 , and R s1 and R f1 Sort by the corresponding intrinsic structure amplitude from large to small.

[0118] In S45, l s1 and l f1 Substituting into the discretized time series t[l], we get R s1 and R f1 The actual arrival time t[l s1 ] and t[l f1 ]. At this point, the identification of the sound ray arrival peak in the first ray cluster is completed.

[0119] In step S46, starting from the nth ray cluster (n≥2), the theoretical maximum peak arrival time of the nth ray cluster obtained in step S34 is As the center, use the max function in T max Inner Seeking The maximum value number of the peak Will Substituting into the discretized time series t[l], we get the actual maximum peak arrival time of ray cluster n:

[0120] In S47, use the findpeaks function to find the time span T n Inside The sequence number corresponding to the peak maximum value l n .

[0121] In step S48, the peak maximum value sequence number obtained in step S46 is As the center, the peak maximum value number l n Greater than The serial number is divided into l sn , the peak maximum number l n Less than The serial number is divided into l fn , will l sn and l fn According to the corresponding The peak maximum values ​​are sorted from large to small.

[0122] In step S49, the sound line number of the sound line of the ray cluster n obtained in step S34 with the largest eigenstructure amplitude is As the center, R n Medium or larger The sound line number is divided into R sn , R n Smaller than The sound line number is divided into R fn , and R sn and R fn Sort by the corresponding intrinsic structure amplitude from large to small.

[0123] In S410, l sn and l fn Substituting into the discretized time series t[l], we get R sn and R fn The actual arrival time t[l sn ] and t[l fn ]. At this point, the identification of the sound ray arrival peak in the nth ray cluster is completed.

[0124] In S411, it is determined whether the number of ray clusters n is greater than N. When n≤N, n=n+1, and steps S46 to S410 are repeated until n>N, and a total of N ray clusters and the actual arrival times of M fundamental characteristic sound rays are output.

[0125] The present invention is further described below with reference to the accompanying drawings and examples.

[0126] Example 1: The data processing of acoustic tomography is divided into the following steps:

[0127] (a) The acoustic signal R(t a ), a total of 172 groups of signals, sampling frequency f s Take 10kHz, perform matched filtering and normalization, and the result is as follows Figure 2 As shown in the figure, the range of the discretized time series t[l] is 11.60s~14.10s, the length is 25000, and l=0,1,…,25000.

[0128] (b) The background sound velocity profile, topography, bottom quality and other data of the experimental sea area, such as Figure 3 and 4 As shown, the experimental geometry and sound source parameters are input into the ray model BELLHOP, including the transmitting depth of 1099m, the receiving depth of 1156m, the propagation distance of 17.3km, the sound source frequency of 5kHz, etc., and the sound line distribution and eigenstructure are calculated. The results are as follows Figure 5 and 6 There are 24 sound lines in total, and the sound line eigenstructures are sorted from small to large according to the theoretical arrival time, as shown in Table 1.

[0129] Table 1

[0130]

[0131]

[0132]

[0133] (c) Press T w Take 0.1s, divide the sound ray eigenstructure into 5 ray clusters in the time domain, and calculate the sound ray number R in each ray cluster n and the time span T of the ray cluster n , the theoretical maximum peak arrival time in each ray cluster is obtained through the max function and the corresponding sound line number (n=1,2,…,5), the results are shown in Table 2.

[0134] Table 2

[0135]

[0136] (d) The divided ray clusters and the sound line numbers in the ray clusters are R n and time span T n , Theoretical maximum peak arrival time and the corresponding sound line number (Table 1 and Table 2), and the matched filtering normalization results ( Figure 3 ) as input, and perform each set of R(t a ) is extracted, and the first group R(t a ) arrives at the corresponding result in the peak extraction process.

[0137] Table 3

[0138]

[0139]

[0140] (e) Use the max function of Matlab to normalize the matched filter results Find the sequence number corresponding to the maximum value of the peak Will Substituting into the discretized time series t[l], the actual maximum peak arrival time of the sound line corresponding to the arrival peak is obtained: Determine the overall sound line arrival time in the sound signal R(t a ) in the maximum disturbance time window T max The time window is 0.02s, which is used as the time window for all subsequent arrival peak extractions.

[0141] (f) Use the findpeaks function to find the time span T 1 within The sequence number l corresponding to the peak maximum value 1 .

[0142] (g) The peak maximum number obtained in step (e) As the center, the peak maximum value number l 1 Greater than The serial number is divided into l s1 , the peak maximum number l 1 Less than The serial number is divided into l f1 , will l s1 and l f1 According to the corresponding The peak maximum values ​​are sorted from large to small.

[0143] (h) The number of the sound line with the largest eigenstructure amplitude of the ray cluster 1 obtained in step (d) As the center, R 1 Medium or larger The sound line number is divided into R s1 , R 1 Smaller than The sound line number is divided into R f1, and R s1 and R f1 Sort by the corresponding intrinsic structure amplitude from large to small.

[0144] (i) s1 and l f1 Substituting into the discretized time series t[l], we get R s1 and R f1 The actual arrival time t[l s1 ] and t[l f1 ](Table 4). The identification of the arrival peak of the sound rays in the first ray cluster is completed.

[0145] (j) Starting from the nth ray cluster (n ≥ 2), the theoretical maximum peak arrival time of the nth ray cluster obtained in step (d) is As the center, use the max function in T max Inner Seeking The maximum value number of the peak Will Substituting into the time t of the acoustic signal R(t), we can get the actual maximum peak arrival time of ray cluster n:

[0146] (k) Use the findpeaks function to find the time span T n Inside The sequence number corresponding to the peak maximum value l n .

[0147] (l) The maximum peak value number obtained in step (j) As the center, the peak maximum value is numbered l n is greater than The serial number is divided into l sn , the peak maximum number l n Less than The serial number is divided into l fn , will l sn and l fn According to the corresponding The peak maximum values ​​are sorted from large to small.

[0148] (m) The number of the sound line with the largest amplitude of the eigenstructure of the sound line in the ray cluster n obtained in step (d) As the center, R n Medium or larger The sound line number is divided into R sn , R n Smaller than The sound line number is divided into R fn , and R sn and R fnSort by the corresponding intrinsic structure amplitude from large to small.

[0149] (n) will l sn and l fn Substituting into the discretized time series t[l], we get R sn and R fn The actual arrival time t[l s1 ] and t[l f1 ](Table 4). The identification of the arrival peak of the sound ray in the nth ray cluster is completed.

[0150] (o) Determine whether the ray cluster n is greater than 5. When n≤5, n=n+1, and continue to repeat steps (j) to (n) until n>5, then output 5 ray clusters, a total of 24 actual arrival times of fundamental sound rays, as shown in Table 4.

[0151] Table 4

[0152]

[0153]

[0154] (p) Repeat steps (e) to (o) for each group of R(t) until the arrival peaks of 172 groups of R(t) are extracted.

[0155] Figure 7 The position of the maximum peak arrival time of ray cluster 1 in the original 172 groups of signals is given. Figure 8-12 The sound ray arrival peak extraction results of all sound rays in 172 groups of signals and five ray clusters are given (the observation time on the vertical axis represents the acquisition time of 172 groups of signals). The extraction results of ray cluster 1 show obvious diurnal tide phenomenon as a whole, and the disturbance of each sound line is basically offset as a whole, indicating that the extraction results of the overall sound line of ray cluster 1 are better; compared with the extraction results of ray cluster 1, the extraction results of ray cluster 2 have poor continuity in the total time length of acoustic tomography, but the whole can still reflect the diurnal tide modulation; although the results of sound line No. 16 in ray cluster 3 reflect the diurnal tide modulation, the subsequent results of sound lines No. 17 to No. 19 do not show the situation of overall offset; only sound line No. 20 in ray cluster 4 reflects the diurnal tide modulation; only sound line No. 24 in ray cluster 5 reflects the diurnal tide modulation. Based on the analysis of the above 24 eigenvalue sound ray arrival peak delay extraction results, it can be seen that the arrival peak delay extraction results of rays 2 to 12 in ray cluster 1, rays 13 to 15 in ray cluster 2, ray 16 in ray cluster 3, ray 20 in ray cluster 4 and ray 24 in ray cluster 5 can be used as input data for acoustic tomography inversion.

[0156] Among them, the sound line trajectory depth coverage range of ray cluster 1 is 300-1100m, and the sound line trajectory depth coverage range of ray clusters 2 to 5 is 0-1100m, which meets the sound line trajectory requirements of full-sea depth acoustic tomography. In the currently commonly used methods, the sound lines in ray clusters 2 to 5 are basically unable to be identified and tracked, and the sound lines with smaller amplitudes in ray cluster 1 are also basically unable to be identified.

[0157] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and should be understood that the protection scope of the present invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific variations and combinations that do not deviate from the essence of the present invention based on the technical revelations disclosed by the present invention, and these variations and combinations are still within the protection scope of the present invention.

Claims

1. A method for extracting fuzzy matching of acoustic tomography arrival peaks based on the eigenstructure of sound rays, characterized in that: The following steps are involved: S1, using an acoustic tomography instrument to receive an acoustic signal, and performing matched filtering processing on the acoustic signal to obtain a matched filtering normalization result; S2, determining a number of sound line eigenstructures; S3, dividing the ray clusters according to the eigenstructures of several sound rays, and determining the number of the sound ray with the largest amplitude of the eigenstructure of the sound ray in each ray cluster and the theoretical maximum peak arrival time; S4. Extract the arrival peak of the acoustic signal according to the normalized result of matched filtering, the number of the sound line with the largest sound line intrinsic structure amplitude in each ray cluster, and the theoretical maximum peak arrival time.

2. The method for extracting the fuzzy matching of the arrival peak of acoustic tomography based on the eigenstructure of sound rays according to claim 1 is characterized in that: The S1 comprises the following sub-steps: S11, using an acoustic tomograph to receive an acoustic signal and calculate the signal amplitude of a matched filter; S12, normalizing the signal amplitude of the matched filter to obtain a matched filter normalization result.

3. The method for extracting the fuzzy matching of the arrival peak of acoustic tomography based on the eigenstructure of sound rays according to claim 2 is characterized in that: In S11, the signal amplitude A of the matched filter C The calculation formula for [t] is: In the formula, C i [t] represents the amplitude of the orthogonal component of the acoustic signal, C q [t] represents the amplitude of the in-phase component of the acoustic signal, and t represents time.

4. The method for extracting the fuzzy matching of the arrival peak of acoustic tomography based on the eigenstructure of sound rays according to claim 2, characterized in that: In S12, the matched filtering normalization result The expression is: In the formula, A C [t] represents the signal amplitude of the matched filter, max(·) represents the maximum value operation, min(·) represents the minimum value operation, and t represents time.

5. The method for extracting the fuzzy matching of the arrival peak of acoustic tomography based on the eigenstructure of sound rays according to claim 1, characterized in that: The S2 comprises the following sub-steps: S21. Set several parameters, input the parameters into the BELLHOP acoustic toolbox, and obtain several sound line eigenstructures; S22. Sort several sound line eigenstructures in ascending order according to theoretical arrival time.

6. The method for extracting the fuzzy matching of the arrival peak of acoustic tomography based on the eigenstructure of sound rays according to claim 1, characterized in that: The S3 comprises the following sub-steps: S31, determining a time window according to a plurality of sound line eigenstructures arranged in ascending order according to theoretical arrival times; S32, dividing the eigenstructures of a plurality of sound rays in the time domain according to time windows to obtain a plurality of ray clusters; S33, accumulating the number of eigenstructures of each ray cluster as the sequence number of the sound ray eigenstructure in each ray cluster; S34, taking the time span between two sound ray eigenstructures with the largest theoretical arrival time difference in each ray cluster as the time span of each ray cluster; S35. According to the time span of each ray cluster, obtain the sound ray number of the sound ray with the largest intrinsic structure amplitude and the theoretical maximum peak arrival time in each ray cluster.

7. The method for extracting the fuzzy matching of the arrival peak of acoustic tomography based on the eigenstructure of sound rays according to claim 1, characterized in that: The S4 comprises the following sub-steps: S41, determining the actual maximum peak arrival time of the first ray cluster according to the sound ray number corresponding to the maximum value of the peak in the matched filtering normalization result, and calculating the maximum disturbance time window according to the theoretical maximum peak arrival time in the remaining ray clusters; S42, determining the sound line number corresponding to the maximum value of the peak in the matched filtering normalization result based on the maximum disturbance window; S43, defining the sound line number corresponding to the maximum value of the peak value in the matched filtering normalization result as l s1 , the sound line number corresponding to the maximum value of the peak in the matched filtering normalization result is defined as l f1 , will l s1 and l f1 Sort the results of matched filtering normalization by their maximum peak values ​​from large to small; S44. Define the number of the sound line with the largest amplitude greater than the sound line eigenstructure in the first ray cluster as R s1 , define the number of the sound line with the largest amplitude smaller than the sound line eigenstructure in the first ray cluster as R f1 , R s1 and R f1 Sort by the amplitude of the sound line eigenstructure from large to small; S45, will l s1 and l f1 Substitute into the discretized time series and get R s1 and R f1 The actual arrival time corresponding to the sound ray completes the identification of the arrival peak of the sound ray in the first ray cluster; S46, repeat S43-S45 until the identification of the arrival peaks of the sound rays in all ray clusters is completed.

8. The method for extracting the fuzzy matching of the arrival peak of acoustic tomography based on the eigenstructure of sound rays according to claim 7, characterized in that: In S41, the calculation formula of the maximum disturbance time window is: In the formula, represents the actual maximum peak arrival time of the first ray cluster, represents the theoretical maximum peak arrival time within the remaining ray clusters.

Images