Unmanned aerial vehicle sound source direction of arrival estimation method and system based on multi-frequency fusion

Abstract

The invention belongs to the field of acoustic sound signal processing, and discloses an unmanned aerial vehicle sound source direction-of-arrival estimation method and system based on multi-frequency fusion, and the method comprises the steps: firstly obtaining the quality, frequency and characteristic value separation degree of a fusion signal, and the multi-dimensional weight of the reliability of a pitch angle and an azimuth angle; a mode order is adaptively selected based on a beam space transformation theory, key frequency points are screened in a target frequency band, and a preliminary DOA estimation result is obtained; and finally, pre-processing the preliminary result in combination with a multi-dimensional weight, and integrating a frequency estimation result through a fusion strategy to realize sound source DOA estimation. According to the evaluation method provided by the invention, the frequency estimation reliability can be accurately distinguished, the multi-frequency complementary information is fully mined, the accuracy and stability of DOA estimation in a broadband acoustic scene are effectively improved, and the technical problem that the overall precision of broadband DOA estimation is limited due to neglect of frequency estimation quality difference and single fusion mode in the prior art is solved.

Description

Technical Field

[0001] This invention belongs to the field of acoustic sound signal processing, specifically relating to a method and system for estimating the direction of arrival of a drone sound source based on multi-frequency fusion. Background Technology

[0002] Most sound source signals in real-world environments (such as drone noise, vehicle noise, and human voices) are broadband signals, with their energy distributed over a wide frequency range, making it difficult to satisfy the narrowband assumption in traditional sound source localization. In broadband direction of arrival (DOA) estimation, incoherent signal subspace methods and coherent signal subspace methods are two classic broadband high-resolution algorithms. Incoherent signal subspace methods perform a simple arithmetic or geometric average of DOA estimation results across multiple frequencies, ignoring the differences in estimation quality at different frequencies. This can lead to contamination by low-quality estimation results and an inability to effectively distinguish the reliability differences between estimation results at different frequencies, resulting in limited overall estimation accuracy. Summary of the Invention

[0003] The purpose of this invention is to solve the problem that the existing broadband direction-of-arrival incoherent methods ignore the differences in frequency estimation quality and limit the overall estimation accuracy, and to provide a method and system for estimating the direction of arrival of UAV sound sources based on multi-frequency fusion.

[0004] To achieve the above objectives, the present invention employs the following technical solution: The present invention proposes a method for estimating the direction of arrival (DOA) of a UAV sound source based on multi-frequency fusion, comprising the following steps: Obtain multi-dimensional weights for fused signal quality, frequency, eigenvalue separation, azimuth cross-span consistency, and elevation cross-span consistency; Based on the beam space transformation theory, the adaptive mode order is selected, key frequency points are screened in the target frequency band, and the direction of arrival estimation results are obtained based on the key frequency points. By combining multi-dimensional weights to preprocess the direction of arrival estimation results, frequency estimation results are obtained. The frequency estimation results are then fused to achieve the direction of arrival estimation of the sound source.

[0005] Preferably, the method for obtaining the multi-dimensional weights of fused signal quality, frequency, eigenvalue separation, azimuth cross-span consistency, and elevation cross-span consistency is as follows: Obtain signal quality weight, frequency weight, eigenvalue separation weight, azimuth cross-consistency weight, and elevation cross-consistency weight; The multi-dimensional weights are obtained based on signal quality weight, frequency weight, eigenvalue separation weight, azimuth cross-consistency weight, and elevation cross-consistency weight, and are expressed as follows:

[0006] in, For signal quality weights, For frequency weights, For eigenvalue separation weights, For pitch angle to cross consistency weights, For azimuth angle consistency weighting.

[0007] Preferably, the signal quality weight is expressed as:

[0008] The eigenvalue separation weights are expressed as follows:

[0009] in, It is the largest eigenvalue of the target frequency band covariance matrix; The mean of the remaining eigenvalues; As the normalization factor, The order of the pattern. Let be the i-th eigenvalue.

[0010] Preferably, the frequency weights are represented as follows:

[0011] in, It is the center frequency; It is the frequency deviation; This is the lower limit of the frequency band. This is the upper limit of the frequency band; This is the attenuation coefficient, used to control the rate at which the weights decay. The center frequency weighting parameter, This is the current frequency.

[0012] Preferably, the azimuth angle spans a consistency weight. Represented as:

[0013] in, Estimate the azimuth angle for the k-th frequency point; It is an initial weighted fused azimuth angle based on signal quality weight, frequency weight, and eigenvalue separation weight. It is the annular angular distance. The azimuth scale parameter controls the sensitivity of the weighting function; The pitch angle cross-consistency weight is represented as follows:

[0014] in, This is the estimated elevation angle value for the k-th frequency point; The initial weighted fused pitch angle based on the first three dimensions. , This is the pitch angle scale parameter, adjusted according to SNR.

[0015] Preferably, the step of selecting the adaptive mode order based on beam space transformation theory, screening key frequency points within the target frequency band, and obtaining the direction-of-arrival estimation result based on the key frequency points specifically involves: Determine the normalized frequency parameters that reflect the matching relationship between the acoustic wavelength and the array size. ;

[0016] in, It is the kth processing frequency. For a uniform circular array radius, The speed of sound under standard conditions, For the corresponding wavelength; Sure Mapping rules:

[0017] Dynamic adjustment The mapping rule selects an adaptive mode order to adapt to the signal characteristics of different frequencies; After completing the adaptive mode order selection, the time-domain signal is converted to the frequency domain using STFT, and then the signal is applied in the target frequency band. Internal screening Key frequency points ; For each selected frequency Extract multi-channel data of this frequency from the STFT output and calculate the covariance matrix. Then, the DOA estimate at this frequency is obtained using the UCA-ESPRIT algorithm:

[0018] Through the Each frequency is processed in the above manner to obtain... Independent direction-of-arrival estimation results: ; in, The optimal mode order, Indicates the first One frequency, These are the array radius and the number of array elements, respectively. For the speed of sound, The algorithm is UCA-ESPRIT. This is the lower limit of the frequency band. This is the upper limit of the frequency band. For the first Elevation angle at a given frequency For the first Azimuth angle at a frequency For frequency band bandwidth, , Sampling frequency, It is the number of transform points in the FFT.

[0019] Preferably, the preprocessing of the direction-of-arrival estimation results by combining multi-dimensional weights to obtain the frequency estimation results specifically involves: Using a multi-dimensional weighted evaluation mechanism, for each frequency The estimation results are used to conduct a reliability assessment:

[0020] The effective frequency estimation set is ; when At that time, weight normalization is performed. ; when The system automatically reverts to single-frequency mode, using the frequency with the highest overall weight for direction-of-arrival estimation. ; in, The weight threshold is determined using a method based on the quantiles of the weight distribution. Let covariance matrix be the variance matrix. For the first Azimuth angle at a frequency For the first Elevation angle at a given frequency For multi-dimensional weights, The index corresponding to the frequency with the highest overall weight. The frequency with the highest overall weight. The weight corresponding to the k-th frequency. The weight corresponding to the j-th frequency, These are the normalized weights.

[0021] Preferably, the process of fusing the frequency estimation results with direction-of-arrival estimation specifically involves: The fusion of direction-of-arrival (DOA) estimates includes azimuth cyclic weighted fusion and elevation weighted fusion. The azimuth weighted fusion is as follows: Map each azimuth angle to a complex number on the unit circle:

[0022] Calculate the weighted average

[0023] in, ,

[0024] The final azimuth angle is as follows:

[0025] like ,but ; Pitch angle weighted fusion is as follows:

[0026] in, It is the final azimuth. For the final pitch angle, For the first Azimuth angle at a frequency For the first Elevation angle at a given frequency The azimuth angle is mapped to the x-coordinate on the unit circle. The y-coordinate is the azimuth angle mapped onto the unit circle.

[0027] This invention proposes a UAV sound source direction-of-arrival estimation system based on multi-frequency fusion, comprising: A multi-dimensional weight acquisition module is used to acquire multi-dimensional weights of fused signal quality, frequency, eigenvalue separation, azimuth cross-continuity, and elevation cross-continuity. The direction of arrival estimation module is used to select the adaptive mode order based on beam space transformation theory, screen out key frequency points in the target frequency band, and obtain the direction of arrival estimation result based on the key frequency points. The direction-of-arrival (DOA) fusion module is used to preprocess the DOA estimation results by combining multi-dimensional weights, obtain the frequency estimation results, and perform DOA estimation fusion on the frequency estimation results to realize the DOA estimation of the sound source.

[0028] A terminal device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of a method for estimating the direction of arrival of a UAV sound source based on multi-frequency fusion.

[0029] Compared with the prior art, the present invention has the following beneficial effects: This invention proposes a method for estimating the direction of arrival (DOA) of a UAV sound source based on multi-frequency fusion. By constructing a multi-dimensional weighted evaluation system, it comprehensively integrates key indicators such as signal quality, frequency characteristics, eigenvalue separation, and consistency across azimuth and pitch angles. This fundamentally solves the core defect of existing incoherent signal subspace methods that fail to distinguish the reliability of estimation results from different frequencies, providing a quantitative basis for accurately selecting and utilizing effective frequency information. Based on this, relying on beamspace transformation theory, it adaptively selects the mode order to accurately select key frequency points within the target frequency band, ensuring the effectiveness and accuracy of the preliminary DOA estimation results. Subsequently, the preliminary estimation results are preprocessed using the aforementioned multi-dimensional weights to enhance the weight of high-quality frequency estimation results. Then, a fusion strategy is used to integrate the preprocessed frequency estimation results, replacing the traditional coarse fusion method of simple arithmetic or geometric averaging. This fully explores and utilizes the complementary information between multiple frequencies, ultimately achieving high-precision and high-stability DOA estimation. This solves the technical problem of limited overall accuracy of broadband DOA estimation caused by neglecting differences in frequency estimation quality and the single fusion method in existing technologies. Therefore, this invention can comprehensively utilize complementary information from multiple frequencies to improve the accuracy and stability of DOA estimation in broadband acoustic scenarios.

[0030] This invention proposes a UAV sound source direction-of-arrival estimation system based on multi-frequency fusion. The system is divided into a multi-dimensional weight acquisition module, a direction-of-arrival estimation module, and a direction-of-arrival fusion module to obtain the fusion result and achieve sound source direction-of-arrival estimation. The modular approach ensures that each module is independent, facilitating unified management of all modules. Attached Figure Description

[0031] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0032] Figure 1 This is a flowchart of the UAV sound source direction-of-arrival estimation method based on multi-frequency fusion according to the present invention.

[0033] Figure 2 This is a diagram of the uniform circular array model of the present invention.

[0034] Figure 3 This is a detailed flowchart of the UAV sound source direction-of-arrival estimation method based on multi-frequency fusion according to the present invention.

[0035] Figure 4This is a diagram of the UAV sound source direction-of-arrival estimation system based on multi-frequency fusion according to the present invention.

[0036] Figure 5 This is a schematic diagram of the structure of an electronic device according to the present invention. Detailed Implementation

[0037] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0038] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0039] The present invention will now be described in further detail with reference to the accompanying drawings: This invention proposes a method for estimating the direction of arrival (DOA) of a UAV sound source based on multi-frequency fusion, such as... Figure 1 As shown, it includes the following steps: S1. Obtain multi-dimensional weights for fused signal quality, frequency, eigenvalue separation, azimuth cross-conformity, and elevation cross-conformity. The method for obtaining the multi-dimensional weights of fused signal quality, frequency, eigenvalue separation, azimuth cross-sectional consistency, and elevation cross-sectional consistency is as follows: Obtain signal quality weight, frequency weight, eigenvalue separation weight, azimuth cross-consistency weight, and elevation cross-consistency weight; The multi-dimensional weights are obtained based on signal quality weight, frequency weight, eigenvalue separation weight, pitch angle cross-consistency weight, and pitch angle cross-consistency weight, and are expressed as follows:

[0040] in, For signal quality weights, For frequency weights, For eigenvalue separation weights, For pitch angle to cross consistency weights, For azimuth angle consistency weighting.

[0041] The signal quality weight is represented as follows:

[0042] The eigenvalue separation weights are expressed as follows:

[0043] in, It is the largest eigenvalue of the target frequency band covariance matrix; The mean of the remaining eigenvalues; As the normalization factor, The order of the pattern. Let be the i-th eigenvalue.

[0044] The frequency weights are represented as follows:

[0045] in, It is the center frequency; It is the frequency deviation; This is the lower limit of the frequency band. This is the upper limit of the frequency band; This is the attenuation coefficient, used to control the rate at which the weights decay. The center frequency weighting parameter, This is the current frequency.

[0046] The azimuth angle crosses the consistency weight Represented as:

[0047] in, Estimate the azimuth angle for the k-th frequency point; It is an initial weighted fused azimuth angle based on signal quality weight, frequency weight, and eigenvalue separation weight. It is the annular angular distance. The azimuth scale parameter controls the sensitivity of the weighting function. The value can be adjusted according to SNR; the larger the SNR, the better. The smaller the value (for azimuth angles, this can be 1°~10°), the better. Initial fusion azimuth angle. The calculation method is as follows:

[0048] in The normalized three-dimensional joint weights.

[0049] The pitch angle cross-consistency weight is represented as follows:

[0050] in, This is the estimated elevation angle value for the k-th frequency point; The initial weighted fused pitch angle based on the first three dimensions. , This is the pitch angle scale parameter, adjusted according to SNR.

[0051] S2. Based on the beam space transformation theory, select the adaptive mode order, screen out key frequency points in the target frequency band, and obtain the direction of arrival estimation results based on the key frequency points. The adaptive mode order is selected based on beam space transformation theory, key frequency points are screened within the target frequency band, and the direction of arrival estimation result is obtained based on the key frequency points. Specifically: Sure Mapping rules:

[0052] Among them, the normalized frequency parameter It reflects the matching relationship between the sound wave wavelength and the array size, specifically:

[0053] Dynamic adjustment The mapping rule selects an adaptive mode order to adapt to the signal characteristics of different frequencies; After completing the adaptive mode order selection, the time-domain signal is converted to the frequency domain using STFT, and then the signal is applied in the target frequency band. Internal screening Key frequency points ; For each selected frequency Extract multi-channel data of this frequency from the STFT output and calculate the covariance matrix. Then, the DOA estimate at this frequency is obtained using the UCA-ESPRIT algorithm:

[0054] Through the Each frequency is processed in the above manner to obtain... Independent direction-of-arrival estimation results: ; in, For normalized frequency parameters, The optimal mode order, Indicates the first One frequency, These are the array radius and the number of array elements, respectively. For the speed of sound, The algorithm is UCA-ESPRIT. This is the lower limit of the frequency band. This is the upper limit of the frequency band. For the first Elevation angle at a given frequency For the first Azimuth angle at a frequency For frequency band bandwidth, , Sampling frequency, It is the number of transform points in the FFT.

[0055] S3. Preprocess the direction of arrival estimation results by combining multi-dimensional weights to obtain frequency estimation results, and fuse the frequency estimation results to realize the direction of arrival estimation of the sound source.

[0056] The preprocessing of the direction-of-arrival estimation results by combining multi-dimensional weights to obtain the frequency estimation results is specifically as follows: Using a multi-dimensional weighted evaluation mechanism, for each frequency The estimation results are used to conduct a reliability assessment:

[0057] The effective frequency estimation set is ; when At that time, weight normalization is performed. ; when The system automatically reverts to single-frequency mode, using the frequency with the highest overall weight for direction-of-arrival estimation. ; in, For weighted thresholds, Let covariance matrix be the variance matrix. For the first Azimuth angle at a frequency For the first Elevation angle at a given frequency For multi-dimensional weights. The index corresponding to the frequency with the highest overall weight. The frequency with the highest overall weight. The weight corresponding to the k-th frequency. The weight corresponding to the j-th frequency, These are the normalized weights. Considering the weight distribution characteristics and adaptability requirements under different signal-to-noise ratio conditions, a method based on the weight distribution quantiles is adopted to determine the weights. The specific strategy is as follows: Take the 25th percentile (i.e., the lower quartile) of the weighted distribution.

[0058] The process of fusing the frequency estimation results with direction-of-arrival estimation specifically involves: The fusion of direction-of-arrival (DOA) estimates includes azimuth cyclic weighted fusion and elevation weighted fusion. The azimuth weighted fusion is as follows: Map each azimuth angle to a complex number on the unit circle:

[0059] Calculate the weighted average

[0060] in, ,

[0061] The final azimuth angle is as follows:

[0062] like ,but ; Pitch angle weighted fusion is as follows:

[0063] in, For the final pitch angle, For the first Azimuth angle at a frequency For the first Elevation angle at a given frequency The azimuth angle is mapped to the x-coordinate on the unit circle. The y-coordinate is the azimuth angle mapped onto the unit circle.

[0064] The following is combined Figure 2 and Figure 3 The method is described in detail below: A method for estimating the direction of arrival (DOA) of a UAV sound source based on multi-frequency fusion is proposed, specifically targeting a uniform circular array and its signal model. The method comprises three parts: multi-dimensional weight calculation, a multi-frequency fusion algorithm, and weighted fusion of DOA estimation, achieving more accurate and robust estimation of the sound source DOA. Figure 3 First, the received signal undergoes STFT transformation, key frequency points are selected within the target frequency band, and the covariance matrix is ​​calculated; then, based on the normalized frequency parameters... Dynamically determine the optimal mode order for each frequency The system obtains single-frequency DOA estimates using the UCA-ESPRIT algorithm; then, it calculates the reliability weights for each frequency using a five-dimensional weight evaluation mechanism, filters the effective estimate set, and performs weight normalization; finally, it achieves azimuth angle cyclic weighted fusion through unit circle mapping, completes pitch angle fusion through standard weighted averaging, and outputs the final DOA estimate. The entire process, with its multi-level anomaly handling and rollback mechanisms, ensures system stability under various conditions.

[0065] First, the geometric and signal model of the circular array is as follows: Consider a uniform circular array consisting of N omnidirectional microphones, with a radius of R, such as... Figure 2 As shown. Elevation angle of the signal source. It is the line connecting the origin to the source and Angle between axes, azimuth It is the line connecting the origin to the source. Projection of a plane and The angle between axes.

[0066] For those located Given a broadband signal source, considering the far-field assumption, the signal received by the array at time t... It can be represented as:

[0067] in, It is the source signal; The array direction matrix is ​​composed of array guide vectors; This is a noise signal. (Received signal) After Short-Time Fourier Transform (STFT), it can be expressed as:

[0068] In DOA estimation, the covariance matrix is ​​used to describe the correlation between received signals. Assuming the signals are received via an array, the covariance matrix... This represents the correlation of a signal between different sensor array elements. Its mathematical representation in the frequency domain is:

[0069] Where f represents frequency information. Assume the target frequency band is... ,in This is the lower limit of the frequency band. If the upper limit of the frequency band is , then the frequency band bandwidth is The center frequency is .

[0070] Second, multi-dimensional weight calculation: The multi-dimensional weight calculation aims to comprehensively evaluate the reliability of the direction-of-arrival (DOA) estimation results at each frequency, and to achieve a fusion strategy of "high-quality estimates contributing more and low-quality estimates having less impact" through dynamic weight allocation. Specifically, it includes five parts: signal quality weight, frequency weight, eigenvalue separation weight, elevation reliability weight, and azimuth weight. The calculation method is as follows: Signal quality weight is defined as:

[0071] in, It is the largest eigenvalue of the target frequency band covariance matrix; The mean of the remaining eigenvalues; This is the normalization factor.

[0072] Frequency weight is defined as:

[0073] in, It is the center frequency; It is the frequency deviation (distance from the center frequency); let This refers to the relative frequency deviation. This is the attenuation coefficient, used to control the rate at which the weights decay. The center frequency weighting parameter directly determines the center frequency point. The initial weights of ) simultaneously satisfy , This represents the maximum relative deviation of frequency points within the target frequency band. Ensure... The function input is always positive, causing the weights to vary. Increasing the monotonically decreasing weight aligns with the physical law that "the further away from the center, the lower the weight."

[0074] The eigenvalue separation weight is defined as follows:

[0075] in, The largest eigenvalue of the target frequency band covariance matrix. The mean of the noise eigenvalues ​​is given. The eigenvalue separation weight reflects the degree of separation between the signal subspace and the noise subspace. The higher the separation, the larger the weight, indicating that the direction of arrival estimation at that frequency is more reliable.

[0076] The azimuth cross-consistency weight is defined as follows:

[0077] in, Estimate the azimuth angle for the k-th frequency point; Based on the first three dimensions weight ( The initial weighted fused azimuth angle, It is the annular angular distance. The azimuth scale parameter controls the sensitivity of the weighting function. The value can be adjusted according to SNR; the larger the SNR, the better. The smaller the value (for azimuth angles, this can be 1°~10°), the better. Initial fusion azimuth angle. The calculation method is as follows:

[0078] in, The normalized three-dimensional joint weights.

[0079] The pitch angle cross-consistency weight is defined as follows:

[0080] in, This is the estimated elevation angle value for the k-th frequency point; The initial weighted fused pitch angle based on the first three dimensions. , This is the pitch angle scale parameter, adjusted according to SNR.

[0081] The overall weight design is as follows:

[0082] The five coefficients in the weighting formula can be determined through experimental optimization.

[0083] Third, the multi-frequency fusion estimation algorithm: This invention uses a uniform circular array (UCA) as the array geometry model and utilizes the Estimation of Signal Parameters via Rotational Invariance Techniques (ESPRIT) to estimate the direction of arrival (DOA) in the same frequency band.

[0084] 1) Dynamic mode order optimization To ensure optimal performance of the UCA-ESPRIT algorithm across all frequencies, this invention expands the array steering vector using a Bessel function based on beamspace transformation theory. Since the m-th order Bessel function... exist Rapid decay occurs, and this invention combines the mode order M with the normalized frequency parameter. Matching, will Defined as

[0085] in, It is the current processing frequency (Hz); The radius of the uniform circular array; The speed of sound under standard conditions. When When the wavelength is small (low frequency or small array), the wavelength is much larger than the array size, the phase difference between array elements is small, and fewer phase modes are required; while when At larger wavelengths (high frequency or large arrays), the wavelength is close to or smaller than the array size, resulting in rich phase variations between array elements, requiring more phase modes for accurate characterization. This invention proposes the following... Mapping rules:

[0086] This dynamic selection rule can effectively avoid the performance degradation caused by a fixed mode order, especially in the case of wideband signals, by dynamically adjusting... The value can better adapt to the characteristics of signals at different frequencies.

[0087] 2) Single-frequency direction of arrival estimation After completing the adaptive mode order selection, the system performs independent DOA estimation for each frequency point. First, the time-domain signal is converted to the frequency domain using STFT, and then the DOA is estimated in the target frequency band. Internal screening The calculation method for the key frequency points is as follows:

[0088] in, , Sampling frequency, This represents the number of transform points in the FFT. For each selected frequency... Perform independent direction-of-arrival estimation using the following steps: Extract multi-channel data of this frequency from the STFT output and calculate the covariance matrix. ; Based on normalized frequency parameters Dynamically determine the optimal mode order ; The DOA estimate at this frequency was obtained using the UCA-ESPRIT algorithm:

[0089] in, Indicates the first One frequency; It is the signal covariance matrix; These are the array radius and the number of array elements, respectively. Speed ​​of sound; This refers to the UCA-ESPRIT algorithm. Through analysis of... Each frequency is processed in the above manner to obtain... Independent direction-of-arrival estimation results:

[0090] 3) Validity screening and weight normalization After completing the single-frequency direction-of-arrival estimation, the established multi-dimensional weighted evaluation mechanism is used to evaluate each frequency. The estimation results are used to conduct a reliability assessment:

[0091] The effective estimate set is defined as:

[0092] in, The weight threshold is determined using a method based on the quantiles of the weight distribution. The specific strategy is as follows: Take the 25th percentile (i.e., the lower quartile) of the weighted distribution. when At that time, perform weight normalization:

[0093] That is, at least three valid frequency estimates are required before proceeding with the fusion of direction-of-arrival (DOA) estimates. When the number of valid frequencies is insufficient ( The system automatically reverts to single-frequency mode, using the frequency with the highest overall weight for direction-of-arrival estimation.

[0094] Using the frequency covariance matrix and the corresponding mode order Using UCA-ESPRIT for estimation ensures that the system can still output reliable direction of arrival results under extreme conditions.

[0095] Fourth, weighted fusion of direction-of-arrival estimation: The direction-of-arrival estimation weighted fusion includes two parts: azimuth cyclic weighted fusion and elevation weighted fusion.

[0096] (a) Azimuth weighted fusion Because the azimuth angle has a cyclic characteristic ( and (Equivalent), the direct arithmetic mean will cross An error occurs at the boundary. The system solves this problem by using a unit circle mapping method, mapping each azimuth angle to a complex number on the unit circle:

[0097] Calculate the weighted average:

[0098] in:

[0099]

[0100] The final azimuth angle is obtained using the arctangent function:

[0101] like ,but

[0102] The unit circle average method is a standard statistical approach for handling annular data, ensuring mathematical consistency when crossing boundaries. (b) Pitch angle weighted fusion Pitch angles are calculated using a standard weighted average.

[0103] Therefore, in the field of source direction of arrival (DOA) estimation, the method proposed in this invention focuses on the use of uniform circular arrays (UCA) due to their unique geometric symmetry and omnidirectional coverage. This invention aims to propose a multi-frequency fusion-based DOA estimation method. Its core innovation lies in constructing a multi-dimensional estimation quality assessment and intelligent fusion mechanism, with the goal of improving the accuracy and robustness of DOA estimation in complex acoustic environments.

[0104] Example 2 This invention proposes a UAV sound source direction-of-arrival estimation system based on multi-frequency fusion, such as... Figure 4 As shown, it includes: A multi-dimensional weight acquisition module is used to acquire multi-dimensional weights of fused signal quality, frequency, eigenvalue separation, azimuth cross-continuity, and elevation cross-continuity. The direction of arrival estimation module is used to select the adaptive mode order based on beam space transformation theory, screen out key frequency points in the target frequency band, and obtain the direction of arrival estimation result based on the key frequency points. The direction-of-arrival (DOA) fusion module is used to preprocess the DOA estimation results by combining multi-dimensional weights, obtain the frequency estimation results, and perform DOA estimation fusion on the frequency estimation results to realize the DOA estimation of the sound source.

[0105] Example 3 Please see Figure 5 As shown, the present invention also provides an electronic device 100 for a method for estimating the direction of arrival of a UAV sound source based on multi-frequency fusion; the electronic device 100 includes a memory 101, at least one processor 102, a computer program 103 stored in the memory 101 and executable on the at least one processor 102, and at least one communication bus 104.

[0106] The memory 101 can be used to store the computer program 103. The processor 102 implements the steps of the UAV sound source direction-of-arrival estimation method based on multi-frequency fusion described in Embodiment 1 by running or executing the computer program stored in the memory 101 and calling the data stored in the memory 101. The memory 101 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the electronic device 100 (such as audio data), etc. In addition, the memory 101 may include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other non-volatile solid-state storage device.

[0107] The at least one processor 102 may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The processor 102 may be a microprocessor or any conventional processor. The processor 102 is the control center of the electronic device 100, connecting various parts of the electronic device 100 via various interfaces and lines.

[0108] The memory 101 in the electronic device 100 stores multiple instructions to implement a method for estimating the direction of arrival (DOA) of a UAV sound source based on multi-frequency fusion, and the processor 102 can execute the multiple instructions to achieve the following: Obtain multi-dimensional weights for fused signal quality, frequency, eigenvalue separation, elevation reliability, and azimuth; Based on the beam space transformation theory, the adaptive mode order is selected, key frequency points are screened in the target frequency band, and the direction of arrival estimation results are obtained based on the key frequency points. By combining multi-dimensional weights to preprocess the direction of arrival estimation results, frequency estimation results are obtained. The frequency estimation results are then fused to achieve the direction of arrival estimation of the sound source.

[0109] Example 4 If the modules / units integrated in the electronic device 100 are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, and a read-only memory (ROM).

[0110] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0111] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0112] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0113] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0114] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for estimating the direction of arrival (DOA) of a UAV sound source based on multi-frequency fusion, characterized in that, Includes the following steps: Obtain multi-dimensional weights for fused signal quality, frequency, eigenvalue separation, azimuth cross-span consistency, and elevation cross-span consistency; Based on the beam space transformation theory, the adaptive mode order is selected, key frequency points are screened in the target frequency band, and the direction of arrival estimation results are obtained based on the key frequency points. By combining multi-dimensional weights to preprocess the direction of arrival estimation results, frequency estimation results are obtained. The frequency estimation results are then fused to achieve the direction of arrival estimation of the sound source.

2. The method for estimating the direction of arrival of a UAV sound source based on multi-frequency fusion according to claim 1, characterized in that, The method for obtaining the multi-dimensional weights of fused signal quality, frequency, eigenvalue separation, azimuth cross-sectional consistency, and elevation cross-sectional consistency is as follows: Obtain signal quality weight, frequency weight, eigenvalue separation weight, azimuth cross-consistency weight, and elevation cross-consistency weight; The multi-dimensional weights are obtained based on signal quality weight, frequency weight, eigenvalue separation weight, azimuth cross-consistency weight, and elevation cross-consistency weight, and are expressed as follows: in, For signal quality weights, For frequency weights, For eigenvalue separation weights, For pitch angle to cross consistency weights, For azimuth angle consistency weighting.

3. The method for estimating the direction of arrival of a UAV sound source based on multi-frequency fusion according to claim 2, characterized in that, The signal quality weight is represented as follows: The eigenvalue separation weights are expressed as follows: in, It is the largest eigenvalue of the target frequency band covariance matrix; The mean of the remaining eigenvalues; As the normalization factor, The order of the pattern. Let be the i-th eigenvalue.

4. The method for estimating the direction of arrival of a UAV sound source based on multi-frequency fusion according to claim 2, characterized in that, The frequency weights are represented as follows: in, It is the center frequency; It is the frequency deviation; This is the lower limit of the frequency band. This is the upper limit of the frequency band; This is the attenuation coefficient, used to control the rate at which the weights decay. The center frequency weighting parameter, This is the current frequency.

5. The method for estimating the direction of arrival of a UAV sound source based on multi-frequency fusion according to claim 1, characterized in that, The azimuth angle crosses the consistency weight Represented as: in, Estimate the azimuth angle for the k-th frequency point; It is an initial weighted fused azimuth angle based on signal quality weight, frequency weight, and eigenvalue separation weight. It is the annular angular distance. The azimuth scale parameter controls the sensitivity of the weighting function; The pitch angle cross-consistency weight is represented as follows: in, This is the estimated elevation angle value for the k-th frequency point; The initial weighted fused pitch angle based on the first three dimensions. , This is the pitch angle scale parameter, adjusted according to SNR.

6. The method for estimating the direction of arrival of a UAV sound source based on multi-frequency fusion according to claim 1, characterized in that, The adaptive mode order is selected based on beam space transformation theory, key frequency points are screened within the target frequency band, and the direction of arrival estimation result is obtained based on the key frequency points. Specifically: Determine the normalized frequency parameters that reflect the matching relationship between the acoustic wavelength and the array size. ; in, It is the kth processing frequency. For a uniform circular array radius, The speed of sound under standard conditions. For the corresponding wavelength; Sure Mapping rules: Dynamic adjustment The mapping rule selects an adaptive mode order to adapt to the signal characteristics of different frequencies; After completing the adaptive mode order selection, the time-domain signal is converted to the frequency domain using STFT, and then the signal is applied in the target frequency band. Internal screening Key frequency points ; For each selected frequency Extract multi-channel data of this frequency from the STFT output and calculate the covariance matrix. Then, the DOA estimate at this frequency is obtained using the UCA-ESPRIT algorithm: Through the Each frequency is processed in the above manner to obtain... Independent direction-of-arrival estimation results: ; in, The optimal mode order, Indicates the first One frequency, These are the array radius and the number of array elements, respectively. For the speed of sound, The algorithm is UCA-ESPRIT. This is the lower limit of the frequency band. This is the upper limit of the frequency band. For the first Elevation angle at a given frequency For the first Azimuth angle at a frequency For frequency band bandwidth, , Sampling frequency, It is the number of transform points in the FFT.

7. The method for estimating the direction of arrival (DOA) of a UAV sound source based on multi-frequency fusion according to claim 1, characterized in that, The preprocessing of the direction-of-arrival estimation results by combining multi-dimensional weights to obtain the frequency estimation results is specifically as follows: Using a multi-dimensional weighted evaluation mechanism, for each frequency The estimation results are used to conduct a reliability assessment: The effective frequency estimation set is ; when At that time, weight normalization is performed. ; when The system automatically reverts to single-frequency mode, using the frequency with the highest overall weight for direction-of-arrival estimation. ; in, The weight threshold is determined using a method based on the quantiles of the weight distribution. Let covariance matrix be the variance matrix. For the first Azimuth angle at a frequency For the first Elevation angle at a given frequency For multi-dimensional weights, The index corresponding to the frequency with the highest overall weight. The frequency with the highest overall weight. The weight corresponding to the k-th frequency. The weight corresponding to the j-th frequency, These are the normalized weights.

8. The method for estimating the direction of arrival of a UAV sound source based on multi-frequency fusion according to claim 1, characterized in that, The process of fusing the frequency estimation results with direction-of-arrival estimation specifically involves: The fusion of direction-of-arrival (DOA) estimates includes azimuth cyclic weighted fusion and elevation weighted fusion. The azimuth weighted fusion is as follows: Map each azimuth angle to a complex number on the unit circle: Calculate the weighted average in, , The final azimuth angle is as follows: like ,but ; Pitch angle weighted fusion is as follows: in, It is the final azimuth. For the final pitch angle, For the first Azimuth angle at a frequency For the first Elevation angle at a given frequency The azimuth angle is mapped to the x-coordinate on the unit circle. The y-coordinate is the azimuth angle mapped onto the unit circle.

9. A direction-of-arrival estimation system for UAV sound sources based on multi-frequency fusion, characterized in that, include: A multi-dimensional weight acquisition module is used to acquire multi-dimensional weights of fused signal quality, frequency, eigenvalue separation, azimuth cross-continuity, and elevation cross-continuity. The direction of arrival estimation module is used to select the adaptive mode order based on beam space transformation theory, screen out key frequency points in the target frequency band, and obtain the direction of arrival estimation result based on the key frequency points. The direction-of-arrival (DOA) fusion module is used to preprocess the DOA estimation results by combining multi-dimensional weights, obtain the frequency estimation results, and perform DOA estimation fusion on the frequency estimation results to realize the DOA estimation of the sound source.

10. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the UAV sound source direction-of-arrival estimation method based on multi-frequency fusion as described in any one of claims 1 to 8.

Images