Hydrophone array multi-sound-source positioning method based on frost ice optimization algorithm

Abstract

The invention discloses a hydrophone array multi-sound-source positioning method based on a frost ice optimization algorithm, and belongs to the field of sound source positioning. According to the method, a linear hydrophone array is constructed, an underwater sound source positioning scene is set, and a sound source signal is collected and processed to obtain a receiving signal; a MUSIC algorithm is used to process the received signal, a spatial spectrum is generated to preliminarily estimate information such as an azimuth angle and a pitch angle of a sound source, normalization processing is carried out on the spatial spectrum, a gravitational force area and a repulsive force area are divided, and a gravitational force-repulsive force type artificial potential energy field is constructed; an independent sub-population is initialized for each sound source by adopting a frost ice RIME optimization algorithm, individual positions are updated in combination with an artificial potential energy field and a frost puncture mechanism in an iterative updating process, and particles with the highest fitness are selected as final estimation positions through iterative search. The method overcomes the limitation of a traditional optimization algorithm, remarkably improves the precision and reliability of sound source positioning, and is particularly suitable for a multi-sound-source positioning scene in a complex underwater sound environment.

Description

Technical Field

[0001] The present invention belongs to the field of sound source localization, and in particular relates to a hydrophone array multi-sound source localization method based on a frost optimization algorithm. Background Art

[0002] With the continuous development of underwater acoustic signal processing and hydrophone array technology, hydrophone arrays are increasingly being used in underwater exploration applications such as oceans and lakes. They play an irreplaceable role in critical tasks such as noise detection and sound source localization. In marine resource exploration, hydrophone arrays can accurately capture and locate weak acoustic signals released by oil and gas resources, providing a reliable basis for resource development. In environmental monitoring, they can monitor marine biological activity and background noise changes in real time, assisting in the protection and assessment of the marine ecological environment. In terms of national defense and security, sound source localization of underwater targets is a crucial technical support for safeguarding national coastal defense. However, the underwater acoustic environment is complex and changeable, with numerous interfering factors such as biological and mechanical noise, posing significant challenges to sound source localization algorithms based on hydrophone arrays. Against this backdrop, improving the robustness, accuracy, and convergence efficiency of algorithms has become a key research focus in the field of underwater sound source localization.

[0003] Traditional particle swarm optimization algorithms are widely used in sound source localization problems, but they also expose many limitations: their global search capabilities are weak, and particles are easily trapped in local optimality during the search process and find it difficult to escape the local extreme value area, thus failing to obtain the global optimal solution; in addition, the algorithm lacks an adaptive parameter adjustment mechanism, and key parameters such as inertia weight remain fixed throughout the optimization process, resulting in slow convergence in the early stage and easy loss of the optimal solution in the later stage; with iterative updates, the population diversity gradually decreases, and premature convergence is prone to occur, which reduces the quality of the final solution.

[0004] The RIME optimization algorithm, a newly proposed intelligent optimization method based on natural physical phenomena, draws inspiration from the growth and spread of frost ice. By constructing a unique "hard frost ice" piercing search mechanism, it offers both global and localized search capabilities, potentially addressing the shortcomings of traditional algorithms. However, in the application of underwater sound source localization, where interference is significant, the RIME algorithm still has shortcomings: its initial search direction is highly random, making it difficult to quickly focus on a valid search area in complex environments, increasing search blindness and time costs.

[0005] Therefore, it is necessary to propose a multi-sound source localization method for a hydrophone array based on a frost optimization algorithm to solve the above technical problems existing in the prior art. Summary of the Invention

[0006] The object of the present invention is to provide a hydrophone array multi-sound source localization method based on a frost optimization algorithm, so as to improve the robustness, accuracy and convergence efficiency of the sound source localization algorithm based on the hydrophone array.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] A method for localizing multiple sound sources using a hydrophone array based on a frost optimization algorithm comprises the following steps:

[0009] Step 1: Set up an underwater sound source positioning scenario, establish a linear hydrophone array and a reference coordinate system, collect sound source signals, and perform data preprocessing on the collected sound source signals to obtain received signals;

[0010] Step 2: Define the azimuth search grid and pitch angle search grid Calculate the covariance matrix of the received signal and perform eigenvalue decomposition on it to obtain eigenvectors and eigenvalues. Sort the eigenvalues ​​from small to large, extract the eigenvectors corresponding to small eigenvalues ​​to form the noise subspace, and then calculate the MUSIC space spectrum.

[0011] Step 3: Normalize the MUSIC spatial spectrum matrix and set the spectrum peak judgment threshold to construct an attraction-repulsion artificial potential field;

[0012] Step 4: Set the parameters of the frost optimization algorithm and initialize an independent subpopulation for each sound source, including variables in three dimensions: azimuth, elevation, and distance.

[0013] Step 5: Calculate the fitness of each individual in each sound source subpopulation and find the optimal individual in each subpopulation; update the individual position according to the artificial potential field and frost penetration mechanism during the iterative update process, while performing boundary constraints, accepting the better solution and updating the fitness;

[0014] Step 6. After each iteration, the total error of the current iteration is counted and the average error is calculated to analyze the convergence of the error. When the iterative convergence condition is reached, the fitness of all particles of each sound source is calculated again, and the particle with the highest fitness is selected as the final estimated position.

[0015] Compared with the prior art, the present invention has the following beneficial effects:

[0016] (1) The present invention can effectively estimate the approximate direction of the sound source by first using the MUSIC algorithm and relying on its orthogonality analysis of the signal subspace and the noise subspace, providing reliable initial information for subsequent searches, thereby avoiding the blind and inefficient search process caused by the lack of such a priori estimation of the sound source direction in traditional algorithms.

[0017] (2) The introduction of an artificial potential energy field with an attractive and repulsive force further enhances the algorithm's search capabilities. By normalizing the spatial spectrum obtained by the MUSIC algorithm and dividing it into attractive and repulsive regions, particles can be guided more specifically during the search process. When particles are close to the possible sound source location, they are attracted by gravity; when they are in an unfavorable area, they are repelled by repulsion. This mechanism enables particles to converge to the sound source location more quickly, greatly improving the efficiency of positioning.

[0018] (3) The frost penetration mechanism in the RIME optimization algorithm solves the problem that traditional PSO algorithms are prone to falling into local optimality. When the normalized fitness of particles meets certain conditions, particles are allowed to jump to the optimal individual, thereby increasing the diversity of particles and giving the algorithm a greater chance of escaping the local optimal solution and finding the global optimal solution, further improving the positioning accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments.

[0020] Figure 1 This is a flow chart of a method for localizing multiple sound sources using a hydrophone array based on a frost optimization algorithm in the present invention;

[0021] Figure 2 This is the trajectory diagram of the frost and ice optimization sound source positioning in the present invention;

[0022] Figure 3 This is a curve diagram of the average error change during the iterative process of the present invention. DETAILED DESCRIPTION

[0023] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0024] Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative work shall fall within the scope of protection of the present invention.

[0025] Example

[0026] like Figure 1 As shown, this embodiment describes a method for localizing multiple sound sources using a hydrophone array based on a frost optimization algorithm. The method includes the following steps:

[0027] Step 1. In the set underwater sound source positioning scenario, the position of the first hydrophone is used as the reference origin. A coordinate system is established based on this reference origin. The remaining hydrophones are arranged in sequence at equal intervals in the coordinate system to form a linear hydrophone array. The linear hydrophone array is used to collect sound source signals, and then the sound source signals are preprocessed to obtain received signals.

[0028] First determine the array element spacing d and the number of hydrophones N sensors , and based on the determined array element spacing d and the number of hydrophones N sensors Construct a linear hydrophone array (linear hydrophone array) and a reference coordinate system. The position of the first hydrophone is used as the origin of the coordinate system, and the subsequent hydrophones are arranged in sequence with equal spacing according to the array element spacing d. The position of the nth hydrophone is x n =(n-1)d, n=1,2,…,N sensors .

[0029] Sample the sound source signal multiple times and set the number of snapshots N snapshots =200 to improve the reliability and stability of the signal. For each sound source k, first its azimuth and pitch angle Convert to radians and then calculate the spatial direction cosine components:

[0030]

[0031] in, Represents the cosine component of the propagation direction of the sound source k in the x-axis direction, Represents the cosine component of the sound source k in the y-axis direction, Represents the cosine component of the sound source k's propagation direction in the z-axis direction.

[0032] Simultaneously calculate the delay of the signal from the sound source to the nth hydrophone

[0033]

[0034] Among them, x n is the position of the nth hydrophone, c is the speed of sound, is the nth hydrophone position x n The corresponding direction cosine component of the sound source k propagates in the direction.

[0035] This generates an analog signal

[0036]

[0037] in, is the random initial phase, and f is the frequency of the signal.

[0038] The signals of all sound sources are superimposed to obtain the sum s of all sound source signals received by the nth hydrophone. n (t):

[0039]

[0040] Among them, N sources Indicates the total number of sound sources;

[0041] Then add Gaussian white noise n n (t) Get the final received signal r n (t) = s n (t)+n n (t).

[0042] Step 2: Define the azimuth search grid and pitch angle search grid The covariance matrix of the received signal is calculated and eigenvalue decomposition is performed on it to obtain eigenvectors and eigenvalues. The eigenvalues ​​are sorted from small to large, and the eigenvectors corresponding to small eigenvalues ​​are extracted to form the noise subspace. Then, the MUSIC space spectrum is calculated.

[0043] This embodiment defines an azimuth search grid The range is 0° to 360° with a step size of 5°, where i = 1, 2, ..., 72; define the pitch angle search grid The range is -30° to 30° with a step size of 2°, where j = 1, 2, …, 30.

[0044] According to the final received signal r n (t), calculate the covariance matrix R xx :

[0045]

[0046] in, is the signal vector received by all hydrophones at the tth sampling time. snapshots is the number of times the signal is sampled. They represent the 1st, 2nd, ..., Nth at time t. snapshots The received signal of a hydrophone, r H (t) represents the conjugate transpose of r(t);

[0047] Covariance matrix R xx Perform eigenvalue decomposition:

[0048] R xx =EΛE H

[0049] Where E is the eigenvector matrix and Λ is the eigenvalue diagonal matrix.

[0050] After sorting the eigenvalues ​​from small to large, take the last N sensors -M eigenvalues ​​(small eigenvalues, i.e. relatively small eigenvalues) corresponding to the eigenvectors to form the noise subspace E n , M is the number of sound sources.

[0051] For each point in the search grid Calculate the direction vector

[0052]

[0053] in, represents the position coordinates of each hydrophone in the coordinate system, f is the frequency of the signal, and c is the speed of sound.

[0054] Then calculate the MUSIC spatial spectrum and find the spectrum peak to estimate the direction of the sound source. The calculation formula is as follows:

[0055]

[0056] in, Represents the direction vector The conjugate transpose of Denotes the noise subspace E n The conjugate transpose of .

[0057] Step 3: Normalize the MUSIC spatial spectrum matrix and set the spectrum peak judgment threshold to construct an attraction-repulsion type artificial potential field.

[0058] MUSIC spatial spectrum Perform normalization to obtain the normalized spectrum The expression is as follows:

[0059]

[0060] in, Represents the calculated MUSIC spatial spectrum The maximum absolute value.

[0061] The spectrum peak judgment threshold T is set to 0.6, and the search space is divided into an attractive region and a repulsive region according to the value of the normalized spectrum, thereby constructing an attractive-repulsive artificial potential energy field. When , the area is divided into the gravitational area. The gravitational area may contain sound sources, which has an attractive effect on the particles in the subsequent optimization algorithm, guiding the particles to move towards the area where the sound source may exist, providing a clear direction for the search of particles, so that the particles can find the location of the sound source more specifically during the search process. When the sound source is detected, the area is divided into a repulsive area, which repels the particles and guides them to move toward the possible sound source.

[0062] Step 4: Set the parameters of the frost optimization algorithm and initialize an independent subpopulation for each sound source, including variables in three dimensions: azimuth, elevation, and distance.

[0063] Depend on Figure 2 As shown, set the population number N and the maximum number of iterations N of the frost optimization algorithm iter =200, position update proportional coefficient β = 0.5, frost penetration intensity parameter E = 0.1, artificial potential field gravity coefficient α potential =1 and repulsion coefficient β potential =1;

[0064] For each sound source k, initialize a pop There are independent subpopulations of particles, and the position of each particle is determined by the azimuth angle Pitch angle and distance r ik express.

[0065] Step 5: Calculate the fitness of each individual in each sound source subpopulation and find the optimal individual in each subpopulation; during the iterative update process, update the individual position based on the artificial potential field and frost penetration mechanism, and at the same time perform boundary constraints, accept the better solution and update the fitness.

[0066] Design the best fitness function F ik , the function is expressed as follows:

[0067]

[0068] in, and r ik Respectively represent the azimuth, pitch angle and distance of each particle in the independent sub-population initialized by each sound source k, represents the true azimuth of the kth sound source, represents the true pitch angle of the kth sound source, r k Indicates the true distance of the kth sound source. ik Select the particle with the highest current fitness as the optimal individual The fitness function evaluates the closeness between the particle position and the true sound source position. The closer the distance, the higher the fitness, which means that the particle position is more likely to be the true position of the sound source.

[0069] Calculate the cosine factor Controls the direction and amplitude of particle position updates; where N iter The maximum number of iterations for the frost optimization algorithm.

[0070] Normalized spectrum value according to the particle position Calculate the potential field guide value U force And update the particle position

[0071]

[0072] Among them, p ik represents the position of the i-th particle in the population corresponding to the k-th sound source, β represents the position update proportional coefficient, and r1 represents a vector with random properties. The particle conducts a more extensive exploration in the search space to avoid the algorithm falling into a local optimal solution.

[0073] Set boundary limits on the updated position. Calculate the fitness of the updated particle if Then accept the updated position; at the same time, in order to prevent particles from falling into local optimality, the frost puncture mechanism is applied. For each particle i, a random number r3 is generated and the normalized fitness is calculated:

[0074]

[0075] when When , particles are allowed to jump to the optimal individual, explore the new search space, and update the particle position And perform boundary restriction and fitness evaluation. Among them, 10 -5 To prevent the denominator from being 0, which would cause the formula to be incorrect.

[0076] Step 6: After each iteration, the total error of the current iteration is counted and the average error is calculated to analyze the convergence of the error. When the convergence condition is reached, the fitness of all particles of each sound source is calculated again, and the particle with the highest fitness is selected as the final estimated position.

[0077] After each iteration, the total error E of the current iteration is calculated. t :

[0078]

[0079] in, is the optimal individual position of the kth sound source, represents the true azimuth of the kth sound source, represents the true pitch angle of the kth sound source, r k Indicates the true distance of the kth sound source, N sources Indicates the total number of sound sources.

[0080] Then calculate the average error And record it for analyzing the convergence of errors.

[0081] Depend on Figure 3 As shown, when the iterative convergence condition is reached, for each sound source k, the fitness F of all particles is calculated again ik , select the particle with the highest fitness as the final estimated position

[0082] In this embodiment, a method for localizing multiple sound sources using a hydrophone array based on a frost-and-ice optimization algorithm is described. First, a linear hydrophone array is constructed, and an underwater sound source localization scenario is set for the array. Source signals are collected and processed to obtain received signals. The received signals are then processed using the MUSIC algorithm to generate a spatial spectrogram to preliminarily estimate information such as the sound source's azimuth and elevation angles. The spatial spectrogram is then normalized, and the attractive and repulsive regions are divided based on a spectrum peak determination threshold to construct an attractive-repulsive artificial potential field. The frost-and-ice RIME optimization algorithm is then used to initialize an independent subpopulation for each sound source. Individual positions are updated during an iterative update process, combining the artificial potential field and a frost-piercing mechanism. The method continues iteratively until the maximum number of iterations is reached, and the final sound source position estimation parameters are output. This method effectively overcomes the problem of traditional PSO algorithms easily falling into local optimality, significantly improving the accuracy and reliability of sound source localization. It is particularly suitable for multi-source localization scenarios in complex underwater acoustic environments.

[0083] The embodiments of the present invention are only used to illustrate the technical solutions of the present invention rather than to limit the present invention. Those skilled in the art will appreciate that various changes, modifications, substitutions and variations may be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for localizing multiple sound sources using a hydrophone array based on a frost optimization algorithm, characterized in that: The following steps are involved: Step 1: Set up an underwater sound source positioning scenario, establish a linear hydrophone array and a reference coordinate system, collect sound source signals, and perform data preprocessing on the collected sound source signals to obtain received signals; Step 2: Define the azimuth search grid and pitch angle search grid Calculate the covariance matrix of the received signal and perform eigenvalue decomposition on it to obtain eigenvectors and eigenvalues. Sort the eigenvalues ​​from small to large, extract the eigenvectors corresponding to small eigenvalues ​​to form the noise subspace, and then calculate the MUSIC space spectrum. Step 3: Normalize the MUSIC spatial spectrum matrix and set the spectrum peak judgment threshold to construct an attraction-repulsion artificial potential field; Step 4: Set the parameters of the frost optimization algorithm and initialize an independent subpopulation for each sound source, including variables in three dimensions: azimuth, elevation, and distance. Step 5: Calculate the fitness of each individual in each sound source subpopulation and find the optimal individual in each subpopulation; update the individual position according to the artificial potential field and frost penetration mechanism during the iterative update process, while performing boundary constraints, accepting the better solution and updating the fitness; Step 6. After each iteration, the total error of the current iteration is counted and the average error is calculated to analyze the convergence of the error. When the iterative convergence condition is reached, the fitness of all particles of each sound source is calculated again, and the particle with the highest fitness is selected as the final estimated position.

2. The method for multi-sound source localization using a hydrophone array based on frost optimization algorithm according to claim 1, characterized in that: The specific process of step 1 is: First determine the array element spacing d and the number of hydrophones N sensors , and construct a linear hydrophone array and a reference coordinate system. The position of the first hydrophone is used as the origin of the coordinate system. Subsequent hydrophones are arranged in sequence with equal spacing according to the array element spacing d. The position of the nth hydrophone is x n =(n-1)d, where n=1, 2, ..., N sensors ; Set the number of snapshots N snapshots , the sound source signal is sampled multiple times, and then the collected sound source signal is preprocessed to obtain the received signal.

3. The method for multi-sound source localization using a hydrophone array based on frost optimization algorithm according to claim 1, characterized in that: In step 1, the data preprocessing process is as follows: For each sound source k, first its azimuth and pitch angle Convert to radians and then calculate the spatial direction cosine components: in, Represents the cosine component of the propagation direction of the sound source k in the x-axis direction, Represents the cosine component of the sound source k in the y-axis direction, Represents the cosine component of the sound source k's propagation direction in the z-axis direction; Simultaneously calculate the delay of the signal from the sound source to the nth hydrophone Among them, x n is the position of the nth hydrophone, c is the speed of sound, is the nth hydrophone position x n The cosine component of the propagation direction of the sound source k corresponding to the direction; This generates an analog signal in, is the random initial phase, f is the frequency of the signal; The signals of all sound sources are superimposed to obtain the sum s of all sound source signals received by the nth hydrophone. n (t): Among them, N sources Indicates the total number of sound sources; Then add Gaussian white noise n n (t) Get the final received signal r n (t) = s n (t)+n n (t).

4. The method for localizing multiple sound sources using a hydrophone array based on a frost optimization algorithm according to claim 1, wherein: In step 2, the contents of the azimuth search grid and the elevation search grid are defined as follows: Define the azimuth search grid The range is 0° to 360° with a step size of 5°, where i = 1, 2, ..., 72; Define the pitch angle search grid The range is -30° to 30° with a step size of 2°, where j = 1, 2, …, 30.

5. The method for multi-sound source localization using a hydrophone array based on frost optimization algorithm according to claim 1, characterized in that: In step 2, the specific process of calculating the covariance matrix and performing eigenvalue decomposition on it is as follows: Calculate the covariance matrix R based on the received signal xx , and its calculation formula is: Among them, N snapshots is the number of times the signal is sampled. r1(t), r2(t), ..., They represent the 1st, 2nd, ..., Nth at time t respectively. snapshots The received signal of a hydrophone, r H (t) represents the conjugate transpose of r(t); Then the covariance matrix R xx Perform eigenvalue decomposition, and the calculation formula is: R xx =EΛE H Where E is the eigenvector matrix and Λ is the eigenvalue diagonal matrix.

6. The method for multi-sound source localization using a hydrophone array based on frost optimization algorithm according to claim 1, characterized in that: In step 2, the specific process of calculating the MUSIC spatial spectrum is as follows: For each point in the search grid Calculate the direction vector in, x1, x2, ..., represents the position coordinates of each hydrophone in the coordinate system, f is the frequency of the signal, and c is the speed of sound; Then calculate the MUSIC spatial spectrum Find the spectrum peak to estimate the direction of the sound source. The calculation formula is as follows: in, express The conjugate transpose of Denotes the noise subspace E n The conjugate transpose of .

7. The method for multi-sound source localization using a hydrophone array based on frost optimization algorithm according to claim 1, characterized in that: The specific process of step 3 is as follows: Normalize the MUSIC spatial spectrum to obtain the normalized spectrum The expression is as follows: in, Represents the calculated MUSIC spatial spectrum The maximum value among the absolute values; Set the peak judgment threshold T, when When , the region is defined as the gravitational region; when When , the area is defined as the repulsive area, thereby constructing an attraction-repulsion type artificial potential energy field.

8. The method for multi-sound source localization using a hydrophone array based on frost optimization algorithm according to claim 1, characterized in that: The specific process of step 4 is as follows: Set the population size N and maximum number of iterations N of the frost optimization algorithm iter , position update proportional coefficient β, frost penetration intensity parameter E, artificial potential field gravity coefficient α potential and the repulsion coefficient β potential ; For each sound source k, initialize a pop An independent subpopulation of particles, each particle's position is determined by its azimuth angle Pitch angle and distance r ik express.

9. The method for multi-sound source localization using a hydrophone array based on frost optimization algorithm according to claim 1, characterized in that: The specific process of step 5 is as follows: First, design the fitness function F with the best distance ik , the function is expressed as follows: in, and r ik Respectively represent the azimuth, pitch angle and distance of each particle in the independent sub-population initialized by each sound source k, represents the true azimuth of the kth sound source, represents the true pitch angle of the kth sound source, r k Indicates the true distance of the kth sound source; through the fitness function F ik Select the particle with the highest current fitness as the optimal individual Then calculate the cosine factor Among them, N iter The maximum number of iterations for the frost optimization algorithm; Normalized spectrum value according to the particle position Calculate the potential field guide value U force And update the particle position Among them, p ik represents the position of the i-th particle in the population corresponding to the k-th sound source, β represents the position update proportional coefficient, and r1 represents a vector with random properties; Calculate the fitness of the updated particles if Then accept the updated position; at the same time, apply the frost puncture mechanism, for each particle i, generate a random number r3, and calculate the normalized fitness: when When , particles are allowed to jump to the optimal individual and update the particle position And perform boundary constraints and fitness evaluation.

10. The method for multi-sound source localization using a hydrophone array based on frost optimization algorithm according to claim 1, characterized in that: The specific process of step 6 is as follows: After each iteration, the total error E of the current iteration is calculated. t : in, is the optimal individual position of the kth sound source, represents the true azimuth of the kth sound source, represents the true pitch angle of the kth sound source, r k Indicates the true distance of the kth sound source, N sourres Indicates the total number of sound sources; Then calculate the average error When the iterative convergence condition is reached, for each sound source k, the fitness F of all particles is calculated again ik , select the particle with the highest fitness as the final estimated position

Images