Single snapshot direction finding method and system based on quantum artificial bee bird mechanism, and storage medium

Abstract

The invention discloses a single-snapshot direction finding method and system based on a quantum artificial bee bird mechanism and a storage medium. The method comprises the following steps: firstly, establishing a uniform linear array single-snapshot sampling signal model, and constructing a covariance matrix and an orthogonal projection matrix to obtain a direction finding equation; initializing a hummingbird group and parameters; the fitness is compared to obtain a global optimal territory equal quantum position; guiding the hummingbird to find food, and updating the position of the optimal hummingbird honey collection target quantum by using a revolving door; hummingbird territorial foraging, updating the quantum position and mapping the quantum position to an actual position, and comparing fitness to obtain a global optimal quantum position; and judging whether the maximum number of iterations is reached or not, if not, repeating the step of guiding foraging, if so, terminating, mapping the global optimal quantum position to an angle interval, and outputting an estimated angle. Compared with an existing single-snapshot direction-of-arrival estimation algorithm, the method has higher accuracy and higher robustness, single-snapshot data is processed, the number of snapshots is greatly reduced, the real-time performance of the system is improved, and the operation complexity is reduced.

Description

Technical Field

[0001] The present invention belongs to the field of array signal processing, and particularly relates to a single snapshot direction finding method, system and storage medium based on a quantum artificial hummingbird mechanism. Background Technique

[0002] The Direction of Arrival (DOA) estimation technique is a key research direction in the field of modern array signal processing. Its core objective is to determine the direction of a signal source relative to an array by analyzing the signals collected by a receiving antenna array. This technique has a wide range of applications in multiple fields, including radar systems, sonar detection, wireless communication networks, electronic warfare, seismic monitoring, and medical imaging. By accurately estimating the direction of the signal source, the DOA technique can improve the performance of these systems in target positioning, signal processing, and data analysis. Traditional DOA estimation techniques are mainly based on spatial spectrum estimation theory. By decomposing the covariance matrix of the received signal, the signal subspace and the noise subspace are separated, and then the arrival direction of the signal is estimated. Typical algorithms such as the MUSIC (Multiple Signal Classification) algorithm and the ESPRIT (Estimation of Signal Parameters via Rotational Invariance Techniques) algorithm. These two typical algorithms both decompose the covariance matrix of the received signal into a signal subspace and a noise subspace, and use the orthogonal relationship between the two to estimate the DOA of the signal. However, since these basic algorithms were proposed in a Gaussian white noise environment, the common problem of these algorithms is that they cannot be extended to an impulsive noise environment, have low real-time performance, and large computational complexity, resulting in inapplicability in many practical environments. With the progress of technology, researchers have proposed a variety of improved algorithms, continuously developing and improving the DOA estimation technique.

[0003] After retrieving the existing literature, Wang Lei, Gao Xiang, etc. proposed constructing a pseudo-covariance matrix with a rank equal to the number of signal sources after rearranging the single snapshot received signals in "Direction of Arrival Estimation of Single Snapshot LDACS System Based on Higher Order Power" published in the Journal of Northwestern Polytechnical University (2024, Vol. 42, No. 2, pp. 362 - 367). He Shun, etc. proposed minimizing the nuclear norm and truncated kernel function regularization in "DOA Estimation Based on Covariance Matrix Weighted Truncated Nuclear Norm" published in Modern Radar (2023, Vol. 45, No. 7, pp. 51 - 55), which improves the resolution to a certain extent and reduces the number of snapshots.

[0004] Single-snapshot DOA estimation can improve the real-time performance of the system, enhance the system adaptability, and reduce the computational complexity compared with multi-snapshot. However, the reduction in the number of snapshots will also lead to inaccurate estimation performance and poor robustness. At the same time, the estimation performance of most direction-finding methods under impulsive noise is also not ideal. Therefore, a high-accuracy single-snapshot direction-finding method under impulsive noise needs to be designed. The present invention designs a single-snapshot direction-finding method based on the quantum artificial hummingbird mechanism, and combines the low-order moment matrix of the cosine kernel correlation entropy to accurately determine the direction of the signal source. The simulation results show that this method can achieve single-snapshot direction-finding, and is more accurate and can decorrelate under impulsive noise compared with the MUSIC algorithm and the multi-pigeon flock algorithm. Summary of the Invention

[0005] The purpose of the present invention is to provide a single-snapshot direction-finding method, system and storage medium based on the quantum artificial hummingbird mechanism, so as to solve the problem that the DOA estimation performance is inaccurate due to the reduction in the number of snapshots and the change of interference noise under single-snapshot and impulsive noise conditions.

[0006] The purpose of the present invention is achieved through the following technical solutions:

[0007] A single-snapshot direction-finding method based on the quantum artificial hummingbird mechanism, the specific steps include:

[0008] Step 1: Establish a single-snapshot sampling signal model of a uniform linear array;

[0009] Step 2: Use the snapshot data received by the array to construct a covariance matrix based on the cosine kernel infinity norm, and then use the steering matrix of the uniform linear array to construct an orthogonal projection matrix to obtain a single-snapshot maximum likelihood direction-finding equation based on the cosine kernel infinity norm;

[0010] Step 3: Initialize the hummingbird flock and corresponding parameters;

[0011] Step 4: Compare the fitness of each position with the average fitness of the territory where each hummingbird is located to obtain the foraging target quantum position of the global optimal territory, the optimal quantum position within the optimal territory, and the global optimal quantum position up to the g-th generation;

[0012] Step 5: Guide the foraging of hummingbirds. Hummingbirds tend to choose positions with high fitness for foraging; update through the quantum rotation gate to obtain the m-th dimension of the k-th foraging target quantum position of the hummingbird in the optimal territory;

[0013] Step 6: Territorial foraging of hummingbirds; map the updated quantum position after territorial foraging of hummingbirds to the corresponding actual position, and calculate the fitness value of the territory where the i-th hummingbird is located; compare the fitness of the newly generated quantum position with the original quantum position; compare the original quantum position, and then obtain the global optimal quantum position up to the (g + 1)-th generation;

[0014] Step 7: Determine whether the maximum number of iterations G is reached max , if not, let g = g + 1, and return to Step 5; if so, terminate the iteration, map the obtained global optimal quantum position to the angular interval, and obtain the estimated angle output.

[0015] Further, the specific content of Step 1 is as follows:

[0016] Establish a uniform linear array, where all array elements are arranged at equal intervals on the same straight line, the number of array elements is N, the element spacing is d, and the wavelength is λ, where There are M far-field narrowband signals incident on the array from the θ i direction respectively, and the incident signals and the noise signals are uncorrelated, i = 1, 2,..., M; the first array element is used as the reference array element, then the signal received by the k-th array element at time t is y k (t):

[0017]

[0018] where s i (t) is the incident signal, and n k (t) is the noise signal of the k-th array element, k = 1, 2,..., N, and the single snapshot signal model received by the array is:

[0019] y(1) = A(θ)s(1) + n(1)

[0020] where y(1) = [y1(1), y2(1),..., y N (1)] T , A(θ) = [a(θ1), a(θ2),..., a(θ M )] N×M is the steering matrix, where the i-th steering vector is θ = [θ1, θ2,..., θ M is the incoming wave azimuth vector; s(1) = [s1(1), s2(1),..., s M (1)] T is the signal vector, and n(1) = [n1(1), n2(1),..., n N (1)] T is the impulse noise vector of the array, and the impulse noise satisfies the noise model with the SαS characteristic exponent of α stable distribution.

[0021] Further, the specific content of Step 2 is as follows:

[0022] Construct a pseudo-covariance matrix using the single snapshot data received by the array:

[0023]

[0024] Process the constructed pseudo-covariance matrix \(R\) y to obtain the low-order moment matrix based on the cosine kernel correlation entropy as follows:

[0025]

[0026] where the element is expressed as:

[0027]

[0028] where \(R\) y [:, k] is the vector composed of all elements in the \(k\)-th column of \(R\) y , \(i = 1, 2, \ldots, N\), \(n = 1, 2, \ldots, N\); \((\cdot)\) * represents taking the conjugate, \(R\) y (i, k) is the element in the \(i\)-th row and \(k\)-th column of \(R\) y , \(R\) y [i, :] is the vector composed of all elements in the \(i\)-th row of \(R\) y , med is the median of the calculation array, \(\sigma\) 2 represents the kernel length; the orthogonal projection matrix is \(P\) A(θ) = \(A(\theta)(A\) H (\theta)A(\theta)) -1 \(A\) H (\theta)\), and the angle estimation value of the infinite norm single snapshot maximum likelihood direction finding equation based on the cosine kernel is where \((\cdot)\) H represents the conjugate transpose, and \(tr(\cdot)\) represents the matrix trace function.

[0029] Furthermore, the specific content of step three is as follows:

[0030] Initialize a hummingbird swarm. The number of hummingbirds in the swarm is \(Q1\). Place \(Q1\) hummingbirds on \(Q1\) food sources. The \(m\)-th dimensional quantum position of the \(i\)-th hummingbird in the \(g\)-th generation is randomly initialized between \([0, 1]\), \(i = 1, 2, \ldots, Q1\), \(m = 1, 2, \ldots, M\), which is expressed as generating the quantum position of the \(i\)-th hummingbird in the \(g\)-th generation in the hummingbird swarm, that is Each hummingbird needs to generate some potential quantum positions within the range \([0, 1]\), that is, the initial honey collection target quantum positions of the hummingbird. Among them, the number of honey collection target quantum positions is \(Q2\), and the range radius \(R=\rho\), \(\rho\in(0, 1)\). The area within the range radius is the territory of the hummingbird; the quantum position of the \(k\)-th honey collection target corresponding to the \(i\)-th hummingbird in the \(g\)-th generation in space is It should be emphasized here that the honey collection target already includes the quantum position where the hummingbird is located after the above processing. Take the quantum position where the hummingbird is located as the first honey collection target position, that is Map the quantum position in the m-th dimension to the actual position in the m-th dimension as follows: where and are the upper and lower bounds of the position in the m-th dimension, and the mapping to the actual position is Hummingbird migration rate r b , number of migration generations G t , maximum number of iterations G max ; Since the range of the quantum position of the honey collection target of the hummingbird needs to be shrunk or expanded after a certain number of generations, the shrinking and expanding generations G s are set, and the shrinking coefficient is μ; Map the quantum position to the actual position Find the corresponding fitness value as The average fitness of the i-th hummingbird and the honey collection target positions around it is the fitness value of the territory where the i-th hummingbird is located as

[0031] Furthermore, the specific content of step four is as follows:

[0032] By comparing the fitness of all positions with the average fitness of the territory where the hummingbird is located, the honey collection target quantum position with the maximum average fitness is obtained as the global optimal territory Obtain the optimal quantum position within the optimal territory until the g-th generation Global optimal quantum position up to the g-th generation

[0033] Furthermore, the specific content of step five is as follows:

[0034] Each honey collection target of each hummingbird randomly selects three flight modes according to the bionic mechanism. Axial flight indicates that the hummingbird flies along any one direction in the m-th dimension. Diagonal flight indicates that the hummingbird moves from one corner to another corner of a rectangle formed by the axes of any two dimensions in the m-th dimension. Omnidirectional flight indicates that flight updates are performed in any direction in the m-th dimension;

[0035] Thus, the determination rule of the m-th dimension direction variable of the k-th honey collection target of the i-th hummingbird:

[0036] Axial flight:

[0037] Diagonal flight:

[0038] Omnidirectional flight:

[0039] where is the direction vector of the k-th honey collection target of the i-th hummingbird in the g-th generation, which is used to characterize the update method of each hummingbird corresponding to the honey collection target. Among them is the m - dimensional direction variable of the k - th nectar - collecting target of the i - th hummingbird in the g - th generation, where m = 1, 2, …, M, and randi([1,M]) represents generating a random integer from 1 to M; h represents the length of the generated sequence, and randperm(h) represents creating a random permutation of integers from 1 to h. P l is the l - th dimension of this random permutation, where l ∈ [1, h]; r i g (k) is a random number in (0, 1]; the quantum rotation angle update equation obtained by implementing the guided foraging mechanism for the m - th dimension of the k - th nectar - collecting quantum position of the i - th hummingbird in a non - optimal territory is:

[0040]

[0041] where, is a random number in the range [0, 1] as the guiding factor; is the m - th dimension of the k - th nectar - collecting quantum position of the i - th hummingbird at the g - th iteration;

[0042] The m - th dimension of the k - th nectar - collecting quantum position of the i - th hummingbird updated through the quantum rotation gate is:

[0043]

[0044] The hummingbirds in the optimal territory will also search for new food sources, and thus explore food sources with greater fitness; therefore, the hummingbirds will move towards the global - optimal quantum position with greater fitness as the goal to search for new food sources as territories with higher fitness values than the current ones; the m - th dimension quantum rotation angle update equation obtained by implementing the territorial foraging mechanism for all nectar - collecting quantum positions in the optimal territory is:

[0045]

[0046] where, is a random number in the range [0, 1] as the regional foraging factor; the m - th dimension of the k - th nectar - collecting quantum position of the hummingbird in the optimal territory updated through the quantum rotation gate is:

[0047]

[0048] Furthermore, the specific content of step six is as follows:

[0049] Map all nectar - collecting target quantum positions to actual positions, calculate the fitness values of each nectar - collecting target quantum position and the position, and the average fitness value of each territory, and sort them. The territory with the smallest average fitness is the worst territory; a migration generation number G t is defined during initialization; if the number of iterations exceeds the migration generation number G t, the hummingbird corresponding to the worst territory will migrate to a new position randomly generated in the entire search space by comparing the average fitness of each territory; at this time, the hummingbird and the nectar - collecting target positions around this hummingbird will all be updated; this process effectively avoids the algorithm falling into a local optimal solution; when the number of iterations exceeds the contraction and expansion algebra G s When, the foraging range of the hummingbird will contract or expand accordingly; when the distance between the k - th nectar - collecting target quantum position of the i - th hummingbird in the g - th generation and the hummingbird quantum position is greater than μ, the nectar - collecting target range contracts. It should be noted that at this time k≠1, that is, the k - th nectar - collecting target position cannot be the position where the hummingbird is currently located, and the position where the hummingbird is located is That is, k = 1, and its quantum rotation angle and quantum position update equations are:

[0050]

[0051] where is a random number in the range of [0, 1] as the contraction factor; when the distance between the k - th nectar - collecting target quantum position of the i - th hummingbird in the g - th generation and the hummingbird quantum position is less than μ, the nectar - collecting target range expands, and the obtained quantum rotation angle and quantum position update equations are:

[0052]

[0053] Here is a random number in [0, ρ] as the expansion factor;

[0054] Map the updated quantum position of the hummingbird after foraging in the territory to the corresponding actual position Calculate the fitness of the k - th position of the i - th hummingbird in the (g + 1)-th generation as The average fitness of the i - th hummingbird and the nectar - collecting target positions around it is the fitness value of the territory where the i - th hummingbird is located If the fitness of the newly generated quantum position is greater than the original quantum position, that is, f i g+1 (k)>f i g (k), then replace the original quantum position with the new quantum position for storage; otherwise, the stored is still the original quantum position, that is, let f i g+1 (k)=f i g (k), Compare that the one with the maximum average fitness is the nectar - collecting target quantum position of the global optimal territory Furthermore, obtain the optimal quantum position within the optimal territory until the (g + 1)-th generation Thus, obtain the global optimal quantum position up to the (g + 1)-th generation

[0055] A computer device / system / apparatus, comprising a memory, a processor, and a computer program stored on the memory, wherein the processor executes the computer program to implement the steps of a single-snapshot direction finding method based on a quantum artificial hummingbird mechanism.

[0056] A computer-readable storage medium, on which a computer program / instructions are stored, and when the computer program / instructions are executed by a processor, the steps of a single-snapshot direction finding method based on a quantum artificial hummingbird mechanism are implemented.

[0057] A computer program product, comprising a computer program / instructions, and when the computer program / instructions are executed by a processor, the steps of a single-snapshot direction finding method based on a quantum artificial hummingbird mechanism are implemented.

[0058] The beneficial effects of the present invention are as follows:

[0059] 1. The present invention realizes the direction-of-arrival estimation under impulse noise conditions based on the correlation entropy constructed by the array received signal based on the cosine kernel, and can accurately estimate the direction of arrival in an environment with very low signal-to-noise ratio and strong impulse noise.

[0060] 2. The present invention processes single-snapshot data, greatly reducing the number of snapshots, improving the real-time performance of the system, and reducing the computational complexity.

[0061] 3. The present invention can maintain excellent performance under both coherent source and independent source conditions, indicating that the algorithm designed by the present invention can effectively decorrelate.

[0062] 4. The designed single-snapshot direction finding method based on the quantum artificial hummingbird mechanism has higher accuracy and higher robustness compared with the existing single-snapshot direction-of-arrival estimation algorithms. Description of the Drawings

[0063] Figure 1 It is a flowchart of a single-snapshot direction finding method based on a quantum artificial hummingbird mechanism;

[0064] Figure 2 It is a comparison diagram of the quantum artificial hummingbird algorithm and the MUSIC algorithm for direction finding in the case of impulse noise of three independent sources;

[0065] Figure 3 It is a comparison diagram of the quantum artificial hummingbird algorithm and the MUSIC algorithm for direction finding in the case of impulse noise of three coherent sources;

[0066] Figure 4 It is a comparison diagram of the quantum artificial hummingbird algorithm and the multi-pigeon flock algorithm for direction finding in the case of impulse noise of three independent sources;

[0067] Figure 5Quantum artificial hummingbird algorithm and multi-pigeon flock algorithm for direction finding in the case of three coherent source impact noises; comparison chart

[0068] Figure 6 Relationship diagram between estimation success probability and characteristic exponent

[0069] Figure 7 Relationship diagram between estimation success probability and generalized signal-to-noise ratio Specific implementation manner

[0070] The present invention will be further described below with reference to the accompanying drawings.

[0071] The overall process of the single snapshot direction finding method based on the quantum artificial hummingbird mechanism of the present invention will be further described in detail below in conjunction with the accompanying drawings and specific implementation manners. The technical solution of the present invention includes the following steps: Figure 1 As shown, the technical solution of the present invention includes the following steps:

[0072] Step 1: Establish a single snapshot sampling signal model for a uniform linear array

[0073] Establish a uniform linear array, satisfying that all array elements are arranged at equal intervals on the same straight line, the number of array elements is N, the array element spacing is d, and the wavelength is λ, where There are M far-field narrowband signals incident on the array from θ i directions respectively, and the incident signals and noise signals are uncorrelated, i = 1, 2,..., M. The first array element is used as the reference array element, then the signal received by the kth array element at time t is y k (t):

[0074]

[0075] where s i (t) is the incident signal, n k (t) is the noise signal of the kth array element, k = 1, 2,..., N, and the single snapshot signal model received by the array is:

[0076] y(1) = A(θ)s(1) + n(1) In the formula, y(1) = [y1(1), y2(1),..., y N (1)] T , A(θ) = [a(θ1), a(θ2),..., a(θ M )] N×M

[0077] is the steering matrix, where the ith steering vector is θ = [θ1, θ2,..., θ M is the incoming wave azimuth vector; s(1) = [s1(1), s2(1),..., s M (1)] Tis the signal vector, n(1) = [n1(1), n2(1), …, n N (1)] T is the impulse noise vector of the array, and the impulse noise satisfies the noise model with the SαS characteristic exponent of α stable distribution.

[0078] Step 2: Use the snapshot data received by the array to construct the covariance matrix based on the cosine kernel infinity norm, then use the steering matrix of the uniform linear array to construct the orthogonal projection matrix, and obtain the single-snapshot maximum likelihood direction finding equation based on the cosine kernel infinity norm.

[0079] Use the single-snapshot data received by the array to construct the pseudo-covariance matrix:

[0080] Process the constructed pseudo-covariance matrix R y to obtain the low-order moment matrix based on the cosine kernel correlation entropy as:

[0081]

[0082] The element is expressed as:

[0083]

[0084] where R y (:, k) is the vector composed of all elements in the k-th column of R y , i = 1, 2, …, N, n = 1, 2, …, N. (·) * represents taking the conjugate, R y (i, k) is the element in the i-th row and k-th column of R y , R y (i, :) is the vector composed of all elements in the i-th row of R y , med is to calculate the median of the array, σ 2 represents the kernel length. The orthogonal projection matrix is P A(θ) = A(θ)(A H (θ)A(θ)) -1 A H (θ), and the angle estimation value of the single-snapshot maximum likelihood direction finding equation based on the cosine kernel infinity norm is where (·) H represents the conjugate transpose, and tr(·) represents the matrix trace function.

[0085] Step 3: Initialize the hummingbird swarm and corresponding parameters

[0086] Initialize a hummingbird swarm. The number of hummingbirds in the swarm is Q1. Place Q1 hummingbirds on Q1 food sources. The m-th dimensional quantum position of the i-th hummingbird in the g-th generation Randomly initialize between [0, 1], i = 1, 2, …, Q1, m = 1, 2, …, M, It is expressed as generating the quantum position of the i-th hummingbird in the g-th generation of the hummingbird swarm, that is, Each hummingbird needs to generate some potential quantum positions within the range of [0, 1], that is, the initial foraging target quantum positions of the hummingbird. Among them, the number of foraging target quantum positions is Q2, and the range radius R = ρ, ρ ∈ (0, 1). The area within the range radius is the territory of the hummingbird. The quantum position of the k-th foraging target corresponding to the i-th hummingbird in the g-th generation in space is Here, it should be emphasized that the foraging target already includes the quantum position where the hummingbird is located after the above processing. Taking the quantum position where the hummingbird is located as the first foraging target position, that is, Map the quantum position of the m-th dimension to the actual position in the m-th dimension as follows: Among them and are the upper and lower bounds of the position in the m-th dimension, and the mapping to the actual position is The migration rate r of the hummingbird b , the migration generation G t , the maximum number of iterations G max . Since the foraging target quantum position range of the hummingbird needs to be shrunk or expanded after a certain number of generations, the shrinkage and expansion generation G s is set, and the shrinkage coefficient is μ. Map the quantum position to the actual position Calculate the corresponding fitness value as The average fitness of the i-th hummingbird and its surrounding foraging target positions, that is, the fitness value of the territory where the i-th hummingbird is located, is

[0087] Step 4: Compare the fitness of each position with the average fitness of the territory where each hummingbird is located to obtain the foraging target quantum position of the global optimal territory, the optimal quantum position within the optimal territory, and the global optimal quantum position up to the g-th generation.

[0088] By comparing the fitness of all positions with the average fitness of the territory where the hummingbird is located, the foraging target quantum position of the global optimal territory with the largest average fitness is obtained Obtain the optimal quantum position within the optimal territory up to the g-th generation The global optimal quantum position up to the g-th generation

[0089] Step 5: Guided foraging of hummingbirds

[0090] Hummingbirds tend to choose locations with high fitness for foraging, so the designed quantum artificial hummingbird algorithm is a maximum fitness optimization problem. Each foraging target of each hummingbird randomly selects three flight modes according to the bionic mechanism: axial, diagonal, and omnidirectional. Axial flight indicates that the hummingbird can fly in any direction in the m-dimensional space; diagonal flight indicates that the hummingbird moves from one corner to another corner of a rectangle formed by the axes of any two dimensions in the m-dimensional space; omnidirectional flight indicates that the hummingbird updates in any direction in the m-dimensional space. Thus, the determination rule of the m-dimensional direction variable of the k-th foraging target of the i-th hummingbird is as follows:

[0091] Axial flight:

[0092] Diagonal flight:

[0093] Omnidirectional flight:

[0094] Among them is the direction vector of the k-th foraging target of the i-th hummingbird in the g-th generation, which is used to characterize the update method of each hummingbird's corresponding foraging target. Among them is the m-dimensional direction variable of the k-th foraging target of the i-th hummingbird in the g-th generation, m = 1, 2,..., M, randi([1, M]) represents generating a random integer from 1 to M; h represents the length of the generated sequence, randperm(h) represents creating a random permutation of integers from 1 to h, and P l is the l-th dimension of this random permutation, l ∈ [1, h]; r i g (k) is a random number in (0, 1]. The quantum rotation angle update equation obtained by implementing the guided foraging mechanism for the m-th dimension of the k-th foraging quantum position of the i-th hummingbird in the non-optimal territory is:

[0095]

[0096] Among them is a random number in the range of [0, 1] as the guiding factor; is the m-th dimension of the k-th foraging target quantum position of the i-th hummingbird at the g-th iteration.

[0097] The m-th dimension of the k-th foraging target quantum position of the i-th hummingbird updated through the quantum rotation gate is:

[0098]

[0099] Hummingbirds in the optimal territory will also search for new food sources, thereby exploring food sources with greater fitness. Therefore, hummingbirds will move towards the global optimal quantum position with greater fitness to search for new food sources as territories that may have a higher fitness value than the current one. The update equation of the m-th dimensional quantum rotation angle obtained by implementing the territorial foraging mechanism on all nectar-foraging quantum positions at the optimal territory is as follows:

[0100]

[0101] where is a random number within the range of [0, 1] as the regional foraging factor; the m-th dimension of the k-th nectar-foraging target quantum position of the optimal territory hummingbird updated through the quantum rotation gate is:

[0102]

[0103] Step Six: Territorial Foraging of Hummingbirds

[0104] Map all nectar-foraging target quantum positions to actual positions, calculate the fitness values of each nectar-foraging target quantum position and the position, and sort the average fitness values of each territory. The territory with the smallest average fitness is the worst territory. A migration generation G t is defined during initialization. If the number of iterations exceeds the migration generation G t , then by comparing the average fitness of each territory, the hummingbird corresponding to the worst territory will migrate to a new position randomly generated in the entire search space. At this time, this hummingbird and the nectar-foraging target positions around this hummingbird will all be updated. This process effectively prevents the algorithm from falling into local optimal solutions. When the number of iterations exceeds the contraction and expansion generation G s , the foraging range of the hummingbird will contract or expand accordingly. When the distance between the k-th nectar-foraging target quantum position of the i-th hummingbird in the g-th generation and the hummingbird quantum position is greater than μ, the foraging target range contracts. It should be noted that at this time k≠1, that is, the k-th nectar-foraging target position cannot be the position where the hummingbird is currently located, and the position where the hummingbird is located is that is, k = 1, and its quantum rotation angle and quantum position update equations are:

[0105]

[0106] where is a random number within the range of [0, 1] as the contraction factor. When the distance between the k-th nectar-foraging target quantum position of the i-th hummingbird in the g-th generation and the hummingbird quantum position is less than μ, the foraging target range expands, and the obtained quantum rotation angle and quantum position update equations are:

[0107]

[0108] Here A random number in [0, ρ] is used as the expansion factor.

[0109] Map the updated quantum position of the hummingbird after territorial foraging to the corresponding actual position Calculate the fitness of the k-th position of the i-th hummingbird in the (g + 1)-th generation as The average fitness of the i-th hummingbird and the nectar-foraging target positions around it is the fitness value of the territory where the i-th hummingbird is located, which is If the fitness of the newly generated quantum position is greater than that of the original quantum position, i.e., f i g+1 (k) > f i g (k), then replace the original quantum position with the new quantum position for preservation; otherwise, the original quantum position is still preserved, that is, let f i g+1 (k) = f i g (k). Compare that the one with the maximum average fitness is the nectar-foraging target quantum position of the global optimal territory Furthermore, obtain the optimal quantum position within the optimal territory until the (g + 1)-th generation Thus, obtain the global optimal quantum position up to the (g + 1)-th generation

[0110] Step Seven: Determine whether the maximum number of iterations G is reached max , if not, let g = g + 1, and return to Step Five; if reached, terminate the iteration, and map the obtained global optimal quantum position to the angular interval to obtain the estimated angle output.

[0111] The present invention is applied to the problem of DOA estimation. In the existing methods, the estimation performance is not very good under impulsive noise, and most of the methods with better estimation performance require hundreds or thousands of snapshots to achieve accurate estimation. By constructing the low-order moment of the correlation entropy based on the cosine kernel function from the received single-snapshot data, the covariance matrix is reconstructed, and the maximum likelihood estimation based on the infinite norm covariance matrix of the cosine kernel function and the quantum artificial hummingbird algorithm are designed to estimate the direction of the signal source with high precision. The steps of this method are as follows: establish a single-snapshot sampling signal model for a uniform linear array; construct a covariance matrix based on the correlation entropy of the cosine kernel according to the single-snapshot data received by the array, so as to obtain the maximum likelihood estimation equation for a single snapshot; after initializing the population and related parameters, the hummingbirds perform guided foraging and territorial foraging; map the quantum positions of the hummingbirds and their nectar-foraging targets to the actual positions, calculate the fitness values of the positions of the hummingbirds and their nectar-foraging targets and the average fitness value of each territory for sorting, update the positions of the hummingbirds and their nectar-foraging targets, and to prevent falling into local optimal solutions, search for foraging by shrinking and expanding the range of the nectar-foraging targets. After multiple iterations, the global optimal quantum position is finally obtained and output, and mapping the quantum position to the actual position is the estimated value of the direction of arrival of the incoming wave. The present invention realizes the processing of single-snapshot data under the condition of impulsive noise to obtain the direction-of-arrival estimation of the array received signal, makes the application environment of DOA estimation more extensive, improves the real-time performance of the system at the same time, and realizes the high-precision direction finding of the target incoming wave.

[0112] In the simulation experiment, three signals are incident on a uniform linear array with the number of array elements N = 128 and the element spacing from three directions of [60°, 20°, -10°] respectively. The number of signal sources M = 3, the kernel length of the cosine kernel function is 5, and the maximum number of iterations G max = 200. The number of experimental runs is 30, the number of hummingbirds Q1 = 20, the number of nectar-foraging target positions of each hummingbird Q2 = 20, the upper bound and the lower bound are 90° and -90° respectively, the range radius R = ρ = 0.06, the migration generation G t = 10, the shrinking and expanding generation G s = 50, and the shrinking coefficient is μ = 10 -7 . The estimation success probability is defined as the ratio of the number of runs where the angle estimation value and the accurate angle value differ by less than or equal to 1° to the total number of experimental runs.

[0113] Figure 2In the simulation experiment, there are three independent signal sources [60°, 20°, -10°], the impact noise characteristic index is 1.4, the generalized signal-to-noise ratio is 3 dB, and the number of experimental runs is 30 times. As can be seen from the figure, the angle estimation value of the MUSIC algorithm is very different from the accurate angle value and has completely failed; there is basically no deviation between the angle estimation value of the quantum artificial hummingbird algorithm, that is, the AHA algorithm, and the accurate angle value. Therefore, under the condition of independent sources, the quantum artificial hummingbird algorithm can accurately estimate the direction of independent signal sources in an impact noise environment.

[0114] Figure 3 In the simulation experiment, there are three coherent signal sources [60°, 20°, -10°], the impact noise characteristic index is 1.4, the generalized signal-to-noise ratio is 3 dB, and the number of experimental runs is 30 times. As can be seen from the figure, the angle estimation value of the MUSIC algorithm is very different from the accurate angle value and has completely failed; there is basically no deviation between the angle estimation value of the quantum artificial hummingbird algorithm, that is, the AHA algorithm, and the accurate angle value. Therefore, under the condition of coherent sources, the quantum artificial hummingbird algorithm can accurately estimate the direction of coherent signal sources in an impact noise environment.

[0115] Figure 4 In the simulation experiment, there are three independent signal sources [60°, 20°, -10°], the impact noise characteristic index is 1.4, the generalized signal-to-noise ratio is 3 dB, and the number of experimental runs is 30 times. For the simulation results of the multi-pigeon flock algorithm, that is, the PIO algorithm, nearly half of them have large deviations, indicating that the robustness of this algorithm is not good; there is only 1 inaccurate estimation in 30 simulation experiments of the quantum artificial hummingbird algorithm, that is, the AHA algorithm. Therefore, under the condition of independent sources, the quantum artificial hummingbird algorithm can accurately and robustly estimate the direction of independent signal sources in an impact noise environment.

[0116] Figure 5 In the simulation experiment, there are three coherent signal sources [60°, 20°, -10°], the impact noise characteristic index is 1.4, the generalized signal-to-noise ratio is 3 dB, and the number of experimental runs is 30 times. For the simulation results of the multi-pigeon flock algorithm, that is, the PIO algorithm, nearly half of them have large deviations, indicating that the robustness of this algorithm is not good; there is basically no deviation between the angle estimation value of the quantum artificial hummingbird algorithm, that is, the AHA algorithm, and the accurate angle value. Therefore, under the condition of coherent sources, the quantum artificial hummingbird algorithm can accurately and robustly estimate the direction of coherent signal sources in an impact noise environment.

[0117] Figure 6 The change curve of the estimation success probability and the characteristic index is plotted. From the figure, as the characteristic index increases, the estimation success probability shows an upward trend, and the success probability of the quantum artificial hummingbird algorithm is higher than that of the other two algorithms when the characteristic indices are the same.

[0118] Figure 7The curve of the estimated success probability versus the signal-to-noise ratio is plotted. From the figure, as the signal-to-noise ratio increases, the estimated success probability shows an upward trend, and at the same signal-to-noise ratio, the success probability of the quantum artificial hummingbird algorithm is higher than that of the other two algorithms.

[0119] The function of a single-snapshot direction-finding system based on the quantum artificial hummingbird mechanism in an embodiment of the present invention can be illustrated by the aforementioned single-snapshot direction-finding method based on the quantum artificial hummingbird mechanism. Therefore, for the parts not described in detail in the system embodiment, reference can be made to the above method embodiment, and details will not be repeated here.

[0120] An embodiment of the present invention also discloses an electronic device, including a memory and a processor. Among them, the memory is used to store a program; the processor is coupled to the memory and is used to execute the program stored in the memory to implement the steps in the aforementioned single-snapshot direction-finding method based on the quantum artificial hummingbird mechanism.

[0121] An embodiment of the present invention also discloses a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the above single-snapshot direction-finding method based on the quantum artificial hummingbird mechanism.

[0122] It should be noted that the logic and / or steps represented in the flowchart or described in other ways herein, for example, can be considered as a definite sequence list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch and execute instructions from the instruction execution system, apparatus, or device), or in combination with these instruction execution systems, apparatus, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. More specific examples (non-exhaustive list) of computer-readable media include the following: an electrical connection portion with one or more wirings (electronic device), a portable computer diskette (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpretation, or other appropriate processing as necessary, and then stored in a computer memory.

[0123] It should be understood that each part of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logic functions on data signals, application specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc.

[0124] An embodiment of the present invention further provides a computer program product, which, when running on a computer, executes a single-snapshot direction finding method based on a quantum artificial hummingbird mechanism. Among them, the computer can be a general-purpose computer, a special-purpose computer, a computer network or other programmable devices.

[0125] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A single-snapshot direction finding method based on the quantum artificial hummingbird mechanism, characterized in that: The specific steps include: Step 1: Establish a uniform linear array single snapshot sampling signal model; Step 2: Use the snapshot data received by the array to construct a covariance matrix based on the infinite norm of the cosine kernel, and then use the steering matrix of the uniform linear array to construct an orthogonal projection matrix to obtain the infinite norm single snapshot maximum likelihood direction finding equation based on the cosine kernel; Step 3: Initialize the hummingbird group and corresponding parameters; Step 4: Compare the fitness of each position with the average fitness of each hummingbird's territory, obtain the nectar collection target quantum position of the global optimal territory, the optimal quantum position in the optimal territory, and the global optimal quantum position of the g-th generation; Step 5: Guided foraging of hummingbirds. Hummingbirds tend to choose locations with high fitness for foraging. The mth dimension of the kth nectar-collecting target quantum position of the hummingbird with the optimal territory is obtained through the quantum revolving door. Step 6: Foraging in the territory of the hummingbird; Map the updated quantum position after foraging in the hummingbird territory to the corresponding actual position, calculate the fitness value of the territory where the i-th hummingbird is located; Compare the fitness of the newly generated quantum position with the original quantum position; Compare the original quantum position, and then obtain the global optimal quantum position up to the g+1th generation; Step 7: Determine whether the maximum number of iterations G has been reached max If it is not reached, set g = g + 1 and return to step 5; if it is reached, terminate the iteration and map the obtained global optimal quantum position to the angle interval to obtain the estimated angle output.

2. A single-snapshot direction finding method based on the quantum artificial hummingbird mechanism according to claim 1, characterized in that: The step 1 is specifically as follows: Establish a uniform linear array so that all array elements are arranged on the same straight line with equal spacing. The number of array elements is N, the array element spacing is d, and the wavelength is λ, where There are M far-field narrowband signals from θ i The direction is incident on the array, and the incident signal and the noise signal are uncorrelated, i = 1, 2, ..., M; the first array element is used as the reference array element, then the signal received by the kth array element at time t is y k (t): Among them, s i (t) is the incident signal, n k (t) is the noise signal of the kth array element, k = 1, 2, ..., N, and the array receives a single snapshot signal model: y(1)=A(θ)s(1)+n(1) where y(1)=[y1(1),y2(1),…,y N (1)] T , A(θ)=[a(θ1),a(θ2),…,a(θ M )] N×M is the steering matrix, where the i-th steering vector is θ=[θ1,θ2,…,θ M ] is the wave direction vector; s(1)=[s1(1),s2(1),…,s M (1)] T is the signal vector, n(1)=[n1(1),n2(1),…,n N (1)] T is the impulse noise vector of the array, and the impulse noise satisfies the noise model of SαS stable distribution with characteristic index α.

3. The single-snapshot direction finding method based on the quantum artificial hummingbird mechanism according to claim 1, characterized in that: The step 2 is specifically as follows: Construct a pseudo-covariance matrix using the single snapshot data received by the array: For the constructed pseudo-covariance matrix R y After processing, the low-order moment matrix based on the cosine kernel correlation entropy is obtained: Among them, the element It is expressed as: Among them, R y (:,k) is R y A vector consisting of all elements in the kth column of , i = 1, 2, ..., N, n = 1, 2, ..., N; (·) * stands for conjugation, R y (i,k) is R y The element in the i-th row and k-th column of R y (i,:) is R y The vector consisting of all elements of the i-th row of , med is the median of the calculated array, σ 2 represents the core length; the orthogonal projection matrix is ​​P A(θ) =A(θ)(A H (θ)A(θ)) -1 A H (θ), the angle estimate based on the infinite norm single-shot maximum likelihood direction finding equation of the cosine kernel is in(·) H represents the conjugate transpose, and tr(·) represents the matrix trace function.

4. The single-snapshot direction finding method based on the quantum artificial hummingbird mechanism according to claim 1, characterized in that: The step three is specifically as follows: Initialize a swarm of hummingbirds, the number of hummingbirds in the swarm is Q1, place Q1 hummingbirds on Q1 food sources, and the m-dimensional quantum position of the i-th hummingbird of the g-th generation is Initialize randomly between [0,1], i=1,2,…,Q1, m=1,2,…,M, It is expressed as the quantum position of the i-th hummingbird in the g-th generation of the hummingbird group, that is, Each hummingbird needs to generate some potential quantum positions in the range [0,1], which are the initial nectar collection target quantum positions of the hummingbird. The number of nectar collection target quantum positions is Q2, and the range radius is R = ρ, ρ∈(0,1). The area within the range radius is the hummingbird’s territory. The quantum position of the kth nectar collection target corresponding to the i-th hummingbird of the g-th generation in space is k=1,2,…,Q2. It should be emphasized here that the nectar collection target has included the quantum position of the hummingbird after the above processing. The quantum position of the hummingbird is taken as the first nectar collection target position. The m-th dimension quantum position is mapped to the m-th dimension actual position as follows: i=1,2,…,Q1,where and is the upper and lower bounds of the m-th dimension position, mapped to the actual position as Hummingbird migration rate b , migration algebra G t , the maximum number of iterations G max ; Since the quantum position range of the hummingbird's honey collection target needs to be shrunk or expanded after a certain number of generations, the contraction and expansion algebra G is set s , the contraction coefficient is μ; Mapping quantum positions to real positions Find the corresponding fitness value The average fitness of the i-th hummingbird and its surrounding nectar collection target locations, that is, the fitness value of the i-th hummingbird’s territory, is 5. The single-snapshot direction finding method based on the quantum artificial hummingbird mechanism according to claim 1, characterized in that: The step 4 is specifically as follows: By comparing the fitness of all positions with the average fitness of the hummingbird's territory, the nectar-collecting target quantum position with the largest average fitness is obtained as the global optimal territory. Get the optimal quantum position in the optimal territory up to the gth generation Up to the g-th generation global optimal quantum position 6. The single-snapshot direction finding method based on the quantum artificial hummingbird mechanism according to claim 1, characterized in that: The step five is specifically as follows: Each hummingbird has three flight modes for each nectar-collecting target randomly selected according to the bionic mechanism. Axial flight means that the hummingbird flies in any direction of m dimensions. Diagonal flight means that the hummingbird moves from one corner of a rectangle formed by the axes of any two dimensions in m dimensions to another corner. Omnidirectional flight means that flight updates are performed in any direction of m dimensions. Therefore, the m-th directional variable of the k-th nectar-collecting target of the i-th hummingbird Determination rules: Axial Flight: Diagonal flight: Omnidirectional flight: in, is the direction vector of the kth nectar collection target of the i-th hummingbird of the g-th generation, which is used to represent the update method of each hummingbird's corresponding nectar collection target. is the m-th directional variable of the k-th nectar collection target of the i-th hummingbird of the g-th generation, m = 1, 2, ..., M, randi ([1, M]) means generating a random integer from 1 to M; h represents the length of the generated sequence, randperm (h) means creating a random permutation of integers from 1 to h, P l is the lth dimension of the random permutation, l∈[1,h]; is a random number between (0,1]; the quantum rotation angle update equation obtained by implementing the guided foraging mechanism for the kth nectar-collecting quantum position of the mth dimension of the i-th hummingbird in the non-optimal territory is: in, is a random number in the range [0,1] as a bootstrap factor; is the m-th dimensional quantum position of the k-th nectar target of the i-th hummingbird at the g-th iteration; The mth dimension of the kth nectar-collecting target quantum position of the i-th hummingbird is updated through the quantum revolving gate as follows: The hummingbird in the optimal territory will also look for new food sources, and then explore food sources with greater fitness; therefore, the hummingbird will move with the global optimal quantum position with greater fitness as the goal, looking for new food sources as a territory with a higher fitness value than the current one; the update equation of the m-th dimension quantum rotation angle obtained by implementing the territory foraging mechanism for all nectar-collecting quantum positions in the optimal territory is: in, is a random number in the range of [0,1] as the regional foraging factor; the mth dimension of the kth nectar-collecting target quantum position of the optimal territory hummingbird is updated through the quantum revolving gate:

7. The single-snapshot direction finding method based on the quantum artificial hummingbird mechanism according to claim 1, characterized in that: The step six is ​​specifically as follows: Map all honey-collecting target quantum positions to actual positions, calculate the fitness value of each honey-collecting target quantum position and the average fitness value of each territory, and sort them. The territory with the smallest average fitness is the worst territory. A migration algebra G is defined during initialization. t ; If the number of iterations exceeds the migration generation G t , then by comparing the average fitness of each territory, the hummingbird corresponding to the worst territory will migrate to a new position randomly generated in the entire search space; at this time, the hummingbird and the nectar collection target positions around it will all be updated; this process effectively prevents the algorithm from falling into a local optimal solution; and when the number of iterations exceeds the contraction and expansion algebra G s , the hummingbird's foraging range will shrink or expand accordingly; when the distance between the k-th nectar collection target quantum position of the i-th hummingbird of the g-th generation and the hummingbird's quantum position is greater than μ, the nectar collection target range shrinks. It should be noted that k≠1 at this time, that is, the k-th nectar collection target position cannot be the current position of the hummingbird. The hummingbird's position is That is, k = 1, and the quantum rotation angle and quantum position update equation are: in, is a random number in the range of [0,1] as the contraction factor; when the distance between the kth nectar collection target quantum position of the i-th hummingbird of the g-th generation and the hummingbird quantum position is less than μ, the nectar collection target range is expanded, and the quantum rotation angle and quantum position update equation are obtained as follows: Here is a random number in [0,ρ] as the expansion factor; Mapping the updated quantum position of a hummingbird's territory after foraging to the corresponding actual position Calculate the fitness of the k-th position of the i-th hummingbird in the g+1th generation as The average fitness of the i-th hummingbird and its surrounding nectar collection target locations, that is, the fitness value of the i-th hummingbird’s territory, is If the fitness of the newly generated quantum position is greater than that of the original quantum position, that is, f i g+1 (k)>f i g (k), the new quantum position replaces the original quantum position and is saved; otherwise, the original quantum position is still saved, that is, let f i g+1 (k) = f i g (k) Compare the quantum positions of the honey-collecting targets with the largest average fitness, which is the global optimal territory Then we can get the optimal quantum position in the optimal territory up to the g+1th generation. Thus, the global optimal quantum position up to the g+1th generation is obtained 8. A computer device / equipment / system comprising a memory, a processor and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program / instruction stored thereon, characterized in that: When the computer program / instructions are executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer program product comprising a computer program / instructions, characterized in that: When the computer program / instructions are executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images