Cylindrical array sonar system and target detection method based on high-resolution beam forming

Abstract

The invention provides a cylindrical array sonar system and a target detection method based on high-resolution beam forming. The method comprises the steps of converting a beam forming weighted vector of a received sound wave signal into a Kronecker product, constructing a negative logarithm likelihood function in combination with a joint probability density model, and optimizing vectors of an azimuth dimension and a pitching dimension through a preset direction constraint condition so as to form a high-resolution beam; correlation analysis is carried out on echo signals reflected by the target at all azimuth angles and pitch angles and transmitted sound wave signals under different time delays; and according to a correlation analysis result, constructing a three-dimensional matrix including the distance, the azimuth angle and the pitch angle, carrying out normalization processing on the three-dimensional matrix so as to eliminate a signal intensity difference, and outputting a target detection positioning result. According to the invention, through combination of the Kronecker integral solution and three-dimensional matrix modeling, high-resolution joint positioning detection of the target in the azimuth angle, the pitch angle and the distance dimension is realized, the signal intensity difference is effectively eliminated, and the positioning precision is improved.

Description

Technical Field

[0001] The present application relates to the field of sonar technology, and in particular to a cylindrical array sonar system and a target detection method based on high-resolution beamforming. Background Art

[0002] Existing active sonar systems are widely used in underwater detection, navigation, imaging and other fields. They usually use array-type sonar transmitting and receiving elements to transmit signals and receive echoes. In order to improve detection accuracy and coverage, different forms of array configurations are used in existing active sonar systems, including linear arrays, planar arrays, circular arrays, etc. Among them, circular arrays are more common in applications. However, the existing array arrangement has limitations in some cases, especially in terms of target detection accuracy, resolution and target recognition. Traditional sonar systems are usually based on simple array designs, such as linear arrays and rectangular arrays. Although these designs meet basic detection requirements, they pose great challenges in high-precision and high-resolution target detection. Especially in application scenarios that require precise positioning or high-resolution imaging, the array configuration of existing systems cannot effectively reduce signal interference and it is difficult to achieve precise beam control.

[0003] At the same time, existing sonar systems' signal processing methods are also subject to certain limitations. Although some sonar systems use beamforming technology to enhance signals from specific directions by adjusting the phase difference of receiving elements, existing beamforming algorithms still suffer from insufficient noise suppression, signal distortion, and echo interference in complex environments. These issues lead to low target detection accuracy and reliability, especially when the target is far away or in high ambient noise.

[0004] Furthermore, traditional sonar systems typically use time-division mode for signal transmission and reception, which can lead to low detection efficiency. Although some systems utilize full-duplex mode (simultaneous signal transmission and reception), effectively avoiding signal interference, reducing echo distortion, and improving target positioning accuracy remain challenges that need to be addressed. Therefore, with the continuous advancement of technology, there is an urgent need for new sonar systems that can effectively improve detection accuracy and resolution, particularly through optimized array design and the use of advanced beamforming techniques to enhance target detection performance. Summary of the Invention

[0005] The present application provides a cylindrical array sonar system and a target detection method based on high-resolution beamforming, which are used to solve the problems of insufficient noise suppression, signal distortion and echo interference in the prior art, as well as the problems of low target detection accuracy and reliability, low detection efficiency, unavoidable signal interference and echo distortion, and insufficient target positioning accuracy.

[0006] In a first aspect, the present application provides a target detection method based on high-resolution beamforming, comprising:

[0007] After the cylindrical transmitting array transmits the acoustic wave signal and the cylindrical receiving array receives the acoustic wave signal, the received acoustic wave signals of all array elements of all layers of the cylindrical receiving array are combined into an original signal vector;

[0008] The beamforming weight vector of the received acoustic signal is converted into a Kronecker product. Based on the Kronecker product and the joint probability density model, a negative log-likelihood function is constructed. The vectors in the azimuth and elevation dimensions are optimized under preset directional constraints to form a high-resolution beam.

[0009] After forming a high-resolution beam, correlation analysis is performed on the echo signals reflected by the target at various azimuth and elevation angles and the transmitted acoustic wave signals at different time delays.

[0010] Based on the correlation analysis results, a three-dimensional matrix containing the distance, azimuth, and elevation angles of the target to the array element in three-dimensional space is constructed, and the three-dimensional matrix is ​​normalized to eliminate signal strength differences and output the target detection and positioning results.

[0011] Optionally, converting the beamforming weighted vector of the received acoustic wave signal into a Kronecker product, constructing a negative log-likelihood function based on the Kronecker product and a joint probability density model, and optimizing the vectors in the azimuth and elevation dimensions under preset directional constraints to form a high-resolution beam includes:

[0012] Decomposing the beamforming weight vector into vectors in azimuth and elevation dimensions based on the structural characteristics of the concentric rings corresponding to the cylindrical transmit array, and expressing the beamforming weight vector using the Kronecker product of the vector in azimuth and the vector in elevation.

[0013] Combining the conjugate transpose of the beamforming weight vector with the original signal vector to obtain the output signal of the beamformer;

[0014] When the output signal conforms to the Gaussian distribution, constructing a probability density function corresponding to each time frame, and combining the probability density functions corresponding to the time frames to obtain a joint probability density model;

[0015] Transforming the joint probability density model to obtain a negative log-likelihood function to quantify the deviation between the output signal and the Gaussian distribution;

[0016] The correspondence between the steering vectors of the cylindrical receiving array in the azimuth and elevation dimensions and the beamforming weighting vector is used to construct a preset directional constraint condition, where the directional constraint condition indicates that the response of the beam main lobe in the target direction is a preset threshold value;

[0017] Combined with the preset directional constraints, the negative log-likelihood function is minimized to obtain the Lagrangian function;

[0018] By alternately optimizing the vectors of the azimuth and elevation dimensions, the minimum value of the Lagrangian function is iteratively solved until the beam mainlobe width and sidelobe level reach the preset performance threshold to form a high-resolution beam.

[0019] Optionally, the alternately optimizing the vectors of the azimuth dimension and the elevation dimension includes:

[0020] Without adjusting the vector of the pitch dimension, a diagonal loading coefficient is introduced to revise the first covariance matrix to obtain a revised first covariance matrix, and the vector of the azimuth dimension is optimized using the revised first covariance matrix;

[0021] Alternatively, without adjusting the vector in the azimuth dimension, a diagonal loading coefficient is introduced to revise the second covariance matrix to obtain a revised second covariance matrix, and the revised second covariance matrix is ​​used to optimize the vector in the elevation dimension.

[0022] Optionally, in the process of iteratively solving the minimum value of the Lagrangian function until the beam main lobe width and the side lobe level reach a preset performance threshold, the method further includes:

[0023] Dynamically adjust the diagonal loading coefficient according to the noise power of the received signal;

[0024] After dynamically adjusting the diagonal loading coefficient, if the beam main lobe width and side lobe level of the optimized azimuth and elevation vectors do not reach the preset performance threshold, a genetic algorithm is used to randomly perturb the optimized azimuth and elevation vectors until the preset performance threshold is reached or the maximum number of iterations is reached.

[0025] Optionally, the performing correlation analysis on the echo signals reflected by the target at various azimuth and elevation angles and the transmitted acoustic wave signals at different time delays includes:

[0026] Perform wavelet transform on the echo signal and the transmitted sound wave signal to extract the time delay characteristics of different frequency bands;

[0027] Based on the delay characteristics, a sparse delay spectrum is reconstructed by using a compressed sensing technology to obtain delay distribution data;

[0028] Based on the time delay distribution data and in combination with the real-time sound velocity profile data of the target environment, the mapping relationship between time delay and distance is dynamically corrected to generate a target distance mapping table;

[0029] Based on the target distance mapping table, the traversal process of the azimuth and elevation angles is parallelized by a graphics processing unit (GPU) cluster. During the parallel calculation process, each GPU node is assigned to process the mapping relationship between the distance between the corresponding azimuth and elevation angles and the signal strength to obtain the correlation analysis result.

[0030] Optionally, reconstructing a sparse delay profile based on the delay feature by using a compressed sensing technology to obtain delay distribution data includes:

[0031] Decomposing the time delay feature into sparse components at multiple scales, performing independent sparse coding on the sparse components at each scale to generate a multi-scale sparse coefficient matrix;

[0032] According to the signal-to-noise ratio of the sparse components at each scale, reconstruction weights are dynamically allocated to perform weighted fusion on the multi-scale sparse coefficient matrix;

[0033] Based on the weighted fusion results, the sparse delay spectrum is reconstructed using the orthogonal matching pursuit algorithm;

[0034] The reconstructed delay spectrum is subjected to energy focusing analysis. If the focusing is lower than the preset focusing threshold, the scale level of the sparse decomposition is adjusted, and the decomposition, encoding, allocation, weighted fusion, reconstruction, and analysis steps are re-executed until the delay distribution data that meets the preset resolution requirements is generated.

[0035] Optionally, dynamically correcting the mapping relationship between time delay and distance based on the time delay distribution data and combining real-time sound velocity profile data of the target environment to generate a target distance mapping table includes:

[0036] Calculate the sound velocity profile function of the target environment based on the ambient temperature data and depth data;

[0037] The sound velocity profile function is solved by the finite element layered method to obtain the correction coefficient used to correct the mapping relationship;

[0038] The mapping relationship between time delay and distance is dynamically corrected using the correction coefficient to generate a target distance mapping table.

[0039] In a second aspect, the present application provides a cylindrical array sonar system, comprising: a cylindrical transmitting array, a cylindrical receiving array, and a controller, wherein the cylindrical transmitting array is coaxially aligned with the cylindrical receiving array;

[0040] The cylindrical transmitting array is composed of a plurality of concentric rings, and each ring is evenly distributed with a plurality of array elements as transmitting units for transmitting sound wave signals;

[0041] The cylindrical receiving array is composed of a plurality of concentric rings, each of which is evenly distributed with a plurality of array elements as receiving units for receiving sound wave signals;

[0042] The control module is used to control the coordinated operation of the cylindrical transmitting array and the cylindrical receiving array, and to execute the target detection method based on high-resolution beamforming as described in any one of the first aspects.

[0043] Optionally, the transmitting unit and the receiving unit are both piezoelectric sensors, the radius of the ring in the cylindrical transmitting array is different from the radius of the ring in the cylindrical receiving array, and the vertical spacing between the rings in the cylindrical transmitting array is different from the vertical spacing between the rings in the cylindrical receiving array, so as to reduce mutual interference between the transmitted sound wave signal and the received sound wave signal.

[0044] Optionally, the multiple concentric rings in the cylindrical transmitting array include a first main concentric ring and a first backup concentric ring, and the multiple concentric rings in the cylindrical receiving array include a second main concentric ring and a second backup concentric ring.

[0045] In a third aspect, the present application provides a computing device comprising a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement a target detection method based on high-resolution beamforming as described in any one of the first aspects.

[0046] In a fourth aspect, the present application provides a computer storage medium storing a computer program. When the computer program is executed by a computer, it implements a target detection method based on high-resolution beamforming as described in any one of the first aspects.

[0047] The present application provides a target detection method based on high-resolution beamforming, which includes: after a cylindrical transmitting array transmits an acoustic wave signal and a cylindrical receiving array receives the acoustic wave signal, combining the received acoustic wave signals of all array elements in all layers of the cylindrical receiving array into an original signal vector; converting the beamforming weighted vector of the received acoustic wave signal into a Kronecker product, constructing a negative log-likelihood function based on the Kronecker product and a joint probability density model, and optimizing the vectors of the azimuth dimension and the elevation dimension through preset directional constraints to form a high-resolution beam; after the high-resolution beam is formed, performing correlation analysis on the echo signals reflected by the target at various azimuth and elevation angles and the transmitted acoustic wave signals at different time delays; based on the correlation analysis results, constructing a three-dimensional matrix containing the distance, azimuth and elevation angles of the target from the array element in three-dimensional space, and normalizing the three-dimensional matrix to eliminate signal strength differences and output a target detection and positioning result.

[0048] The present application has the following beneficial effects: (1) Through the correlation analysis of the echo signal and the matrix normalization processing, high-precision target detection can be achieved in a complex underwater environment. (2) By adopting the decomposition technology of the Kronecker product, the computational complexity is greatly reduced, so that the method can efficiently process large-scale array data and meet the needs of real-time target detection. (3) The signal strength differences caused by different angles and distances are eliminated through normalization processing, and the stability and consistency of the system in different detection environments are enhanced. (4) By combining the distance, azimuth and pitch angle information of the target, the embodiment of the present application can provide a complete three-dimensional target detection matrix, which is suitable for various three-dimensional space detection applications such as underwater sonar and aerial sonar.

[0049] Furthermore, based on the concentric ring structure of a cylindrical array, this method decomposes the beamforming weight vector into the Kronecker product form of the azimuth and elevation dimensions. The output signal is generated by combining the conjugate transposed weight vector with the original signal. The output signal is defined as conforming to a Gaussian distribution, and a time frame joint probability density model is constructed. This is converted into a negative log-likelihood function to quantify the distribution deviation. A Lagrangian function is constructed in conjunction with directional constraints. The minimum value is iteratively solved by alternately optimizing the azimuth and elevation vectors until the mainlobe width and sidelobe level meet the preset threshold, forming a high-resolution beam. Through azimuth-elevation decoupling optimization and diagonal loading of the covariance matrix, the beam mainlobe width is compressed and the sidelobe level is suppressed, thereby improving the azimuth and elevation resolution. Combining probability density modeling with directional constraints, the method enhances robustness to noise and interference while ensuring the stability of the target directional response. Ultimately, high-precision joint estimation of the target azimuth, elevation, and range parameters is achieved, effectively improving the reliability and accuracy of three-dimensional positioning in complex environments.

[0050] These and other aspects of the present application will become more readily apparent from the description of the following embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0052] Figure 1 A schematic structural diagram of a cylindrical array sonar system 200 provided in an embodiment of the present application;

[0053] Figure 2 A flow chart of a target detection method 100 based on high-resolution beamforming provided in an embodiment of the present application;

[0054] Figure 3 A schematic diagram of the structure of a cylindrical receiving array and an azimuth distance matrix provided in an embodiment of the present application;

[0055] Figure 4 A flowchart of another target detection method based on high-resolution beamforming provided in an embodiment of the present application;

[0056] Figure 5 A schematic diagram of the structure of a computing device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0057] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0058] In some of the processes described in the specification and claims of this application and the above-mentioned figures, multiple operations that appear in a specific order are included, but it should be clearly understood that these operations may not be executed in the order in which they appear in this document or may be executed in parallel. The serial numbers of the operations, such as 11, 12, etc., are only used to distinguish between different operations, and the serial numbers themselves do not represent any order of execution. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this document are used to distinguish different messages, devices, modules, etc., and do not represent a sequential order, nor do they limit "first" and "second" to being different types.

[0059] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without making creative efforts are within the scope of protection of this application.

[0060] In order to solve the problems of insufficient noise suppression, signal distortion and echo interference in the prior art, as well as the problems of low target detection accuracy and reliability, low detection efficiency, unavoidable signal interference and echo distortion, and insufficient target positioning accuracy, the embodiment of the present application provides a target detection method based on high-resolution beamforming, which adopts the following concept: first, based on the multi-layer array element structure of the cylindrical receiving array, the acoustic wave signals received by each layer are integrated into the original signal vector; on this basis, the Kronecker integral decomposition characteristics of the beamforming weighted vector are utilized, and the negative log-likelihood function is constructed in combination with the joint probability density model, and the weighted vectors of the azimuth / elevation dimensions are alternately optimized through directional constraints to achieve high-resolution beamforming with mainlobe width compression and sidelobe suppression; further, the correlation analysis of the transmitted signal and the target echo at different time delays is combined to extract the target distance information, and construct a three-dimensional spatial matrix that integrates the azimuth, elevation and distance; finally, the signal strength difference is eliminated through normalization processing, and a high-resolution, multi-dimensional jointly optimized target three-dimensional positioning result is output.

[0061] Figure 1 A schematic structural diagram of a cylindrical array sonar system 200 provided in an embodiment of the present application is shown in FIG. Figure 1 As shown, the cylindrical array sonar system 200 includes:

[0062] A cylindrical transmitting array 210, a cylindrical receiving array 220 and a controller are provided. The cylindrical transmitting array is aligned with the cylindrical receiving array coaxially 230. The cylindrical transmitting array is composed of a plurality of concentric rings 211 with a radius of R1. The number of concentric rings 211 with a radius of R1 is M1. Each ring is evenly distributed with a plurality of array elements as transmitting units for transmitting acoustic wave signals. Figure 1 As shown in 240 in FIG. The cylindrical receiving array is composed of a plurality of concentric rings 212 with a radius of R2. The number of concentric rings 212 with a radius of R2 is M2. Each ring is evenly distributed with a plurality of array elements as receiving units for receiving acoustic wave signals, in other words, for receiving beams, as shown in FIG. Figure 1 As shown in 250 in FIG. A control module is used to control the coordinated operation of the cylindrical transmitting array and the cylindrical receiving array, and to execute the target detection method based on high-resolution beamforming.

[0063] In one possible embodiment, both the transmitting unit and the receiving unit are piezoelectric sensors. The radius R1 of the rings in the cylindrical transmitting array is different from the radius R2 of the rings in the cylindrical receiving array. For example, R1 is greater than R2. The vertical spacing h1 between the rings in the cylindrical transmitting array is different from the vertical spacing h2 between the rings in the cylindrical receiving array to reduce mutual interference between the transmitted and received acoustic wave signals.

[0064] Specifically, in this embodiment, the cylindrical array sonar system comprises a cylindrical transmitting array and a cylindrical receiving array. Specifically, the cylindrical transmitting array and the cylindrical receiving array are configured in space with different radii and vertical separation, thereby effectively reducing interference and improving target detection accuracy.

[0065] In the embodiments of the present application, for the sake of simplicity, in the following description, the cylindrical transmitting array 210 may be referred to as the transmitting array, the cylindrical receiving array 220 may be referred to as the receiving array, the concentric rings 211 with a radius of R1 may be understood as the annular structure, annular array, etc. of the transmitting array, and the concentric rings 212 with a radius of R2 may be understood as the annular structure, annular array, etc. of the receiving array.

[0066] In this embodiment, within each annular structure of the transmitting array, the annular array of the transmitting array is composed of N1 transmitting units. These transmitting units are evenly distributed on a ring with a radius of R1, forming a circular array of the transmitting array. Each transmitting unit is an independent transmitter, such as a piezoelectric sensor, which converts electrical signals into acoustic signals for transmission. In this circular array, all transmitting units are evenly spaced, that is, the distance between array elements is equal, to ensure uniform signal transmission and coverage.

[0067] The structure of the transmitting array consists of multiple identical circular arrays with a radius of R1, and the number of circular arrays is M1. There is a spacing between each circular array and the previous circular array, which is set to h1. That is, each circular array is parallel to the central axis of the previous circular array, and the spacing in the vertical direction is h1. All circular arrays are structurally coaxially aligned, that is, all circular arrays in the transmitting array share a central axis. The total height of the transmitting array is (M1-1)×h1, which is the vertical distance between all circular arrays multiplied by the number of circular arrays. The entire transmitting array consists of M1×N1 transmitting units, which has a larger coverage range and stronger signal transmission capability.

[0068] In this embodiment, within each annular structure of the receiving array, the annular array of the receiving array is composed of N2 receiving units. These receiving units are evenly distributed on a ring with a radius of R2, forming a circular array of the receiving array. Each receiving unit is also an independent receiver, such as a piezoelectric sensor, which is used to receive the acoustic wave signal reflected from the target and convert it into an electrical signal. In this circular array, all receiving units are evenly spaced, that is, the distance between array elements is equal, to ensure good coverage and uniformity of the received signal.

[0069] The structure of the receiving array consists of multiple identical circular arrays with a radius of R2, and the number of circular arrays is M2. There is a spacing between each circular array and the previous circular array, which is set to h2. That is, each circular array is parallel to the central axis of the previous circular array, and the spacing in the vertical direction is h2. All circular arrays are coaxially aligned in structure, that is, all circular arrays in the receiving array share a central axis to ensure the consistency of the received signal. The total height of the receiving array is (M2-1)×h2, which is the vertical distance between all circular arrays multiplied by the number of circular arrays. The entire receiving array consists of M2×N2 receiving units, which can effectively receive echo signals from different directions and provide a wide detection range.

[0070] In order to provide the relative position of the transmitting array and the receiving array, the embodiment of the present application locates the transmitting array and the receiving array on the same cylindrical structure, and their central axes coincide, that is, they are coaxially aligned. The transmitting array is placed on a larger radius R1, while the receiving array is placed on a smaller radius R2, that is, R1>R2. This configuration can enable the transmitting array to cover a wider area, while the receiving array is more concentrated, which helps to improve the reception accuracy of the echo signal. The vertical positions of the transmitting array and the receiving array are different, the receiving array is located below the transmitting array, and the receiving array is located at the bottom of the cylinder, so that the reflected echo signal can be transmitted from the radiation area of ​​the transmitting array to the receiving array. By separating the transmitting array and the receiving array and adjusting their positions, the signal interference between the two arrays can be effectively reduced, and the target detection capability of the cylindrical array sonar system can be improved.

[0071] In one possible embodiment, the multiple concentric rings in the cylindrical transmitting array include a first primary concentric ring and a first backup concentric ring, and the multiple concentric rings in the cylindrical receiving array include a second primary concentric ring and a second backup concentric ring. To address the issue of poor signal quality at different layers, embodiments of the present application can implement primary / backup switching control.

[0072] Figure 2A flow chart of a target detection method 100 based on high-resolution beamforming is provided in an embodiment of the present application, as shown in FIG. Figure 2 As shown, the method is implemented based on the above cylindrical array sonar system, and the method includes the following steps:

[0073] 110. The system controls all transmitting elements of the cylindrical transmitting array to transmit a signal simultaneously, and the signal is an acoustic wave signal.

[0074] 120. Signal reception includes the following process: Step 121. The system waits to avoid self-interference and 122. All elements of the cylindrical receiving array receive the signal.

[0075] In this step, the cylindrical receiving array adopts a coaxial cylindrical structure. After a waiting time to avoid self-interference, all receiving array elements synchronously collect the echo signal reflected by the target and convert it into an electrical signal.

[0076] 130. Cylindrical array reduced complexity waveform formation, including the following process: step 131, calculating the weight vector w1, step 132, calculating the weight vector w2, step 133, calculating the variance a(t), step 134, scanning the pitch angle and azimuth angle to obtain a beamforming signal.

[0077] 140. Imaging matrix construction includes the following steps: Step 141: Constructing a three-dimensional azimuth distance matrix containing range, azimuth, and elevation information; Step 142: Calculating a normalized azimuth matrix; and Step 143: Performing target detection. This embodiment of the present application eliminates intensity differences through correlation analysis and normalization, ultimately achieving high-precision target detection. For a detailed description of the above process, please refer to the following alternative target detection method based on high-resolution beamforming.

[0078] In summary, low-complexity, high-resolution receive beamforming is achieved by simultaneously transmitting signals from elements in a sonar array. The system employs a low-complexity, high-resolution receive beamforming method proposed in this embodiment, which is particularly suitable for efficient target detection in cylindrical array sonar systems. This method decomposes the beamforming weight vector, effectively reducing computational complexity while maintaining high resolution during signal processing.

[0079] Based on the above method flow, the embodiments of the present application can respectively describe the low-complexity high-resolution receive beamforming process and the three-dimensional range-azimuth matrix construction and normalization process as follows:

[0080] (1) The low-complexity, high-resolution receive beamforming process is as follows:

[0081] S1. The embodiment of the present application first represents the received signal. Specifically, the signal of the j-th array element in the i-th layer at time t is defined as X ij(t), the signals of all array elements can be stacked to form an N2×M2 dimensional vector x(t), or can be represented by an N2×M2 dimensional matrix X(t), where the element in the i-th row and the j-th column is X ij (t).

[0082] S2. Design a beamforming weight vector. Specifically, to reduce computational complexity, the embodiment of the present application defines that the beamforming weight vector w can be decomposed into the Kronecker product form of two vectors: Where w1 is a vector of dimension N2, and w2 is a vector of dimension M2.

[0083] S3, beamformer output, specifically: the output signal Z(t) of the beamformer is Z(t)=w H x(t). Where w H represents the conjugate transpose of the weight vector.

[0084] S4. Define the output signal. Specifically, in this embodiment of the application, the output signal Z(t) of the beamformer is defined to obey a Gaussian distribution. Its probability density function is: Where a(t) is the variance to be estimated at time t.

[0085] S5. Construct a joint probability density function. Specifically, if T time frames of data are collected for beamforming, the output joint probability density is:

[0086] S6. Construct a negative log-likelihood function. Exemplarily, the negative log-likelihood function is:

[0087] S7, determine the constraints of target detection, specifically, the relative benchmark and origin reference in three-dimensional space (such as Figure 3 311 in), the direction angle Θ (as Figure 3 312), and the pitch angle Φ (as Figure 3 313), when performing target detection, the weighted vector is required to satisfy: w H d(Θ,Φ)=1。

[0088] Where d(Θ,Φ) is the steering vector of the cylindrical array in the directions Θ and Φ, d(Θ) is the steering vector of the cylindrical array in the direction Θ, and d(Φ) is the steering vector of the cylindrical array in the direction Φ, which can be expressed as:

[0089] S8. Construct the Lagrangian function. Specifically, considering the direction constraint, minimize the negative log-likelihood function to obtain the Lagrangian function:

[0090] S9, optimizing the weighted vector, specifically, includes the following steps S91 and S92, wherein S91, optimizing w1 by fixing other values ​​to obtain a beamformer. The formula used to optimize w1 is:

[0091]

[0092] R2 represents the revised covariance matrix, which is calculated as:

[0093]

[0094] Where I is the unit matrix, ρ is the diagonal loading coefficient, and the stability of the beamformer can be further improved by increasing τI.

[0095] S92. Fix other values ​​and optimize w2 to obtain a beamformer. The formula used to optimize w2 is:

[0096]

[0097] Among them, R1 represents the revised covariance matrix, and the calculation formula of R1 is:

[0098]

[0099] S10. Optimize the scalar a(t). Specifically, optimize a(t) to obtain a(t) = |w H x(t)| 2 .

[0100] S11. When calculating S91 for the first time, w2 corresponding to the fixed beamformer may be used for initialization, and a(t)=1 may be used to calculate the revised covariance matrix.

[0101] This example proposes a method for constructing a target range and azimuth matrix based on receive beamforming and signal correlation analysis. This method is suitable for target detection in cylindrical array sonar systems. By analyzing the correlation between echo and transmitted signals, this method gradually constructs a three-dimensional matrix of the target, including its range, azimuth, and elevation information, further improving the system's detection accuracy and resolution.

[0102] (2) The construction and normalization process of the three-dimensional distance-azimuth matrix is ​​as follows:

[0103] S21: Receive beamforming and target distance matrix construction, including S211 and S212.

[0104] In step S211, for each given azimuth angle Θ and elevation angle Φ, receive beamforming is performed by combining the received signals of all elements in the array in the aforementioned manner to enhance the target signal. Through this beamforming process, an echo signal in the specified direction is obtained.

[0105] S212: For each azimuth angle Θ and elevation angle Φ, correlation analysis is performed between the echo signal and the transmitted signal at different time delays τ. The cross-correlations between the echo signal and the transmitted signal at different time delays τ are calculated, and the speed of sound is used to infer the echo strength at different distances.

[0106] S22, construction of a three-dimensional distance and orientation matrix, including S221 and step 222.

[0107] In step S221, for each azimuth angle θ and pitch angle Φ, the distance and azimuth information of the target is obtained by analyzing the correlation between the echo signal and the transmitted signal. Specifically, by calculating the correlation between the echo signal and the transmitted signal, the signal reflection intensity in different directions can be obtained.

[0108] S222: Combine the reflection intensity information calculated for each direction angle θ with the corresponding distance value to form a distance and orientation matrix of the target (e.g. Figure 3 321 in ). By performing similar operations on each pitch angle Φ, the reflection intensity of the target at different pitch angles is obtained. Finally, the distance and azimuth information are combined to obtain a three-dimensional matrix, which contains the position, azimuth and pitch angle of the target in three-dimensional space (such as Figure 3 320 in ).

[0109] S23, normalization of three-dimensional distance and orientation matrices, including S231 and S232.

[0110] Among them, S231, for each pitch angle Φ, the embodiment of the present application calculates the sum of the echo signal strength (i.e., the corresponding element value in the distance azimuth matrix) at different angles at each given distance value. Specifically, for each distance position, the embodiment of the present application accumulates the echo intensities at all azimuth angles θ to obtain the total intensity value at that position. Then, the total intensity value is divided by the number of angles (i.e., the number of azimuth angles in 360 degrees) to obtain the average echo intensity of the distance position at that pitch angle.

[0111] S232, after completing the average value calculation of the echo intensity, the embodiment of the present application divides the original echo intensity matrix by the average value to obtain a normalized matrix. Through this normalization process, the intensity differences at different angles and distances can be eliminated, so that the values ​​in the matrix are in a unified dimension in all directions. By normalizing the three-dimensional distance and orientation matrix, the embodiment of the present application can obtain a standardized matrix, making the reflection intensity of the target more comparable. This normalization step eliminates the intensity differences caused by different angles or distances, thereby improving the accuracy and consistency of target detection.

[0112] In summary, this embodiment proposes a method for constructing a target range and azimuth matrix based on receive beamforming and signal correlation analysis, suitable for target detection in cylindrical array sonar systems. This method gradually constructs a three-dimensional matrix of the target, including range, azimuth, and elevation information, by analyzing the correlation between the echo and transmitted signals, further improving the system's detection accuracy and resolution.

[0113] Figure 4 A flowchart of another target detection method based on high-resolution beamforming provided in an embodiment of the present application is shown in FIG. Figure 4 As shown, the method includes:

[0114] S41. After the cylindrical transmitting array transmits the acoustic wave signal and the cylindrical receiving array receives the acoustic wave signal, the received acoustic wave signals of all array elements of all layers of the cylindrical receiving array are combined into an original signal vector.

[0115] S42. Convert the beamforming weighted vector of the received acoustic wave signal into a Kronecker product. Based on the Kronecker product and the joint probability density model, construct a negative log-likelihood function. Optimize the vectors of the azimuth and elevation dimensions through preset directional constraints to form a high-resolution beam.

[0116] S43. After forming a high-resolution beam, correlation analysis is performed on the echo signals reflected by the target at various azimuth and elevation angles and the transmitted acoustic wave signals at different time delays.

[0117] S44. Based on the correlation analysis results, a three-dimensional matrix is ​​constructed that includes the distance, azimuth, and elevation angles of the target to the array element in three-dimensional space, and the three-dimensional matrix is ​​normalized to eliminate signal strength differences and output target detection and positioning results.

[0118] The following is a specific example: After a cylindrical transmitting array synchronously transmits an acoustic signal, the receiving array captures the echo reflected by the target and arranges all array element signals by layer and position into a 48-dimensional original signal vector x(t). The beamforming weight vector w is decomposed into the Kronecker product of the azimuth dimension vector w1 and the elevation dimension vector w2. A negative log-likelihood function is constructed based on the Gaussian distribution definition. Combined with directional constraints, w1 and w2 are iteratively updated through Lagrangian optimization to generate a high-resolution beam. Subsequently, the cross-correlation between the beam signal and the transmitted signal at different time delays τ is calculated. For example, a delay of τ = 0.05m corresponds to a target distance of 3.75m, ultimately determining the actual distance to be 75m. A three-dimensional matrix containing the azimuth, elevation, and distance is constructed and normalized to obtain the normalized intensity value 44. After eliminating environmental interference, the three-dimensional coordinates of the target are output.

[0119] By executing S41 to S44, the embodiment of the present application integrates all element signals in the receiving array to construct a high-density raw vector. This process, combined with Kronecker integral decomposition of weighted vectors and probability model optimization, reduces computational complexity and improves beam resolution. It achieves sub-meter range accuracy and three-dimensional information fusion through delay correlation analysis. Finally, it constructs a normalized three-dimensional matrix, eliminates environmental interference, and outputs the target's three-dimensional coordinates, improving the signal-to-noise ratio. This entire process reduces the computational burden while achieving high-precision, highly anti-interference three-dimensional target detection, making it particularly suitable for high-resolution imaging and real-time positioning in complex underwater environments.

[0120] In one possible embodiment, S42 converts the beamforming weight vector of the received acoustic wave signal into a Kronecker product, constructs a negative log-likelihood function based on the Kronecker product and a joint probability density model, and optimizes the vectors in the azimuth and elevation dimensions using preset directional constraints to form a high-resolution beam, including:

[0121] Step a1: Decompose the beamforming weight vector into vectors of azimuth and elevation dimensions based on the structural characteristics of the concentric rings corresponding to the cylindrical transmitting array, and express the beamforming weight vector using the Kronecker product of the vector of the azimuth dimension and the vector of the elevation dimension.

[0122] Step a2: Combine the conjugate transpose of the beamforming weight vector with the original signal vector to obtain the output signal of the beamformer.

[0123] Step a3: When the output signal conforms to the Gaussian distribution, construct the probability density function corresponding to each time frame, and combine the probability density functions corresponding to the time frames to obtain a joint probability density model.

[0124] Step a4: transform the joint probability density model to obtain a negative log-likelihood function to quantify the deviation between the output signal and the Gaussian distribution.

[0125] Step a5: The corresponding relationship between the steering vectors of the cylindrical receiving array in the azimuth and elevation dimensions and the beamforming weighting vector is used to construct a preset directional constraint condition, where the directional constraint condition indicates that the response of the beam main lobe in the target direction is a preset threshold.

[0126] Step a6: Combine the preset directional constraints and minimize the negative log-likelihood function to obtain the Lagrangian function.

[0127] Step a7: By alternately optimizing the vectors of the azimuth dimension and the elevation dimension, the minimum value of the Lagrangian function is iteratively solved until the beam main lobe width and side lobe level reach the preset performance threshold, thereby forming a high-resolution beam.

[0128] The following is a specific example: After a cylindrical transmitting array synchronously transmits an acoustic signal, the receiving array captures the echo reflected by the sunken ship, arranging all the array element signals by layer and azimuth into a 48-dimensional original signal vector x(t). This is achieved by decomposing the beamforming weight vector w into the Kronecker product of the azimuth dimension vector w1 and the elevation dimension vector w2. A joint probability density model is constructed based on the Gaussian distribution definition. Combined with the directional constraint wHd(30°,10°)=1, w1 and w2 are iteratively optimized using the Lagrangian function to ultimately generate a high-resolution beam.

[0129] By executing steps a1 to a7, the present embodiment achieves high-resolution beamforming through weighted vector decomposition, probabilistic modeling, directional constraint optimization, and alternating iterations. In underwater shipwreck detection, the system accurately locates the 3D coordinates of a shipwreck through 3D matrix construction and normalization, maintaining stable output in strong noise environments.

[0130] In a possible embodiment, step a7, alternately optimizing the vectors of the azimuth dimension and the elevation dimension, includes:

[0131] Step b1: Without adjusting the vector of the pitch dimension, introduce the diagonal loading coefficient to revise the first covariance matrix to obtain the revised first covariance matrix, and use the revised first covariance matrix to optimize the vector of the azimuth dimension.

[0132] The first covariance matrix is ​​a signal statistical characteristic matrix in the azimuth dimension, which is used to describe the correlation of the received signal in the horizontal direction. The fixed weight vector w2 based on the pitch dimension and the received signal data is calculated.

[0133] Step b2: Alternatively, without adjusting the vector in the azimuth dimension, introduce a diagonal loading coefficient to revise the second covariance matrix to obtain a revised second covariance matrix, and use the revised second covariance matrix to optimize the vector in the elevation dimension.

[0134] The second covariance matrix is ​​a signal statistical characteristic matrix in the elevation dimension, which is used to describe the correlation of the received signal in the vertical direction. A fixed weight vector w1 based on the azimuth dimension and the received signal data is calculated.

[0135] The following is a specific example: After synchronously transmitting an acoustic signal in a cylindrical transmitting array, the receiving array captures the echo signal reflected by the sunken ship, forming a 48-dimensional original signal vector. Through step b1, fix the pitch dimension vector w2, calculate the first covariance matrix of the azimuth dimension, and introduce the diagonal loading coefficient ρ = 0.1, revised to Among them, R1 represents the orientation dimension, is the first covariance matrix, ρ is the diagonal loading coefficient, I is the unit matrix, and the azimuth vector is optimized so that the main lobe of the beam is precisely aligned with the azimuth angle of 30°. Then, perform step b2, fix the optimized w1, and calculate the second covariance matrix in the pitch dimension. Also add ρ = 0.1, and revise it to Among them, R2 represents the pitch dimension, is the second covariance matrix, ρ is the diagonal loading coefficient, I is the identity matrix, and the pitch vector is optimized so that the main lobe of the beam is aligned with the pitch angle of 10°.

[0136] By executing steps b1 to b2, the embodiment of the present application improves the stability and pointing accuracy of beamforming by alternately optimizing the weight vectors in the azimuth and elevation dimensions, combined with diagonal loading technology. In complex underwater environments, this method effectively suppresses the interference of multipath reflections and background noise, making the main lobe of the beam more focused and the sidelobe interference lower, thereby achieving high-resolution three-dimensional target positioning. This solution is particularly suitable for scenarios requiring precise detection of underwater targets, enhancing the system's reliability and anti-interference capabilities.

[0137] In a possible embodiment, in step b1, during the process of iteratively solving the minimum value of the Lagrangian function until the beam main lobe width and the side lobe level reach a preset performance threshold, the method further includes:

[0138] Step b11: Dynamically adjust the diagonal loading coefficient according to the noise power of the received signal.

[0139] The noise power of the received signal is a measure of the energy of non-target components in the received signal. It is typically calculated as the signal power during target-free periods or the noise power outside the signal's frequency band. The diagonal loading coefficients adjust the parameters for covariance matrix stability by adding fixed values ​​to the matrix diagonal to avoid numerical computational issues.

[0140] Step b12: After dynamically adjusting the diagonal loading coefficient, if the beam main lobe width and side lobe level of the optimized azimuth and elevation vectors do not reach the preset performance threshold, a genetic algorithm is used to randomly perturb the optimized azimuth and elevation vectors until the preset performance threshold is reached or the maximum number of iterations is reached.

[0141] The mainlobe width of a beam refers to the angular range of the area where the beam energy is concentrated, reflecting directional resolution. The sidelobe level refers to the energy intensity in the non-mainlobe direction of the beam, reflecting the anti-interference capability. Random perturbation refers to the introduction of random changes to the vector parameters to escape the local optimal solution.

[0142] A specific example is shown below: During a 3D shipwreck imaging mission, the receiving array detected high noise power. ρ was adjusted from 0.1 to 0.2, and the covariance matrix was revised to optimize the beam vector. However, the mainlobe width remained at 1.3°. Subsequently, a genetic algorithm was used to generate 10 perturbation vectors. After 15 iterations, the optimal solution was found, reducing the mainlobe width to 0.7° and the sidelobe level to -30 dB. Finally, combined with delay correlation analysis, the 3D coordinates of the shipwreck were output, meeting the requirements for high-precision detection.

[0143] By executing steps b11 and b12, this embodiment dynamically adjusts the diagonal loading coefficient to adaptively adapt to varying noise environments, improving the stability of the beamforming algorithm. Combined with the global search capabilities of the genetic algorithm, this further optimizes the performance of the main lobe and side lobes. These two methods work together to achieve high-resolution target detection in complex underwater environments while avoiding performance bottlenecks caused by local optimal solutions.

[0144] In a possible embodiment, S43, performing correlation analysis on the echo signals reflected by the target at various azimuth angles and elevation angles and the transmitted acoustic wave signals at different time delays, includes:

[0145] Step c1: Perform wavelet transform on the echo signal and the transmitted sound wave signal to extract the time delay characteristics of different frequency bands.

[0146] Wavelet transform is a time-frequency analysis technique that decomposes signals into sub-signals of different frequency bands through scaling and translation of basis functions, achieving both time and frequency resolution. Echo signals are acoustic signals reflected from targets, carrying information such as their range and direction. Transmitted acoustic signals are the original acoustic signals actively transmitted by the sonar system, typically in the form of linear frequency modulation or pulse signals.

[0147] Step c2: Based on the delay characteristics, the sparse delay spectrum is reconstructed through compressed sensing technology to obtain delay distribution data.

[0148] Compressed sensing technology is a signal sampling and reconstruction method that exploits signal sparsity to recover the complete signal from a small amount of observation data. A sparse delay profile is a sparse representation of delay distribution data with a small number of nonzero elements in a specific basis. Delay distribution data is a collection of data describing the signal energy distribution corresponding to different delays.

[0149] Step c3: Based on the time delay distribution data and in combination with the real-time sound velocity profile data of the target environment, the mapping relationship between the time delay and the distance is dynamically corrected to generate a target distance mapping table.

[0150] Real-time sound velocity profile data refers to the real-time measurement of how sound velocity changes with depth in a body of water. The target distance mapping table is a table that maps time delays to actual target distances, dynamically adjusted based on sound velocity.

[0151] Step c4: Based on the target distance mapping table, the traversal process of the azimuth and elevation angles is parallelized by the GPU cluster. During the parallel calculation process, each GPU node is assigned to process the mapping relationship between the distance between the corresponding azimuth and elevation angles and the signal strength to obtain the correlation analysis result.

[0152] A GPU cluster is a parallel computing platform composed of multiple GPUs, suitable for large-scale parallel data processing. Correlation analysis results, which measure the degree of match between signals in each direction and the target distance mapping table, are used to locate the target.

[0153] The following is a specific example: In the three-dimensional imaging task of a shipwreck, the wavelet transform of the sonar echo extracts the multi-frequency delay characteristics, and the sparse delay spectrum is reconstructed through compressed sensing to eliminate the multipath interference peaks. The target distance mapping table is generated by combining the real-time sound velocity profile, and the delay τ = 0.05ms is corrected to a distance of 76.2m. The GPU cluster is used to scan 360×61 directions in parallel, and a heat map is output within 5 minutes, showing that the shipwreck is located at an azimuth of 30°±0.1°, a pitch angle of 10°±0.1°, and a distance of 76.2m±0.2m, meeting the centimeter-level detection accuracy requirements.

[0154] By executing steps c1-c4, this embodiment achieves high-resolution, low-latency three-dimensional target positioning through wavelet multi-frequency time delay extraction, compressed sensing sparse reconstruction, dynamic sound velocity correction, and GPU parallel computing. In complex underwater environments, this system can accurately map target range and direction, improving detection efficiency and is suitable for scenarios such as military reconnaissance and marine resource exploration.

[0155] In a possible embodiment, step c2, reconstructing a sparse delay profile based on the delay characteristics using a compressed sensing technique to obtain delay distribution data, includes:

[0156] Step c21: Decompose the time delay feature into sparse components at multiple scales, perform independent sparse coding on the sparse components at each scale, and generate a multi-scale sparse coefficient matrix.

[0157] A sparse component refers to a small number of components in the delay signature with concentrated energy at a certain scale. Sparse coding represents the signal as a linear combination of a sparse basis, preserving the primary components. A multi-scale sparse coefficient matrix is ​​a matrix composed of the encoding results of the sparse components at each scale.

[0158] Step c22: dynamically assign reconstruction weights based on the signal-to-noise ratio of the sparse components at each scale, and perform weighted fusion on the multi-scale sparse coefficient matrix.

[0159] The signal-to-noise ratio (SNR) is the ratio of signal power to noise power and is used to measure signal quality. The reconstruction weight is the fusion weight assigned to the sparse components at each scale based on the SNR. A higher weight indicates a more reliable scale.

[0160] Step c23: Based on the weighted fusion result, reconstruct the sparse delay spectrum using the orthogonal matching pursuit algorithm.

[0161] Orthogonal matching pursuit is an iterative greedy algorithm that reconstructs sparse signals by gradually selecting optimal basis functions. Sparse delay profile is a sparse representation of delay distribution data, containing only a few peaks.

[0162] Step c24: Perform energy focusing analysis on the reconstructed delay spectrum. If the focusing is lower than the preset focusing threshold, adjust the scale level of the sparse decomposition and re-execute the decomposition, encoding, allocation, weighted fusion, reconstruction, and analysis steps until delay distribution data that meets the preset resolution requirements is generated.

[0163] Energy focusing refers to the ratio of the main peak energy in the delay spectrum to the total energy, which measures the delay resolution. The preset focusing threshold is the minimum requirement for energy focusing.

[0164] The following is a specific example: In the shipwreck positioning task, the delay features are decomposed into three scales and sparsely encoded, and fused based on the weights assigned according to the signal-to-noise ratio; the delay spectrum is reconstructed through orthogonal matching pursuit, and the main peak τ = 0.05ms is found; due to insufficient focus, the scale level is adjusted to 5 layers, and the fusion weights are updated after re-decomposition to 0.7 for high frequency, 0.2 for medium frequency, and 0.1 for low frequency. The final reconstructed delay spectrum has a focus of 78%, and the main peak corresponds to a distance of 75.2m, which meets the detection requirements.

[0165] By executing steps c21-c24, this embodiment of the present application improves the resolution and anti-interference capabilities of the delay spectrum through multi-scale decomposition, signal-to-noise ratio weighted fusion, and focus feedback adjustment. In complex underwater environments, the system can accurately isolate the main peak of the target, improving energy focus and making it suitable for high-precision target detection.

[0166] In one possible embodiment, step c3, based on the delay distribution data and in combination with real-time sound velocity profile data of the target environment, dynamically correcting the mapping relationship between delay and distance to generate a target distance mapping table, includes:

[0167] Step c31: Calculate the sound velocity profile function of the target environment based on the ambient temperature data and depth data.

[0168] The sound velocity profile function refers to a function that describes the change of sound velocity with depth, usually expressed as c(z), where z is the depth.

[0169] Step c32: Solve the sound velocity profile function by using the finite element layered method to obtain a correction coefficient for correcting the mapping relationship.

[0170] The finite element layering method divides the water body into several horizontal layers based on depth. The sound velocity within each layer is treated as a constant, and correction parameters for the sound wave propagation path are calculated through layer-by-layer integration. The correction coefficient is an adjustment factor that describes the effect of sound velocity stratification on the time-delay-distance mapping relationship and is used to compensate for the effects of non-uniform sound velocity.

[0171] Step c33: Dynamically correct the mapping relationship between time delay and distance using the correction coefficient to generate a target distance mapping table.

[0172] The following is a specific example: In a 3D shipwreck location task, conductivity, temperature, and depth sensors collect temperature and salinity data from a depth of 0-100 meters to calculate the sound velocity profile function. The water body is divided into 100 layers, and the finite element layering method is used to calculate the correction coefficient k = 0.0658 s / m. In step c33, the time delay τ = 0.05 seconds is dynamically corrected to the distance, generating a target distance mapping table. Finally, GPU parallel computing is combined to output the shipwreck coordinates, reducing the error by 60% compared to traditional methods.

[0173] By executing steps c31 to c33, the embodiment of the present application improves the accuracy and reliability of time delay and distance conversion through sound speed profile modeling, layered correction calculation, and dynamic mapping table generation. In complex underwater environments, the system can adaptively adjust the influence of the sound speed propagation path, eliminating the systematic errors caused by the traditional uniform sound speed definition, thereby more accurately mapping the target distance. This solution is suitable for high-precision target detection and marine science research, effectively addressing the challenges brought by the non-uniform distribution of sound speed in practical applications, and improving the stability and environmental adaptability of detection results.

[0174] Figure 1 The cylindrical array sonar system can be implemented Figure 2 、 Figure 4 The implementation principles and technical effects of the target detection method based on high-resolution beamforming described in the illustrated embodiment are not further elaborated. The specific manner in which the various modules and units of the cylindrical array sonar system in the aforementioned embodiment perform their operations has been described in detail in the relevant embodiments of the method and will not be further elaborated here.

[0175] In one possible design, the control module in the above system can be implemented as a computing device, such as Figure 5 As shown, the computing device may include a storage component 51 and a processing component 52 .

[0176] The storage component 51 stores one or more computer instructions, wherein the one or more computer instructions are called and executed by the processing component 52 .

[0177] The processing component 52 is used to: after the cylindrical transmitting array transmits the acoustic wave signal and the cylindrical receiving array receives the acoustic wave signal, combine the received acoustic wave signals of all array elements in all layers of the cylindrical receiving array into an original signal vector; convert the beamforming weighted vector of the received acoustic wave signal into a Kronecker product, construct a negative log-likelihood function based on the Kronecker product and the joint probability density model, and optimize the vectors of the azimuth dimension and the elevation dimension through preset directional constraints to form a high-resolution beam; after the high-resolution beam is formed, perform correlation analysis on the echo signals reflected by the target at various azimuth and elevation angles and the transmitted acoustic wave signals at different time delays; based on the correlation analysis results, construct a three-dimensional matrix containing the distance, azimuth and elevation angles of the target from the array element in three-dimensional space, and normalize the three-dimensional matrix to eliminate signal strength differences and output the target detection and positioning results.

[0178] The processing component 52 may include one or more processors to execute computer instructions to complete all or part of the steps in the above method. Of course, the processing component may also be implemented as one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above method.

[0179] The storage component 51 is configured to store various types of data to support operations at the terminal. The storage component can be implemented by any type of volatile or non-volatile memory device, or a combination thereof, such as random access memory (RAM), static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0180] Of course, a computing device may also include other components, such as input / output interfaces, display components, communication components, etc.

[0181] The input / output interface provides an interface between the processing component and the peripheral interface module, which can be an output device, an input device, etc.

[0182] The communication component is configured to facilitate, among other things, wired or wireless communications between the computing device and other devices.

[0183] Among them, the computing device can be a physical device or an elastic computing host provided by a cloud computing platform, etc. In this case, the computing device can refer to a cloud server, and the above-mentioned processing components, storage components, etc. can be basic server resources rented or purchased from the cloud computing platform.

[0184] The present application also provides a computer storage medium storing a computer program, wherein the computer program can achieve the above-mentioned Figure 2 or Figure 4 The illustrated embodiment is a target detection method based on high-resolution beamforming.

[0185] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0186] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0187] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, or of course, by hardware. Based on this understanding, the essence of the above technical solution or the part that contributes to the existing technology can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or certain parts of the embodiments.

[0188] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A target detection method based on high-resolution beamforming, characterized in that: include: After the cylindrical transmitting array transmits the acoustic wave signal and the cylindrical receiving array receives the acoustic wave signal, the received acoustic wave signals of all array elements of all layers of the cylindrical receiving array are combined into an original signal vector; The beamforming weight vector of the received acoustic signal is converted into a Kronecker product. Based on the Kronecker product and the joint probability density model, a negative log-likelihood function is constructed. The vectors in the azimuth and elevation dimensions are optimized under preset directional constraints to form a high-resolution beam. After forming a high-resolution beam, correlation analysis is performed on the echo signals reflected by the target at various azimuth and elevation angles and the transmitted acoustic wave signals at different time delays. Based on the correlation analysis results, a three-dimensional matrix containing the distance, azimuth, and elevation angles of the target to the array element in three-dimensional space is constructed, and the three-dimensional matrix is ​​normalized to eliminate signal strength differences and output the target detection and positioning results.

2. The method according to claim 1, characterized in that The method converts the beamforming weight vector of the received acoustic wave signal into a Kronecker product, constructs a negative log-likelihood function based on the Kronecker product and a joint probability density model, and optimizes the vectors of the azimuth and elevation dimensions under preset directional constraints to form a high-resolution beam, including: Decomposing the beamforming weight vector into vectors in azimuth and elevation dimensions based on the structural characteristics of the concentric rings corresponding to the cylindrical transmit array, and expressing the beamforming weight vector using the Kronecker product of the vector in azimuth and the vector in elevation. Combining the conjugate transpose of the beamforming weight vector with the original signal vector to obtain the output signal of the beamformer; When the output signal conforms to the Gaussian distribution, constructing a probability density function corresponding to each time frame, and combining the probability density functions corresponding to the time frames to obtain a joint probability density model; Transforming the joint probability density model to obtain a negative log-likelihood function to quantify the deviation between the output signal and the Gaussian distribution; The correspondence between the steering vectors of the cylindrical receiving array in the azimuth and elevation dimensions and the beamforming weighting vector is used to construct a preset directional constraint condition, where the directional constraint condition indicates that the response of the beam main lobe in the target direction is a preset threshold value; Combined with the preset directional constraints, the negative log-likelihood function is minimized to obtain the Lagrangian function; By alternately optimizing the vectors of the azimuth and elevation dimensions, the minimum value of the Lagrangian function is iteratively solved until the beam mainlobe width and sidelobe level reach the preset performance threshold to form a high-resolution beam.

3. The method according to claim 2, characterized in that The alternately optimizing the vectors of the azimuth dimension and the elevation dimension includes: Without adjusting the vector of the pitch dimension, a diagonal loading coefficient is introduced to revise the first covariance matrix to obtain a revised first covariance matrix, and the vector of the azimuth dimension is optimized using the revised first covariance matrix; Alternatively, without adjusting the vector in the azimuth dimension, a diagonal loading coefficient is introduced to revise the second covariance matrix to obtain a revised second covariance matrix, and the revised second covariance matrix is ​​used to optimize the vector in the elevation dimension.

4. The method according to claim 3, characterized in that In the process of iteratively solving the minimum value of the Lagrangian function until the beam main lobe width and side lobe level reach a preset performance threshold, the method further includes: Dynamically adjust the diagonal loading coefficient according to the noise power of the received signal; After dynamically adjusting the diagonal loading coefficient, if the beam main lobe width and side lobe level of the optimized azimuth and elevation vectors do not reach the preset performance threshold, a genetic algorithm is used to randomly perturb the optimized azimuth and elevation vectors until the preset performance threshold is reached or the maximum number of iterations is reached.

5. The method according to claim 1, characterized in that The correlation analysis of the echo signal reflected by the target at various azimuth and elevation angles and the transmitted acoustic wave signal at different time delays includes: Perform wavelet transform on the echo signal and the transmitted sound wave signal to extract the time delay characteristics of different frequency bands; Based on the delay characteristics, a sparse delay spectrum is reconstructed by using a compressed sensing technology to obtain delay distribution data; Based on the time delay distribution data and in combination with the real-time sound velocity profile data of the target environment, the mapping relationship between time delay and distance is dynamically corrected to generate a target distance mapping table; Based on the target distance mapping table, the traversal process of the azimuth and elevation angles is parallelized by a graphics processing unit (GPU) cluster. During the parallel calculation process, each GPU node is assigned to process the mapping relationship between the distance between the corresponding azimuth and elevation angles and the signal strength to obtain the correlation analysis result.

6. The method according to claim 5, characterized in that The reconstructing a sparse delay spectrum based on the delay characteristics by using a compressed sensing technology to obtain delay distribution data includes: Decomposing the time delay feature into sparse components at multiple scales, performing independent sparse coding on the sparse components at each scale to generate a multi-scale sparse coefficient matrix; According to the signal-to-noise ratio of the sparse components at each scale, reconstruction weights are dynamically allocated to perform weighted fusion on the multi-scale sparse coefficient matrix; Based on the weighted fusion results, the sparse delay spectrum is reconstructed using the orthogonal matching pursuit algorithm; The reconstructed delay spectrum is subjected to energy focusing analysis. If the focusing is lower than the preset focusing threshold, the scale level of the sparse decomposition is adjusted, and the decomposition, encoding, allocation, weighted fusion, reconstruction, and analysis steps are re-executed until the delay distribution data that meets the preset resolution requirements is generated.

7. The method according to claim 5, characterized in that The method of dynamically correcting the mapping relationship between time delay and distance based on the time delay distribution data and combining the real-time sound velocity profile data of the target environment to generate a target distance mapping table includes: Calculate the sound velocity profile function of the target environment based on the ambient temperature data and depth data; The sound velocity profile function is solved by the finite element layered method to obtain the correction coefficient used to correct the mapping relationship; The mapping relationship between time delay and distance is dynamically corrected using the correction coefficient to generate a target distance mapping table.

8. A cylindrical array sonar system, characterized in that: include: a cylindrical transmitting array, a cylindrical receiving array, and a controller, wherein the cylindrical transmitting array is coaxially aligned with the cylindrical receiving array; The cylindrical transmitting array is composed of a plurality of concentric rings, and each ring is evenly distributed with a plurality of array elements as transmitting units for transmitting sound wave signals; The cylindrical receiving array is composed of a plurality of concentric rings, each of which is evenly distributed with a plurality of array elements as receiving units for receiving sound wave signals; The control module is used to control the coordinated operation of the cylindrical transmitting array and the cylindrical receiving array, and to execute the target detection method based on high-resolution beamforming as described in any one of claims 1 to 7.

9. The system according to claim 8, characterized in that The transmitting unit and the receiving unit are both piezoelectric sensors. The radius of the ring in the cylindrical transmitting array is different from the radius of the ring in the cylindrical receiving array, and the vertical spacing between the rings in the cylindrical transmitting array is different from the vertical spacing between the rings in the cylindrical receiving array, so as to reduce mutual interference between the transmitted sound wave signal and the received sound wave signal.

10. The system according to claim 8 or 9, characterized in that The multiple concentric rings in the cylindrical transmitting array include a first main concentric ring and a first backup concentric ring, and the multiple concentric rings in the cylindrical receiving array include a second main concentric ring and a second backup concentric ring.

Images