CN113591310B - Ship radiation noise line spectrum accurate extraction method based on statistic analysis - Google Patents
Ship radiation noise line spectrum accurate extraction method based on statistic analysis Download PDFInfo
- Publication number
- CN113591310B CN113591310B CN202110876663.XA CN202110876663A CN113591310B CN 113591310 B CN113591310 B CN 113591310B CN 202110876663 A CN202110876663 A CN 202110876663A CN 113591310 B CN113591310 B CN 113591310B
- Authority
- CN
- China
- Prior art keywords
- vector
- value
- line spectrum
- frequency
- lofar
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/10—Noise analysis or noise optimisation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Hardware Design (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
- Complex Calculations (AREA)
Abstract
The invention provides a ship radiation noise line spectrum accurate extraction method based on statistic analysis, which is characterized in that a maximum value, a minimum value, an average value, a median, a range and a variance are used as evaluation indexes, a method for removing a continuous spectrum trend is used for extracting suspected line spectrums possibly containing pseudo line spectrums, then the energy statistics of the line spectrums in the time period is obtained, the statistics are ranked, assigned and weighted, the pseudo line spectrums in the suspected line spectrums can be effectively removed, and line spectrum components in a Lofar image are accurately extracted.
Description
Technical Field
The invention relates to the technical field of ship radiation noise extraction, in particular to a ship radiation noise line spectrum accurate extraction method based on statistic analysis.
Background
The ship radiation noise comprises mechanical noise, propeller noise and hydrodynamic noise, and abundant line spectrum components exist on a Lofar graph obtained through short-time Fourier transform, so that the ship radiation noise is relatively stable characteristic information. The accurate extraction of the line spectrum components has important engineering significance for detecting and identifying underwater targets.
According to a common ship radiation noise line spectrum extraction method, a Lofar image is obtained through signal preprocessing and short-time Fourier transform, and then a line spectrum is extracted manually or by designing a threshold value. However, the marine environment is too complex, and interference of many other noises, such as noise of other ship targets, marine environment noise, marine animal cry, etc., can cause that line spectra in the Lofar image are dark and bright, break points occur, and even cross overlapping occurs. Due to the factors, a method for extracting a line spectrum manually or by designing a threshold value is weak in generalization, a pseudo line spectrum may exist, the subsequent underwater acoustic target detection or identification task is not facilitated, and the accuracy of the underwater acoustic target detection or identification is seriously influenced.
The energy statistics of different frequency positions of the Lofar graph, such as energy maximum value, energy minimum value, energy average value, energy median, energy range, energy variance and the like, can help researchers to screen out valuable information and remove pseudo-line spectrum components mixed in a real line spectrum. In a ship radiation noise Lofar graph, a line spectrum shows that the frequency position has a higher energy value in the whole time period, the total energy is larger after accumulation in time, and the line spectrum has a larger total energy after accumulation in time due to the fact that a certain or a plurality of time points have a special value with particularly large energy at the frequency position. The line spectrum and the pseudo-line spectrum have higher energy after time accumulation, and the line spectrum and the pseudo-line spectrum cannot be distinguished by manually designing a threshold or manually extracting the line spectrum. But since the pseudo-line spectrum is "outlier", the energy distribution of the line spectrum is more stable over the entire time period than the pseudo-line spectrum.
Disclosure of Invention
Based on the reasons, a large number of experimental analyses are combined, and aiming at the problem that a pseudo line spectrum exists in a manually designed threshold or manually extracted line spectrum, the invention provides a ship radiation noise line spectrum accurate extraction method based on statistic analysis.
The technical scheme of the invention is as follows:
the ship radiation noise line spectrum accurate extraction method based on statistic analysis comprises the following steps:
step 1: acquiring a radiation noise signal x sent by a ship, and carrying out mean value normalization processing on the radiation noise signal x sent by the ship:
normalizing the mean value of the processed signalPerforming Short Time Fourier Transform (STFT) to obtain a data matrix corresponding to the signal Lofar map:the X dimension is M multiplied by N; x has a time span of T (unit: s) and a frequency span of f (unit: Hz);
carrying out maximum and minimum normalization processing on the data matrix X of the Lofar graph:
step 2: to pairAnd (3) performing time accumulation to obtain the sum of energies of different frequency points in the T time period, and recording the sum as a vector p, wherein the jth element in the vector p is as follows:and (3) performing maximum and minimum normalization processing on the energy vectors p corresponding to different frequency points:
and step 3: vector takingMinimum value point p of min And its corresponding frequency point position f min Using f min And p min Fitting a continuous spectrum trend curve to obtain a function phi: p is a radical of formula min =Φ(f min ) (ii) a Substituting the obtained function phi into the vectorCorresponding frequency position f, obtaining the corresponding energy value of each frequency point pair of the continuous spectrum, and recording the energy value as
And 4, step 4: will vectorIs subtracted from each element ofThe energy value corresponding to each frequency point after the trend of the continuous spectrum is removed is obtained and is marked as p d :The energy vector p after removing the continuous spectrum trend d Carrying out maximum and minimum normalization processing:
and 5: will be provided withThe middle elements are arranged from large to small to form a new vector p' d Is taken to be p' d The element of the top-middle 1/2 is taken as 2 times of the average value as the extracted line spectrum threshold s:INT (N/2) is to round N/2;
step 6: will be provided withElements smaller than the threshold s are zeroed: the frequency corresponding to the element position without zero setting is the frequency f corresponding to the proposed line spectrum p :f p =(f p1 ,f p2 ,…,f pn ) N is the number of extracted line spectra;
and 7: calculating the data matrix corresponding to the Lofar graphThe maximum value for each column, denoted as vector a: a ═ a 1 ,a 2 ,…,a N ) Wherein
Calculating the data matrix corresponding to the Lofar graphThe minimum value for each column, denoted as vector b: b ═ b 1 ,b 2 ,…,b N ) Wherein
Calculating the data matrix corresponding to the Lofar graphThe average of each column, denoted as vector c: c ═ c 1 ,c 2 ,…,c N ) Wherein
Calculating the data matrix corresponding to the Lofar graphMedian of each column, denoted vector d: d ═ d (d) 1 ,d 2 ,…,d N ) In whichX' isEach row of elements in the matrix are arranged from large to small;
calculating the data matrix corresponding to the Lofar graphRange for each column, denoted as vector g: g is a-b;
calculating the data matrix corresponding to the Lofar graphThe variance of each column is recorded asVector h: h ═ h (h) 1 ,h 2 ,…,h N ) Wherein
And 8: vector a is brought at spectral frequency f p The value of the corresponding column is taken out and recorded as The vectors b, c, d, g, h are similarly aligned at the spectral frequency f p The value of the corresponding column is taken out and recorded as
And step 9: will be provided withThe minimum k-INT (n/5) elements are sequentially assigned as (k, k-1, …,1) from small to large, and the rest elements are set as 0; in the same way willThe minimum k-INT (n/5) elements are sequentially assigned as (k, k-1, …,1) from small to large, and the rest elements are set as 0; will be provided withThe largest k-INT (n/5) elements are assigned with values of (k, k-1, …,1) in sequence from large to small;
step 10: generating new vectorWeighted and summed to produce a new vector, denoted as l, preferablyAnd designing a threshold value, preferably a threshold value q ═ 0.6, (k × 4 × 2+ k × 2) ((i))When the number is more than or equal to q, the pseudo-line spectrum is positioned; will f is p Removing the frequency of the medium-false line spectrum to obtain the frequency f corresponding to the actually existing line spectrum p '。
Advantageous effects
The invention provides a ship radiation noise line spectrum accurate extraction method based on statistic analysis by using a maximum value, a minimum value, an average value, a median, a range and a variance as evaluation indexes.
Additional aspects and advantages of the invention will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the invention.
Drawings
The above and/or additional aspects and advantages of the present invention will become apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings of which:
FIG. 1: flow chart of the method of the invention.
FIG. 2 is a schematic diagram: the simulation signal Lofar diagram used in the method of the invention.
FIG. 3: the simulation signal used in the method of the invention is removed from the line spectrogram after the pseudo-line spectrum.
FIG. 4: lofar plot of measured signals for the method of the invention.
FIG. 5: the measured signal used in the method of the invention is the line spectrogram after the pseudo-line spectrum is removed.
Detailed Description
Reference will now be made in detail to embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein like reference numerals refer to the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the drawings are illustrative and intended to be illustrative of the invention and are not to be construed as limiting the invention.
The embodiment provides a method for extracting a suspected line spectrum by removing a continuous spectrum trend aiming at the problem of line spectrum extraction of radiation ship noise, and evaluates and analyzes the extracted suspected line spectrum by utilizing a maximum value, a minimum value, a mean value, a median, a range and a variance in statistics so as to achieve the purpose of removing a pseudo line spectrum. The method can more accurately extract the frequency position of the suspected line spectrum in the Lofar image, and evaluate and analyze the energy statistics of the extracted suspected line spectrum, thereby more accurately extracting the ship radiation noise line spectrum.
The signal x used in this embodiment is an analog signal simulated by matlab, and the simulation time duration is 84 s.
(1) Obtaining a radiation noise signal x ═ x 1 +x 2 +x 3 +x 4 +x 5 +x 6 +x 7 +x 8 ,
Wherein the signal x 1 =2*sin(2*π*f 1 *t),x 2 =10*sin(2*π*f 2 *t),x 3 =5*cos(2*π*f 3 *t),x 4 =7*sin(2*π*f 4 *t),x 5 =sin(2*π*f 5 *t),x 6 =3*sin(2*π*f 6 *t),x 7 =4*cos(2*π*f 7 *t),x 8 =cos(2*π*f 8 *t);f 1 =100Hz,f 2 =200Hz,f 3 =30Hz,f 4 =400Hz,f 5 =45Hz,f 6 =254Hz,f 7 =85Hz,f 8 213 Hz. The radiation noise signal x is subjected to noise addition processing, and the signal-to-noise ratio SNR is-30 dB.
(2) And (4) preprocessing a training sample. Carrying out mean normalization processing, short-time Fourier transform and maximum and minimum normalization processing on x in sequence to obtain a data matrix of the Lofar image Dimension of (e) is 159 × 500, time span is T-84 s, and frequency span is f-500 Hz.
(3) For is toAnd performing accumulation in time to obtain that the sum of the energies of different frequency points in 84s is p, and the jth element in the vector p is:and carrying out maximum and minimum normalization processing on p to obtain
(4) Vector takingMinimum value point p of min And its corresponding frequency point position f min Using f min And p min Fitting a continuous spectrum trend curve to obtain a function phi: p is a radical of min =Φ(f min )。
(5) Substituting the obtained function phi into the vectorCorresponding to the frequency position f, obtaining the corresponding energy value of each frequency point of the continuous spectrum, and recording the corresponding energy value as
(6) Will vectorIs subtracted from each element ofThe energy value corresponding to each frequency point after the trend of the continuous spectrum is removed is obtained and is marked as p d :And to p d Is carried out to the maximumMinimum normalization processing to obtain
(7) Will be provided withThe middle elements are arranged from large to small to form a new vector p' d Is taken to be p' d The element of the top-middle 1/2 is taken as 2 times of the average value as the extracted line spectrum threshold s:
(8) will be provided withElements smaller than threshold s are set to 0: the frequency corresponding to the element position with 0 being set is the frequency f corresponding to the proposed pseudo-line spectrum possibly containing pseudo-line spectrum p :f p The number of extracted line spectra n is 9 at (30,45,85,100,200,213,254,400,461) Hz.
(9) Calculating the data matrix corresponding to the Lofar graphMaximum value a for each column: a ═ a 1 ,a 2 ,…,a 500 ) Wherein a is j =max(X 1j ,X 2j ,…,X 159j ),j=1,2,…,500。
(10) Calculating the data matrix corresponding to the Lofar graphMinimum value b of each column: b ═ b 1 ,b 2 ,…,b 500 ) Wherein b is j =min(X 1j ,X 2j ,L,X 159j ),j=1,2,…,N。
(11) Calculating the data matrix corresponding to the Lofar graphAverage value c for each column: c ═ c 1 ,c 2 ,…,c 500 ) Wherein
(12) Calculating the data matrix corresponding to the Lofar graphMedian of each column, denoted vector d: d ═ d (d) 1 ,d 2 ,…,d 500 ) Wherein d is j =X' 79j X' isWherein each column of elements is arranged from small to large in the matrix.
(13) Calculating the data matrix corresponding to the Lofar graphRange for each column, denoted as vector g: g-a-b.
(14) Calculating the data matrix corresponding to the Lofar graphThe variance of each column, denoted as vector h: h ═ h (h) 1 ,h 2 ,…,h 500 ) Wherein
(15) Extracting line spectrum frequency f from vector a p The value of the corresponding column is taken out and recorded as The line spectrum frequency f extracted from the vectors b, c, d, g and h is extracted in the same way p The value of the corresponding column is taken out and recorded as
(16) Will be provided withThe minimum 2 elements are sequentially assigned with values (2,1) from small to large, and the rest elements are set to be 0; in the same way willThe smallest 2 elements are sequentially assigned with values (2,1) from small to large, and the rest elements are set to be 0; will be provided withThe largest 2 elements are sequentially assigned with values (2,1) from large to small, and the rest elements are set to be 0.
(17) Generating new vectorWeighting and adding to obtain a new vector, and recording the new vector as l:the design threshold q is 12, when l (i) is greater than 12, then this is the pseudowire spectrum.
(18) Will f is p Removing the frequency 461Hz of the medium-false line spectrum to obtain the frequency f corresponding to the actually existing line spectrum p '=(30,45,85,100,200,213,254,400)Hz。
Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention, and that variations, modifications, substitutions and alterations can be made in the above embodiments by those of ordinary skill in the art without departing from the principle and spirit of the present invention.
Claims (2)
1. A ship radiation noise line spectrum accurate extraction method based on statistic analysis is characterized by comprising the following steps: the method comprises the following steps:
step 1: acquiring a radiation noise signal x emitted by a ship, and carrying out mean value normalization processing on the radiation noise signal x emitted by the ship:
carrying out short-time Fourier transform on the signal x after the mean value normalization processing to obtain a data matrix corresponding to a signal Lofar diagram:the X dimension is M multiplied by N;
carrying out maximum and minimum normalization processing on the data matrix X of the Lofar graph:
step 2: to pairAnd (3) performing time accumulation to obtain the sum of energies of different frequency points in a time span T time period, and recording the sum as a vector p, wherein the jth element in the vector p is as follows:and (3) performing maximum and minimum normalization processing on the energy vectors p corresponding to different frequency points:
and step 3: vector takingMinimum value point p of min And its corresponding frequency point position f min Using f min And p min Fitting a continuous spectrum trend curve to obtainTo the function Φ: p is a radical of min =Φ(f min ) (ii) a Substituting the obtained function phi into the vectorCorresponding frequency position f, obtaining the corresponding energy value of each frequency point pair of the continuous spectrum, and recording the energy value as
And 4, step 4: will vectorIs subtracted from each element ofThe energy value corresponding to each frequency point after the trend of the continuous spectrum is removed is obtained and is marked as p d :The energy vector p after removing the continuous spectrum trend d And (3) performing maximum and minimum normalization processing:
and 5: will be provided withThe middle elements are arranged from large to small to form a new vector p' d Is taken to be p' d The element of the top-middle 1/2 is taken as 2 times of the average value as the extracted line spectrum threshold s:INT (N/2) is to round N/2;
step 6: will be provided withElements smaller than the threshold s are zeroed: the frequency corresponding to the element position without zero setting is the frequency f corresponding to the proposed line spectrum p :f p =(f p1 ,f p2 ,…,f pn ) N is the number of extracted line spectra;
and 7: calculating the data matrix corresponding to the Lofar graphThe maximum value for each column, denoted as vector a: a ═ a 1 ,a 2 ,…,a N ) Wherein
Calculating the minimum value of each row of the data matrix X corresponding to the Lofar graph, and recording the minimum value as a vector b: b ═ b 1 ,b 2 ,…,b N ) Wherein
Calculating the data matrix corresponding to the Lofar graphThe average for each column, denoted as vector c: c ═ c 1 ,c 2 ,…,c N ) Wherein
Lof findingData matrix corresponding to ar diagramMedian of each column, denoted vector d: d ═ d (d) 1 ,d 2 ,…,d N ) In whichX' isEach row of elements in the matrix are arranged from large to small;
calculating the data matrix corresponding to the Lofar graphRange for each column, denoted as vector g: g is a-b;
calculating the data matrix corresponding to the Lofar graphThe variance of each column, denoted as vector h: h ═ h (h) 1 ,h 2 ,…,h N ) Wherein
And 8: vector a is brought at spectral frequency f p The value of the corresponding column is taken out and recorded as The vectors b, c, d, g, h are similarly aligned at the spectral frequency f p The value of the corresponding column is taken out and recorded as
And step 9: will be provided withThe minimum k-INT (n/5) elements are sequentially assigned as (k, k-1, …,1) from small to large, and the rest elements are set as 0; in the same way willThe minimum k-INT (n/5) elements are sequentially assigned as (k, k-1, …,1) from small to large, and the rest elements are set as 0; will be provided with The largest k-INT (n/5) elements are assigned with values of (k, k-1, …,1) in sequence from large to small;
step 10: generating new vectorWeighting and adding to obtain a new vector, marking as l, and designing a threshold value q, wherein when l (i) is more than or equal to q, the position is a pseudo-line spectrum; will f is p Removing the frequency of the medium-false line spectrum to obtain the frequency f 'corresponding to the actually existing line spectrum' p 。
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110876663.XA CN113591310B (en) | 2021-07-31 | 2021-07-31 | Ship radiation noise line spectrum accurate extraction method based on statistic analysis |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110876663.XA CN113591310B (en) | 2021-07-31 | 2021-07-31 | Ship radiation noise line spectrum accurate extraction method based on statistic analysis |
Publications (2)
Publication Number | Publication Date |
---|---|
CN113591310A CN113591310A (en) | 2021-11-02 |
CN113591310B true CN113591310B (en) | 2022-09-16 |
Family
ID=78253231
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110876663.XA Active CN113591310B (en) | 2021-07-31 | 2021-07-31 | Ship radiation noise line spectrum accurate extraction method based on statistic analysis |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113591310B (en) |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109655148A (en) * | 2018-12-19 | 2019-04-19 | 南京世海声学科技有限公司 | A kind of autonomous extracting method of ship noise non-stationary low frequency spectrum lines |
CN112560949A (en) * | 2020-12-15 | 2021-03-26 | 南京航空航天大学 | Hyperspectral classification method based on multilevel statistical feature extraction |
-
2021
- 2021-07-31 CN CN202110876663.XA patent/CN113591310B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109655148A (en) * | 2018-12-19 | 2019-04-19 | 南京世海声学科技有限公司 | A kind of autonomous extracting method of ship noise non-stationary low frequency spectrum lines |
CN112560949A (en) * | 2020-12-15 | 2021-03-26 | 南京航空航天大学 | Hyperspectral classification method based on multilevel statistical feature extraction |
Non-Patent Citations (1)
Title |
---|
舰船辐射噪声特征线谱提取方法研究;王盼盼等;《海洋技术》;20100915(第03期);全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN113591310A (en) | 2021-11-02 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Roch et al. | Classification of echolocation clicks from odontocetes in the Southern California Bight | |
Mellinger et al. | A method for detecting whistles, moans, and other frequency contour sounds | |
Baumgartner et al. | A generalized baleen whale call detection and classification system | |
Rosenberg et al. | Application of the K+ Rayleigh distribution to high grazing angle sea-clutter | |
Yack et al. | Comparison of beaked whale detection algorithms | |
CN106772331A (en) | Target identification method and Target Identification Unit | |
Manzano-Roth et al. | Impacts of US Navy training events on Blainville's beaked whale (Mesoplodon densirostris) foraging dives in Hawaiian waters | |
Chen et al. | Wind speed estimation from X-band marine radar images using support vector regression method | |
Pradhan et al. | Ship detection using Neyman-Pearson criterion in marine environment | |
CN113591310B (en) | Ship radiation noise line spectrum accurate extraction method based on statistic analysis | |
CN110554392B (en) | Moving target line spectrum identification method based on frequency-energy fluctuation characteristics | |
Liu et al. | Analysis of the effects of rain on surface wind retrieval from X-band marine radar images | |
Huang et al. | Automated detection and identification of blue and fin whale foraging calls by combining pattern recognition and machine learning techniques | |
de BA Barros et al. | Development of a ship classification method based on Convolutional neural network and Cyclostationarity Analysis | |
Lin et al. | Listening to the deep: Exploring marine soundscape variability by information retrieval techniques | |
Gunawan et al. | Repurposing transfer learning strategy of computer vision for owl sound classification | |
Zhang et al. | A machine learning approach to quality-control Argo temperature data | |
CN110988985B (en) | Seismic signal detection method based on waveform characteristics | |
CN112033656A (en) | Mechanical system fault detection method based on broadband spectrum processing | |
Madhusudhana et al. | Automatic detectors for low-frequency vocalizations of Omura's whales, Balaenoptera omurai: A performance comparison | |
Hood et al. | Improved passive acoustic band-limited energy detection for cetaceans | |
Li et al. | A novel denoising method for acoustic signal | |
RU2736188C9 (en) | Hydroacoustic information displaying method | |
Conant et al. | Silbido profundo: An open source package for the use of deep learning to detect odontocete whistles | |
Klinck et al. | Detection of leopard seal (Hydrurga leptonyx) vocalizations using the Envelope-Spectrogram Technique (TEST) in combination with a Hidden Markov model |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
TR01 | Transfer of patent right |
Effective date of registration: 20221215 Address after: 315103 Zone A, Zhizaogang, Lane 218, Qingyi Road, High tech Zone, Ningbo City, Zhejiang Province Patentee after: Ningbo Institute of Northwest University of Technology Address before: 710072 No. 127 Youyi West Road, Shaanxi, Xi'an Patentee before: Northwestern Polytechnical University |
|
TR01 | Transfer of patent right |