CN-121978617-A - Unmanned aerial vehicle sound source direction of arrival estimation method and system based on multi-frequency fusion

CN121978617ACN 121978617 ACN121978617 ACN 121978617ACN-121978617-A

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, wherein multi-dimensional weights of fusion signal quality, frequency, eigenvalue separation degree, pitch angle and azimuth angle reliability are acquired; and finally, preprocessing the preliminary result by combining with the multidimensional weight, and integrating the frequency estimation result through a fusion strategy to realize the DOA estimation of the sound source. The evaluation method provided by the invention can accurately distinguish the reliability of frequency estimation, fully excavate multi-frequency complementary information, effectively improve the accuracy and stability of DOA estimation in a broadband acoustic scene, and solve the technical problem that the whole accuracy of broadband DOA estimation is limited due to neglecting the quality difference of frequency estimation and single fusion mode in the prior art.

Inventors

GUO MIN
LU YIXIONG
ZHOU SHENGLI

Assignees

西北工业大学

Dates

Publication Date: 20260505
Application Date: 20260114

Claims (10)

1. The unmanned aerial vehicle sound source direction of arrival estimation method based on multi-frequency fusion is characterized by comprising the following steps: Acquiring multi-dimensional weights of fusion signal quality, frequency, eigenvalue separation degree, azimuth crossing consistency and pitch crossing consistency; selecting an adaptive mode order based on a beam space transformation theory, screening out a key frequency point in a target frequency band, and obtaining a direction-of-arrival estimation result based on the key frequency point; Preprocessing the direction of arrival estimation result by combining with the multidimensional weight, obtaining a frequency estimation result, and carrying out direction of arrival estimation fusion on the frequency estimation result to realize sound source direction of arrival estimation.
2. The unmanned aerial vehicle sound source direction of arrival estimation method based on multi-frequency fusion according to claim 1, wherein the multi-dimensional weight acquisition method of fusion signal quality, frequency, eigenvalue separation, azimuth crossing consistency and pitch crossing consistency is as follows: acquiring a signal quality weight, a frequency weight, a characteristic value separation degree weight, an azimuth crossing consistency weight and a pitch crossing consistency weight; Acquiring a multidimensional weight according to the signal quality weight, the frequency weight, the eigenvalue separation degree weight, the azimuth crossing consistency weight and the pitch crossing consistency weight, wherein the multidimensional weight is expressed as follows: Wherein, the As a weight of the signal quality, As the weight of the frequency is given, For the feature value separation degree weight, For the pitch angle to span the consistency weight, For azimuth crossing consistency weights.
3. The method for estimating the direction of arrival of a sound source of an unmanned aerial vehicle based on multi-frequency fusion according to claim 2, wherein the signal quality weight is expressed as: The eigenvalue separation degree weight is expressed as: Wherein, the Is the maximum eigenvalue of the covariance matrix of the target frequency band; The average value of the rest characteristic values; as a result of the normalization factor, In order to be a mode order number, Is the i-th eigenvalue.
4. The method for estimating the direction of arrival of a sound source of an unmanned aerial vehicle based on multi-frequency fusion according to claim 2, wherein the frequency weights are expressed as follows: Wherein, the Is the center frequency; Is the degree of frequency deviation; Is the lower limit of the frequency band, Is the upper limit of the frequency band; Is an attenuation coefficient and is used for controlling the weight attenuation speed; as a parameter of the weight of the center frequency, Is the current frequency.
5. The method for estimating the direction of arrival of a sound source of an unmanned aerial vehicle based on multi-frequency fusion according to claim 1, wherein the azimuth spans a consistency weight Expressed as: Wherein, the Estimating the azimuth angle of the kth frequency point; Is based on the initial weighted fusion azimuth of signal quality weights, frequency weights, eigenvalue separation weights, Is the annular angular distance of the two-dimensional lens, Controlling the sensitivity of the weight function for azimuth scale parameters; Pitch across consistency weights are expressed as: Wherein, the A pitch angle estimated value of the kth frequency point; for initial weighted fusion pitch angle based on previous three-dimensional weights , And (5) adjusting the pitch angle scale parameter according to the SNR.
6. The unmanned aerial vehicle sound source direction of arrival estimation method based on multi-frequency fusion according to claim 1, wherein the adaptive mode order is selected based on a beam space transformation theory, key frequency points are screened out in a target frequency band, and a direction of arrival estimation result is obtained based on the key frequency points, specifically: determining normalized frequency parameters reflecting the matching relationship of the wavelength of the sound wave to the array size ; Wherein, the Is the kth processing frequency and, In order to make the radius of the circular array uniform, Is the speed of sound under standard conditions, Is of a corresponding wavelength; Determination of Mapping rules: Dynamic adjustment The mapping rule selects the order of the self-adaptive mode, and adapts to the signal characteristics of different frequencies; After completing the adaptive mode order selection, the STFT is utilized to convert the time domain signal into the frequency domain, and then the target frequency band is obtained Screening out Critical frequency points ; For each selected frequency Extracting multi-channel data of the frequency from STFT output, calculating covariance matrix And obtaining DOA estimation under the frequency through UCA-ESPRIT algorithm: By aligning The above-mentioned treatments are respectively implemented on the frequencies so as to obtain Group independent direction of arrival estimation results: ; Wherein, the For the order of the optimal mode, Represent the first The frequency of the signal at which the signal is transmitted, Respectively the array radius and the array element number, Is the speed of sound, For the UCA-ESPRIT algorithm, Is the lower limit of the frequency band, Is the upper limit of the frequency band, Is the first The elevation angle at the frequency of the signal, Is the first The azimuth angle at the frequency of the wave, For the bandwidth of the frequency band, , For the sampling frequency to be the same, Is the number of transform points of the FFT.
7. The unmanned aerial vehicle sound source direction of arrival estimation method based on multi-frequency fusion according to claim 1, wherein the preprocessing of the direction of arrival estimation result by combining with multi-dimensional weight is performed to obtain a frequency estimation result, specifically: for each frequency, using a multidimensional weight estimation mechanism Reliability evaluation is carried out on the estimation result of the (a): The set of effective frequency estimates is ; When (when) When in use, weight normalization is carried out ; When (when) The system automatically backs to a single frequency mode and performs direction of arrival estimation by using the frequency with the highest comprehensive weight ; Wherein, the For the weight threshold value, adopting a method based on weight distribution quantile to determine, In the form of a covariance matrix, Is the first The azimuth angle at the frequency of the wave, Is the first The elevation angle at the frequency of the signal, For the multi-dimensional weights to be used, For the index corresponding to the frequency with the highest comprehensive weight, For the frequency with the highest integrated weight, For the weight corresponding to the kth frequency, For the weight corresponding to the jth frequency, Is the normalized weight.
8. The unmanned aerial vehicle sound source direction of arrival estimation method based on multi-frequency fusion according to claim 1, wherein the direction of arrival estimation fusion is performed on the frequency estimation result, specifically: the fusion of the direction of arrival estimation comprises azimuth cyclic weighting fusion and pitch weighting fusion; The azimuthal weighting fusion is as follows: mapping each azimuth to a complex number on a unit circle: Calculating a weighted average Wherein, the , The final azimuth angle is as follows: If it is Then ; The pitch angle weighted fusion is as follows: Wherein, the Is the final azimuth angle and, For the final pitch angle, Is the first The azimuth angle at the frequency of the wave, Is the first The elevation angle at the frequency of the signal, For the azimuth mapping to the x-coordinate on the unit circle, Mapping the azimuth angle to the y coordinate on the unit circle.
9. Unmanned aerial vehicle sound source direction of arrival estimation system based on multifrequency fuses, characterized by comprising: the multi-dimensional weight acquisition module is used for acquiring multi-dimensional weights of fusion signal quality, frequency, eigenvalue separation degree, azimuth crossing consistency and pitch crossing consistency; The direction of arrival estimation module is used for selecting an adaptive mode order based on a beam space transformation theory, screening out key frequency points in a target frequency band and obtaining a direction of arrival estimation result based on the key frequency points; the direction of arrival fusion module is used for preprocessing the direction of arrival estimation result by combining the multidimensional weight, obtaining the frequency estimation result, and carrying out the direction of arrival estimation fusion on the frequency estimation result to realize the direction of arrival 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 the processor, when executing the computer program, realizes the steps of the unmanned aerial vehicle sound source direction of arrival estimation method based on multi-frequency fusion according to any one of claims 1 to 8.

Description

Unmanned aerial vehicle sound source direction of arrival estimation method and system based on multi-frequency fusion Technical Field The invention belongs to the field of acoustic sound signal processing, and particularly relates to an unmanned aerial vehicle sound source direction-of-arrival estimation method and system based on multi-frequency fusion. Background Most sound source signals (such as unmanned plane noise, vehicle noise, human voice and the like) in real environments belong to broadband signals, the energy of the signals is distributed in a wide frequency range, and the narrowband assumption in traditional sound source positioning is difficult to meet. In wideband direction of arrival (direction of arrival, DOA) estimation, the incoherent signal subspace method and the coherent signal subspace method are two classical types of wideband high resolution algorithms. The incoherent signal subspace method carries out simple arithmetic average or geometric average on the arrival direction estimation results of a plurality of frequencies, ignores the difference of estimation quality of different frequencies, is possibly polluted by low-quality estimation results, and cannot effectively distinguish the reliability difference of the estimation results of different frequencies, so that the overall estimation accuracy is limited. Disclosure of Invention The invention aims to solve the problems that a broadband direction-of-arrival incoherent method ignores frequency estimation quality difference and limits overall estimation accuracy in the prior art, and provides an unmanned aerial vehicle sound source direction-of-arrival estimation method and system based on multi-frequency fusion. In order to achieve the purpose, the invention is realized by adopting the following technical scheme: The invention provides an unmanned aerial vehicle sound source direction of arrival estimation method based on multi-frequency fusion, which comprises the following steps: Acquiring multi-dimensional weights of fusion signal quality, frequency, eigenvalue separation degree, azimuth crossing consistency and pitch crossing consistency; selecting an adaptive mode order based on a beam space transformation theory, screening out a key frequency point in a target frequency band, and obtaining a direction-of-arrival estimation result based on the key frequency point; Preprocessing the direction of arrival estimation result by combining with the multidimensional weight, obtaining a frequency estimation result, and carrying out direction of arrival estimation fusion on the frequency estimation result to realize sound source direction of arrival estimation. Preferably, the multi-dimensional weight acquisition method for the fusion signal quality, the frequency, the eigenvalue separation degree, the azimuth crossing consistency and the pitch crossing consistency is as follows: acquiring a signal quality weight, a frequency weight, a characteristic value separation degree weight, an azimuth crossing consistency weight and a pitch crossing consistency weight; Acquiring a multidimensional weight according to the signal quality weight, the frequency weight, the eigenvalue separation degree weight, the azimuth crossing consistency weight and the pitch crossing consistency weight, wherein the multidimensional weight is expressed as follows: Wherein, the As a weight of the signal quality,As the weight of the frequency is given,For the feature value separation degree weight,For the pitch angle to span the consistency weight,For azimuth crossing consistency weights. Preferably, the signal quality weights are expressed as: The eigenvalue separation degree weight is expressed as: Wherein, the Is the maximum eigenvalue of the covariance matrix of the target frequency band; The average value of the rest characteristic values; as a result of the normalization factor, In order to be a mode order number,Is the i-th eigenvalue. Preferably, the frequency weights are expressed as follows: Wherein, the Is the center frequency; Is the degree of frequency deviation; Is the lower limit of the frequency band, Is the upper limit of the frequency band; Is an attenuation coefficient and is used for controlling the weight attenuation speed; as a parameter of the weight of the center frequency, Is the current frequency. Preferably, the azimuth spans a consistency weightExpressed as: Wherein, the Estimating the azimuth angle of the kth frequency point; Is based on the initial weighted fusion azimuth of signal quality weights, frequency weights, eigenvalue separation weights, Is the annular angular distance of the two-dimensional lens,Controlling the sensitivity of the weight function for azimuth scale parameters; Pitch across consistency weights are expressed as: Wherein, the A pitch angle estimated value of the kth frequency point; for initial weighted fusion pitch angle based on previous three-dimensional weights ,And (5) adjusting the pitch angle sca