CN-116413708-B - DEMON spectrum analysis axial leaf frequency extraction method

CN116413708BCN 116413708 BCN116413708 BCN 116413708BCN-116413708-B

Abstract

The invention discloses a DEMON spectrum analysis axial leaf frequency extraction method, belongs to the field of underwater sound target identification, and can be used for passive sonar DEMON spectrum analysis, extracting target axial leaf frequency information and improving target identification capacity. The method comprises the steps of carrying out spatial filtering on received data through beam synthesis in a preprocessing stage to improve signal to noise ratio, carrying out background equalization on a DEMON spectrum through a sequencing cut-off average method to extract a line spectrum more accurately, obtaining more accurate weight of a line spectrum difference frequency through an improved greatest common divisor method to accurately extract shaft frequency and leaf frequency information, wherein the weight of the line spectrum is weighted under the condition that the number of line spectrum at an unstable moment is insufficient, and improving the data utilization rate. The method effectively improves the accuracy of the axial frequency leaf frequency extraction of the DEMON spectrum analysis.

Inventors

LIANG HAOQIAN
LI JIN
WANG XIAOQING

Assignees

中国电子科技集团公司第五十四研究所

Dates

Publication Date: 20260508
Application Date: 20230217

Claims (6)

1. The DEMON spectrum analysis axial leaf frequency extraction method is characterized by comprising the following steps of: (1) Performing frequency domain beam synthesis on the array received data to obtain a plurality of groups of demodulation DEMON spectrum data in a target direction; (2) Carrying out background equalization processing on each group of demodulation DEMON spectrum data by a sequencing cut-off average method, extracting line spectrum, and obtaining a line spectrum frequency point set which is marked as P= { f 1 ,f 2 ,...,f K }, by each group of data; (3) Sequencing the extracted line spectrums according to intensity, taking I line spectrums with highest intensity, sequencing from small to large according to frequency, and marking the obtained I line spectrum frequency point set as Q= { f 1 ,f 2 ,...,f I }, wherein I is a set value; (4) The frequency points in the line spectrum frequency point set Q are subjected to difference frequency to obtain a difference frequency point set, and the weight corresponding to each difference frequency point is calculated; (5) Updating the difference frequency weight, taking the difference frequency corresponding to the maximum weight as the axis frequency of the data, obtaining the axis frequency value of each group of data, and putting the axis frequency value into an axis frequency set; (6) Taking the axial frequency with the largest occurrence number in the axial frequency set as an axial frequency extraction result f Z ; (7) And extracting the leaf frequency according to the axial frequency extraction result to obtain a leaf frequency extraction result f Y .
2. The method for extracting axial leaf frequency from DEMON spectrum analysis according to claim 1, wherein the specific steps of step (2) include: (201) For each set of demodulated DEMON spectrum data X (N) = { X (1), X (2),. The term, X (N) }, each element X (N) in X (N), n=1, 2, &., N, N is the number of demodulated DEMON spectrum data in each set, and the data are expanded by a window length M to obtain 2m+1 data: x(n-M),...,x(n-1),x(n),x(n+1),...,x(n+M) The data are arranged from small to large as: y(1),y(2),...,y(2M+1) the median of the sequence is y (m+1), then the truncated average of the sequence is: truncated mean value correction is carried out on the sequence, and Is removed to obtain: wherein, the rejection threshold parameter alpha is regulated to be a constant, and M is a set value; each element X (N) in X (N), n=1, 2, after N processing is completed, a new set of data Z (N) after background equalization is obtained; (202) An axis frequency leaf frequency detection extraction interval [ f min ,f max ] is set, line spectrums in the detection interval are extracted from Z (N), the frequency interval of the extracted line spectrums is greater than or equal to f min when the line spectrums are extracted, the frequency resolution of Z (N) is f re ＝f s /N d , the line spectrums are extracted from Z (N) at a line spectrum point extraction interval of N span ＝floor(f min /f re ), a line spectrum frequency set P= { f 1 ,f 2 ,...,f K }, wherein N d is the number of points for DFT when X (N) performs frequency domain beam synthesis, f s is the sampling rate, and floor is a downward rounding.
3. The method for extracting axial leaf frequency of DEMON spectrum analysis according to claim 1, wherein the step (3) specifically includes: For each group of extracted line spectrum sets P, the line spectrums are ordered according to intensity, under the condition of stable signal and receiving environment, I line spectrums with highest intensity are selected, the obtained I line spectrum frequency sets are ordered from small to large according to frequency, the line spectrum intensities corresponding to the frequencies in the frequency sets are S= { S 1 ,s 2 ,...,s I }, and the normalized weight corresponding to each frequency f i in the line spectrum frequency sets is w i ＝s i /max (S); Under the condition of unstable signal and receiving environment, only I s line spectrums can be extracted, and at the moment, the weight corresponding to the line spectrum data at the moment is proportionally adjusted to w i ＝s i I/max(S)I s , wherein I s < I.
4. A method for extracting axial leaf frequency of DEMON spectrum analysis according to claim 3, wherein the specific steps of step (4) include: (401) The frequencies in each line spectrum set Q are subjected to difference frequency in sequence: F ij ＝|f i -f j |,i=1,2,...,I-1;j=2,3,...,I;i<j W ij ＝w i +w j ,i=1,2,...,I-1;j=2,3,...,I;i<j Wherein F ij is the obtained difference frequency, and W ij is the weight corresponding to the difference frequency; (402) For the obtained difference frequency point set F, unifying elements with the frequency value difference value smaller than a set value into a frequency value; If the difference frequency weights are set as the difference frequencies F a and F b of W a and W b and |F a -F b | < epsilon, the difference frequencies F a and F b are unified as F ab ,F ab , and the calculation method is as follows: After unification, only one element with the same frequency value in the difference frequency set F is reserved and the corresponding weights are added, wherein the difference frequency unification parameter epsilon is adjusted to be a constant.
5. The method for extracting axial leaf frequency from DEMON spectrum analysis according to claim 4, wherein the step (5) specifically includes: (501) For each set F of difference frequencies, the line frequency difference frequency ratio β=f i /F ij is used to update the difference frequency weight value if β satisfies: then the weight W ij corresponding to the difference frequency F ij is added with 1, wherein round is rounded; (502) After the weights corresponding to each group of the difference frequencies are updated, the difference frequency corresponding to the maximum weight in each group of the weights is used as the axis frequency of the group of data, and the axis frequency is put into an axis frequency set.
6. The method for extracting axial leaf frequency of DEMON spectrum analysis according to claim 1, wherein the step (7) specifically includes: The leaf frequency is equal to the axial frequency multiplied by the number of the paddles in value, and the frequency with the maximum line spectrum intensity in the frequency f i ∈[3f Z ,7f Z is taken as a leaf frequency result f Y for the obtained axial frequency result f Z .

Description

DEMON spectrum analysis axial leaf frequency extraction method Technical Field The invention belongs to the field of underwater sound target identification, and particularly relates to a DEMON spectrum analysis axial leaf frequency extraction method which can be used for passive sonar noise demodulation (Detection ofEnvelope Modulation on Noise, DEMON) spectrum analysis, and can be used for accurately extracting target axial leaf frequency information and improving target identification capability. Background The marine vessel target identification has important significance in the fields of marine security, marine resource environment monitoring and the like. At present, the ship target identification mainly depends on the characteristic that the ship radiates noise, and the radiated noise mainly comprises three parts of propeller noise, internal mechanical noise and hydrodynamic noise. Propeller noise has a higher energy level and a variety of stable frequency components than mechanical noise and hydrodynamic noise. The propeller noise consists of a discrete line spectrum and a high-frequency continuous spectrum, the rotation of the propeller enables the axial frequency and the blade frequency of the propeller to modulate cavitation noise, the axial frequency is equal to the rotation frequency of the propeller, the blade frequency is equal to the product of the axial frequency and the blade number in value, and the important characteristic information of various ship identifications such as the rotating speed of the propeller, the blade number of the propeller of a target and the like can be obtained through the axial frequency blade frequency information. But the axial frequency leaf frequency information of the target cannot be directly obtained from the target power spectrum, the continuous spectrum needs to be subjected to spectrum shifting through DEMON spectrum analysis, and the axial frequency and the harmonic frequency thereof are extracted, so that the information such as the propeller rotating speed, the propeller leaf number and the like of the ship target is obtained, and the target is classified and identified. The traditional DEMON spectrum axial frequency leaf frequency extraction method comprises a greatest common divisor method, a remainder threshold method, a frequency multiplication extraction method, a neural network-based method and the like. The greatest common divisor method and the remainder threshold method are easy to cause the condition of line spectrum omission, and the overall extraction accuracy is not high at low signal-to-noise ratio. The frequency multiplication extraction method has the problems that the axial frequency hypothesis value is not easy to select, and the influence on the algorithm extraction effect is large. The artificial intelligence method is severely dependent on the training sample size, but the acquisition difficulty of the samples of the hostile targets is great, and the extraction result is influenced. Therefore, it is necessary to design and propose a DEMON spectrum analysis method with low probability of line spectrum missing report, high accuracy of axial frequency leaf frequency extraction and no dependence on training samples. Disclosure of Invention The invention aims to provide a DEMON spectrum analysis axial leaf frequency extraction method. The method solves the problem that the traditional DEMON spectrum analysis method has low accuracy in extracting the axial frequency leaf frequency under the condition of low signal-to-noise ratio. The technical problems to be solved by the invention are realized by the following technical scheme: a DEMON spectrum analysis axial leaf frequency extraction method comprises the following steps: (1) Performing frequency domain beam synthesis on the array received data to obtain a plurality of groups of demodulation DEMON spectrum data in a target direction; (2) Carrying out background equalization processing on each group of demodulation DEMON spectrum data by a sequencing cut-off average method, extracting line spectrum, and obtaining a line spectrum frequency point set which is marked as P= { f 1,f2,...,fK }, by each group of data; (3) Sequencing the extracted line spectrums according to intensity, taking I line spectrums with highest intensity, sequencing from small to large according to frequency, and marking the obtained I line spectrum frequency point set as Q= { f 1,f2,...,fI }, wherein I is a set value; (4) The frequency points in the line spectrum frequency point set Q are subjected to difference frequency to obtain a difference frequency point set, and the weight corresponding to each difference frequency point is calculated; (5) Updating the difference frequency weight, taking the difference frequency corresponding to the maximum weight as the axis frequency of the data, obtaining the axis frequency value of each group of data, and putting the axis frequency value into an axis frequency set; (6) Tak