CN114882901A - Shrimp acoustic signal time-frequency feature extraction method based on frequency domain convolution and marginal spectrum feedback - Google Patents
Shrimp acoustic signal time-frequency feature extraction method based on frequency domain convolution and marginal spectrum feedback Download PDFInfo
- Publication number
- CN114882901A CN114882901A CN202210446792.XA CN202210446792A CN114882901A CN 114882901 A CN114882901 A CN 114882901A CN 202210446792 A CN202210446792 A CN 202210446792A CN 114882901 A CN114882901 A CN 114882901A
- Authority
- CN
- China
- Prior art keywords
- frequency
- time
- spectrum
- feedback
- frequency domain
- 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.)
- Pending
Links
- 238000001228 spectrum Methods 0.000 title claims abstract description 157
- 238000000605 extraction Methods 0.000 title claims abstract description 26
- 241000143060 Americamysis bahia Species 0.000 title 1
- 241000238557 Decapoda Species 0.000 claims abstract description 59
- 238000012545 processing Methods 0.000 claims abstract description 34
- 230000014509 gene expression Effects 0.000 claims abstract description 30
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 29
- 239000013598 vector Substances 0.000 claims abstract description 22
- 230000000295 complement effect Effects 0.000 claims abstract description 20
- 238000010183 spectrum analysis Methods 0.000 claims abstract description 17
- 230000005236 sound signal Effects 0.000 claims abstract description 13
- 238000010606 normalization Methods 0.000 claims abstract description 6
- 230000002194 synthesizing effect Effects 0.000 claims abstract description 4
- 238000009499 grossing Methods 0.000 claims abstract 2
- 238000000034 method Methods 0.000 claims description 25
- 238000004458 analytical method Methods 0.000 claims description 12
- 230000008569 process Effects 0.000 claims description 9
- 230000010354 integration Effects 0.000 claims description 4
- 239000011159 matrix material Substances 0.000 claims description 4
- 238000005457 optimization Methods 0.000 claims description 4
- 238000007781 pre-processing Methods 0.000 claims description 3
- 238000012935 Averaging Methods 0.000 claims description 2
- 238000004364 calculation method Methods 0.000 claims description 2
- 238000011049 filling Methods 0.000 claims description 2
- 230000009466 transformation Effects 0.000 description 9
- 230000008901 benefit Effects 0.000 description 3
- 241000238553 Litopenaeus vannamei Species 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 238000002156 mixing Methods 0.000 description 2
- 238000004383 yellowing Methods 0.000 description 2
- 230000006399 behavior Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000009395 breeding Methods 0.000 description 1
- 230000001488 breeding effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 239000012634 fragment Substances 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000004611 spectroscopical analysis Methods 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
- G10L25/03—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the type of extracted parameters
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
- G10L25/27—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the analysis technique
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A40/00—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production
- Y02A40/80—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production in fisheries management
- Y02A40/81—Aquaculture, e.g. of fish
Landscapes
- Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Signal Processing (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- Acoustics & Sound (AREA)
- Multimedia (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
Abstract
The invention relates to a shrimp acoustic signal time-frequency feature extraction method based on frequency domain convolution and marginal spectrum feedback. S1: performing complementary set empirical mode decomposition on the shrimp sound signals to obtain a plurality of intrinsic mode function components, performing Hilbert spectrum analysis on the intrinsic mode function components, and performing normalization processing to obtain a time-frequency spectrogram of the shrimp sound signals; s2: constructing a frequency domain convolution vector according to actual conditions, and performing frequency domain convolution processing on the normalized time-frequency spectrogram to obtain a time-frequency spectrogram after frequency domain smoothing; s3: calculating to obtain a marginal spectrum, and performing primary feedback and secondary feedback on the time-frequency spectrogram after the frequency domain is smoothed by using the characteristic of reflecting the time-frequency spectrum characteristics to obtain optimized time-frequency spectrograms under different marginal spectrum feedback times; s4: and comparing and analyzing the optimized time-frequency spectrograms under different marginal spectrum feedback times, and synthesizing the spectrogram expression to obtain the time-frequency characteristics of the shrimp acoustic signals.
Description
Technical Field
The invention relates to the field of shrimp acoustic signal processing, in particular to a shrimp acoustic signal time-frequency feature extraction method based on frequency domain convolution and marginal spectrum feedback.
Background
Shrimp-like acoustic signals are mainly present in the eating and communication behaviors of shrimps and have rich biological information. In the actual production life, the shrimp acoustic signal analysis can provide reference for the habit research and the high-density intelligent breeding of the shrimps. The shrimp sound signals are taken as non-stationary signals, and single global time domain analysis or global frequency domain analysis such as Fourier transform cannot show the signal characteristics of the shrimp sound signals comprehensively, so that time-frequency double-domain analysis is needed. In addition, shrimp acoustic signals are often short and jerky, so that the analysis method needs to ensure sufficient frequency resolution.
The most common existing time-frequency analysis methods are short-time fourier transform, wavelet transform and hilbert-yellow transform. The window shape of the short-time Fourier transform is fixed and is not suitable for the characteristic of high and low frequency change of an actual signal, the wavelet transform has the advantage of multi-resolution, but is subject to the uncertainty principle as the Fourier transform, and the frequency resolution is limited. Compared with the first two kinds of transformation, the Hilbert-Huang transformation has the advantages that the Hilbert-Huang transformation breaks out of the framework of Fourier idea, is used as self-adaptive transformation based on the characteristics of signals, does not need to consider the selection of basis functions, and has high frequency resolution, so that the frequency resolution of a dense distribution area in a time frequency spectrum is obviously higher than that of short-time Fourier transformation and wavelet transformation.
However, the original hilbert yellowing transform algorithm still has the problems of modal mixing, time-frequency spectrum information redundancy and the like. The intermittent phenomenon is often caused by signal abnormal events, and extreme point distribution of signals can be changed in empirical mode decomposition of Hilbert-Huang transformation, so that local envelopes of the abnormal events appear in a fitting envelope, and the obtained intrinsic mode function components of the signals have modal mixing, which causes wrong intrinsic mode function components; the information contained in the time frequency spectrum is over accurate due to the high frequency resolution of the Hilbert-Huang transform, so that the frequency spectrum contains excessive information, the information redundancy of the time frequency spectrum is caused, and the signal time frequency feature is difficult to directly extract from the time frequency spectrum. In order to better extract the time-frequency characteristics of shrimp acoustic signals, the problems of mode aliasing and time-frequency spectrum information redundancy existing in the original Hilbert-Huang transform algorithm need to be solved.
Disclosure of Invention
The invention aims to provide a shrimp acoustic signal time-frequency feature extraction method based on frequency domain convolution and marginal spectrum feedback for solving the problems of modal aliasing and time-frequency spectrum information redundancy existing in a Hilbert-Huang transform algorithm.
The purpose of the invention can be realized by the following technical scheme:
a shrimp acoustic signal time-frequency feature extraction method based on frequency domain convolution and marginal spectrum feedback is characterized in that the time-frequency spectrum of a shrimp acoustic signal is subjected to feature analysis and extraction by adopting frequency domain convolution processing and marginal spectrum feedback, and the method comprises the following steps:
s1: performing complementary set empirical mode decomposition on the shrimp sound signals to obtain a plurality of intrinsic mode function components, performing Hilbert spectrum analysis on the intrinsic mode function components, and performing normalization processing to obtain a time-frequency spectrogram of the shrimp sound signals;
s2: constructing a frequency domain convolution vector according to actual conditions, and performing frequency domain convolution processing on the normalized time-frequency spectrogram to smooth a frequency domain of the time-frequency spectrogram to obtain the time-frequency spectrogram after the frequency domain is smoothed;
s3: calculating to obtain a marginal spectrum, and performing primary feedback and secondary feedback on the time-frequency spectrogram after the frequency domain is smoothed by using the characteristic of reflecting the time-frequency spectrum characteristics to obtain optimized time-frequency spectrograms under different marginal spectrum feedback times;
s4: and comparing and analyzing the optimized time-frequency spectrograms under different marginal spectrum feedback times, and synthesizing the spectrogram expression to obtain the time-frequency characteristics of the shrimp acoustic signals.
Further, the shrimp acoustic signal time-frequency feature extraction method based on frequency domain convolution and marginal spectrum feedback is characterized in that the shrimp acoustic signal time-frequency feature extraction method in the step S1 includes complementary set empirical mode decomposition and Hilbert spectrum analysis in the early stage; optionally, the time frequency spectrum obtained by hilbert spectrum analysis is subjected to normalization processing.
Further, the shrimp acoustic signal time-frequency feature extraction method based on frequency domain convolution and marginal spectrum feedback is characterized in that in the step S1, shrimp acoustic signals are subjected to complementary set empirical mode decomposition to obtain eigenmode function components, and then subjected to Hilbert spectrum analysis to obtain normalized time-frequency spectrums. The complementary ensemble empirical mode decomposition and hilbert spectral analysis comprises the steps of:
s11: let the original signal be x (i), add a pair of positive and negative complementary white noise sequences to x (t)
x b1 (t)＝x(t)+(-1) b σn(t)
S12: decomposing x by empirical mode b1 (t) decomposition into n eigenmode function components and the remainder R (t)
S13: repeating the steps S11 and S12m times and using different white noise sequences, the j-th decomposition result can be expressed as
S14: summing and averaging the corresponding intrinsic mode function components in each decomposition result to obtain a final intrinsic mode function:
where b is 1, 2, n (t) is a white noise sequence with an expected value of 0 and a variance of 1, n is the number of IMFs, m is the integration order, σ is the standard deviation, and σ is generally 0.1 to 0.2 times the standard deviation of the signal. And (4) ending the complementary set empirical mode decomposition, and entering a Hilbert spectrum analysis step.
S16: the generated intrinsic mode function component is analyzed by a Hilbert spectrum to obtain a Hilbert amplitude spectrum, and the expression is
Wherein w is the angular frequency,
y i (t) is IMF i (t) a Hilbert transform value,
p is the value of the cauchy principle,
further, the method for extracting the time-frequency characteristics of the shrimp acoustic signals based on the frequency domain convolution and the marginal spectrum feedback is characterized in that in the step S2, a frequency domain convolution vector is constructed according to actual requirements, and the vector is used for carrying out frequency domain convolution processing on the time-frequency spectrum to smooth the time-frequency spectrum; optionally, different frequency domain convolution vectors are constructed according to actual requirements aiming at the time-frequency feature extraction of the shrimp acoustic signal based on frequency domain convolution and marginal spectrum feedback, and different time-frequency spectrum smooth expressions are correspondingly obtained; the frequency domain convolution process comprises the following steps:
s21: the hilbert amplitude spectrum obtained in step S16 may be rewritten into a discrete form, and its expression is:
in the formula s ij ，i∈[1，k]，j∈[1，l]Representing amplitude values of corresponding points in a time-frequency domain; give a
And
constructs a frequency domain convolution vector K, denoted as:
s22: according to given
Zero-filling S to obtain matrix S 0 ，S 0 The expression of (a) is:
s23: convolving a matrix S with a frequency domain convolution vector K 0 . Center point of K is from s 11 Slide one by one to s lk . Is s' pq ，p∈[1，k]，q∈[1，l]Indicating that it was originally at S0
Row and q column points, the convolution value of which is:
the new value of the convolved point is:
the time spectrum after the frequency domain convolution process can be expressed as:
further, the method for extracting the time-frequency characteristics of the shrimp acoustic signal based on frequency domain convolution and marginal spectrum feedback is characterized in that the marginal spectrum is obtained by calculation in the step S3, and primary feedback and secondary feedback are performed on the time-frequency spectrogram after the frequency domain is smoothed by using the characteristic that the marginal spectrum reflects the frequency characteristics of the time-frequency spectrum, so as to obtain the optimized time-frequency spectrogram under different marginal spectrum feedback times, wherein the feedback process specifically comprises the following steps:
s31: calculating a marginal spectrum of the Hilbert-time spectrum of the step S16, wherein the expression of the marginal spectrum is as follows:
where w is the signal angular frequency, f is the signal frequency, and T is the signal time span; the marginal spectrum may be seen as a spectrogram of the signal, reflecting the intensity of different frequency components in the signal. Therefore, by performing the marginal spectrum feedback processing on the signal by using the value of the marginal spectrum, the contrast between high-intensity and low-intensity frequency components in the time spectrum can be increased, and instantaneous frequency points with different intensities reflected in the frequency spectrum can be more clearly distinguished.
S32: in order to avoid the loss of strong and weak energy discrimination caused by too large feedback amplitude, the average marginal spectrum is used as a feedback factor. The time frequency spectrum expression after the first feedback optimization is as follows:
the discrete expression form is as follows:
in which l is S p K is S p The number of rows of (a) to (b),
is S p (ii) a Hilbert marginal spectrum of;
s33: in order to highlight the strong frequency component, performing secondary feedback processing on the time frequency spectrum by using the marginal frequency, wherein the obtained optimized time frequency spectrum expression is as follows:
whose discrete expression is
In which l is S z K is S z Number of lines of (A) z Is S z The Hilbert marginal spectrum of (A) is,
further, the method for extracting the time-frequency characteristics of the shrimp acoustic signals based on frequency domain convolution and marginal spectrum feedback is characterized in that in the step S4, optimization time-frequency spectrograms under different marginal spectrum feedback times are compared and analyzed, and the time-frequency characteristics of the shrimp acoustic signals are obtained through comprehensive spectrogram expression. The discrimination between different frequency components in the time frequency spectrum is improved by the primary feedback of the marginal spectrum, and the intuition of the result is improved; and (3) secondary feedback of the marginal spectrum is processed again on the basis of the primary feedback result, key frequency components of the time spectrum are highlighted, other components are filtered out, and the time-frequency characteristics of the shrimp acoustic signals can be extracted by analyzing the time-frequency spectrograms of the primary feedback and the secondary feedback in a combined manner.
The invention discloses a shrimp acoustic signal time-frequency feature extraction method based on frequency domain convolution and marginal spectrum feedback, which is characterized by comprising complementary set empirical mode decomposition, Hilbert spectrum analysis, frequency domain convolution processing of a time-frequency spectrum, primary frequency domain marginal spectrum feedback and secondary frequency domain marginal spectrum feedback of the time-frequency spectrum, wherein the shrimp acoustic signal adopts the feature detection extraction method to execute the steps of the method.
Compared with the prior art, the invention has the following advantages:
the method introduces complementary set empirical mode decomposition to decompose the signal, effectively avoids the mode aliasing problem which is often caused by empirical mode decomposition of shrimp acoustic signals, improves the discrimination between different frequency components by using frequency domain convolution vectors and primary feedback of a marginal spectrum to a time frequency spectrum, reduces the spectrum information redundancy brought by the high frequency resolution of the original Hilbert spectrum analysis, emphasizes the strong frequency components of the time frequency spectrum of the signal by secondary feedback of the marginal spectrum to the time frequency spectrum, reflects the strong frequency components to the extraction of the time frequency characteristics of the shrimp acoustic signals, can effectively improve the intuition of the time frequency spectrum and the convenience of the extraction of the time frequency characteristics, and improves the accuracy of the characteristic extraction by the integration of primary feedback and secondary feedback results.
Drawings
FIG. 1 is a time domain waveform diagram of a valid sample of a shrimp-like acoustic signal;
FIG. 2 is a signal time-frequency spectrum of a Hilbert-Huang transform algorithm using complementary ensemble empirical mode decomposition according to an embodiment;
FIG. 3 shows an example of a time-frequency spectrogram-frequency domain convolution vector of a signal using a Hilbert-Huang transform algorithm with an added frequency domain convolution process as [1, 1, 1, 1 ]] T ；
FIG. 4 shows an example of a time-frequency spectrogram-frequency domain convolution vector of a signal using a Hilbert-Huang transform algorithm with an added frequency domain convolution process as [1, 2, 4, 2, 1 ]] T ；
FIG. 5 shows a signal time-frequency spectrogram-frequency domain convolution vector of the Hilbert-Huang transform algorithm with a feedback process as [1, 1, 1, 1 ]] T ；
FIG. 6 shows a signal time-frequency spectrogram-frequency domain convolution vector of a Hilbert-Huang transform algorithm using quadratic feedback processing as [1, 1, 1, 1 ]] T ；
FIG. 7 is a flowchart of a shrimp acoustic signal time-frequency feature extraction method based on frequency domain convolution and marginal spectrum feedback according to the present invention; in the figure, t is time and f is frequency.
Detailed Description
The invention is described in detail below with reference to the figures and specific embodiments. The present embodiment is implemented on the premise of the technical solution of the present invention, and a detailed implementation manner and a specific operation process are given, but the scope of the present invention is not limited to the following embodiments.
Example 1
As shown in FIG. 7, a method for extracting time-frequency characteristics of shrimp acoustic signals based on frequency domain convolution and marginal spectrum feedback, which adopts frequency domain convolution processing and marginal spectrum feedback to analyze and extract the characteristics of the time-frequency spectrum of shrimp acoustic signals,
the shrimp acoustic signals belong to non-stationary signals, and conventional single-domain transformation such as Fourier transformation cannot completely show signal characteristics, so time-frequency double-domain analysis is needed. In this embodiment, taking the sound signal of the penaeus vannamei boone as an example, the hydrophone is used to acquire a series of data, and then preprocessing operations such as sound discrimination, short-time fourier analysis and interception are performed on the data to obtain a sample sound signal fragment, which is provided to step S1.
The method comprises the following steps:
s1: performing complementary set empirical mode decomposition on the shrimp sound signals to obtain a plurality of intrinsic mode function components, performing Hilbert spectrum analysis on the intrinsic mode function components, and performing normalization processing to obtain a time-frequency spectrogram of the shrimp sound signals;
s2: constructing a frequency domain convolution vector according to actual conditions, and performing frequency domain convolution processing on the normalized time-frequency spectrogram to smooth a frequency domain of the time-frequency spectrogram to obtain the time-frequency spectrogram after the frequency domain is smoothed;
s3: calculating to obtain a marginal spectrum, and performing primary feedback and secondary feedback on the time-frequency spectrogram after the frequency domain is smoothed by using the characteristic of reflecting the time-frequency spectrum characteristics to obtain optimized time-frequency spectrograms under different marginal spectrum feedback times;
s4: and comparing and analyzing the optimized time-frequency spectrograms under different marginal spectrum feedback times, and synthesizing the spectrogram expression to obtain the time-frequency characteristics of the shrimp acoustic signals.
Each portion is described in detail below.
1. Signal decomposition and Hilbert spectroscopy
S11: the method comprises the steps of acquiring series data through a hydrophone, carrying out preprocessing operations such as sound discrimination, short-time Fourier analysis and interception on the data, and obtaining an effective signal segment as shown in a time-domain oscillogram of an effective sample of the shrimp acoustic signal shown in figure 1, wherein the effective signal segment is used as a sample signal for extracting the time-frequency characteristics of the shrimp acoustic signal.
S12: performing complementary set empirical mode decomposition on the signal x (t) to obtain an intrinsic mode function:
wherein, the expression of the j-th decomposition is as follows:
where b is 1, 2, n (t) is a white noise sequence with an expected value of 0 and a variance of 1, n is the number of IMFs, m is the integration order, σ is the standard deviation, and σ is generally 0.1 to 0.2 times the standard deviation of the signal.
S13: the obtained intrinsic mode function component is analyzed by a Hilbert spectrum to obtain a Hilbert amplitude spectrum, and the expression is as follows:
where w is the angular frequency of the signal,
y i (t) is IMF i (t) a Hilbert transform value expressed as
P is the value of the cauchy principle,
i.e. using Hilbert-Huang transform based on complementary set empirical mode decompositionThe time-frequency feature of the signal is extracted by the algorithm, and the obtained time-frequency spectrogram is shown in fig. 2 by using a hilbert yellowing transform algorithm of complementary set empirical mode decomposition.
2. Frequency domain convolution processing of signal time spectrum
The time-frequency spectrum shown in fig. 2 is weak in intuitiveness, and cannot directly extract effective time-frequency characteristics of shrimp acoustic signals, and a mean frequency-domain convolution vector K is constructed as [1, 1, 1, 1 ]] T The time-frequency spectrum and the time-frequency spectrum are subjected to frequency domain convolution processing, and an obtained time-frequency spectrogram is shown in fig. 3 by using a signal time-frequency spectrogram of a hilbert-yellow transform algorithm added with frequency domain convolution processing. Reconstructing a median frequency domain convolution vector K ═ 1, 2, 4, 2, 1] T The time-frequency spectrum and the time-frequency spectrum are subjected to frequency domain convolution processing, and an obtained time-frequency spectrogram is shown in fig. 4 by using a signal time-frequency spectrogram of a hilbert-yellow transform algorithm added with frequency domain convolution processing.
3. Frequency domain margin spectrum feedback of signal time frequency spectrum
S31: observing fig. 3 and fig. 4, the information representation of fig. 3 is more comprehensive and the information contrast of fig. 4 is higher, so that fig. 3 is selected for subsequent frequency domain marginal spectrum feedback processing to ensure sufficient display of the information. Calculating a marginal spectrum of the time frequency spectrum of the graph 3, wherein the expression of the marginal spectrum is as follows:
where f is the signal frequency and T is the signal time span; the marginal spectrum may be seen as a spectrogram of the signal, reflecting the intensity of different frequency components in the signal. Therefore, the marginal spectrum feedback processing can be performed on the signal by using the value of the marginal spectrum to increase the contrast between high-intensity and low-intensity frequency components in the time spectrum so as to better distinguish instantaneous frequency points with different intensities reflected in the frequency spectrum.
S32: in order to avoid the loss of strong and weak energy discrimination caused by overlarge feedback amplitude, an average marginal spectrum is used as a feedback factor. The time frequency spectrum expression after the first feedback optimization is as follows:
the discrete expression form is as follows:
in which l is S p Number of columns of (k) is S p The number of rows of (a) to (b),
is S p The Hilbert marginal spectrum of (1). The time-frequency spectrum obtained by the primary marginal spectrum feedback processing is shown in fig. 5, which is a signal time-frequency spectrum obtained by using a hilbert-yellow transform algorithm added with frequency domain convolution and primary feedback processing.
S33: in order to highlight the strong frequency component, performing secondary feedback processing on the time frequency spectrum by using the marginal frequency, and obtaining an optimized time frequency spectrum expression as follows:
whose discrete expression is
In which l is S z K is S z Number of lines of (A) z Is S z The Hilbert marginal spectrum of (A) is,
the time-frequency spectrum obtained by the secondary marginal spectrum feedback processing is shown in fig. 6, which is a signal time-frequency spectrum obtained by using a hilbert-yellow transform algorithm added with frequency domain convolution and secondary feedback processing.
4. Shrimp acoustic signal time-frequency feature extraction
As can be seen from the reading of FIGS. 5 and 6, when the signal frequency is limited to 0-24 kHz, the duration of the sound signal of Penaeus vannamei Boone is about 5.5ms, and the main frequency is within 12-18 kHz. The energy of the signal is strong in the first 0.6ms, during which the signal continuously outputs strong energy about 0.06ms and the frequency is about in the range of 14.9-15.3 kHz. There is then also a small tail in the signal, which is not analyzed due to equipment limitations.
The preferred embodiment method of the present invention is described in detail above. The method introduces complementary set empirical mode decomposition to decompose the signal, effectively solves the problem of mode aliasing existing in the empirical mode decomposition of the prawn acoustic signal, improves the discrimination between different frequency components by using frequency domain convolution vector and the primary feedback of the marginal spectrum to the time frequency spectrum, reduces the spectrum information redundancy brought by the high frequency resolution of Hilbert spectrum analysis, highlights the strong frequency component of the signal time frequency spectrum by the secondary feedback of the marginal spectrum to the time frequency spectrum, reflects the strong frequency component to the extraction of the time frequency characteristic of the prawn acoustic signal, can effectively improve the intuition of the time frequency spectrum and the convenience of the extraction of the time frequency characteristic, and improves the accuracy of the characteristic extraction by the synthesis of the primary feedback and the secondary feedback result. It should be understood that numerous modifications and variations could be devised by those skilled in the art in light of the present teachings without departing from the inventive concepts. Therefore, the technical solutions available to those skilled in the art through logic analysis, reasoning and limited experiments based on the prior art according to the concept of the present invention should be within the scope of protection defined by the claims.
Claims (5)
1. A shrimp acoustic signal time-frequency feature extraction method based on frequency domain convolution and marginal spectrum feedback is characterized in that the method adopts frequency domain convolution processing and marginal spectrum feedback to perform feature analysis and extraction on the time-frequency spectrum of a shrimp acoustic signal;
the method specifically comprises the following steps:
s1: performing complementary set empirical mode decomposition on the shrimp sound signals to obtain a plurality of intrinsic mode function components, performing Hilbert spectrum analysis on the intrinsic mode function components, and performing normalization processing to obtain a time-frequency spectrogram of the shrimp sound signals;
s2: constructing a frequency domain convolution vector according to actual conditions, and performing frequency domain convolution processing on the normalized time-frequency spectrogram to smooth a frequency domain of the time-frequency spectrogram to obtain the time-frequency spectrogram after the frequency domain is smoothed;
s3: calculating to obtain a marginal spectrum, and performing primary feedback and secondary feedback on the time-frequency spectrogram after the frequency domain is smoothed by using the characteristic of reflecting the frequency characteristic of the time-frequency spectrum to obtain optimized time-frequency spectrograms under different marginal spectrum feedback times;
s4: and comparing and analyzing the optimized time-frequency spectrograms under different marginal spectrum feedback times, and synthesizing the spectrogram expression to obtain the time-frequency characteristics of the shrimp acoustic signals.
2. The method for extracting shrimp acoustic signal time-frequency characteristics based on frequency domain convolution and marginal spectrum feedback as claimed in claim 1, wherein the shrimp acoustic signal time-frequency characteristics extraction method pre-processing of step S1 includes complementary set empirical mode decomposition and hilbert spectrum analysis; optionally, the time frequency spectrum obtained by hilbert spectrum analysis is subjected to normalization processing.
3. The shrimp acoustic signal time-frequency feature extraction method based on frequency domain convolution and marginal spectrum feedback as claimed in claim 2, wherein the step S1 is to perform complementary set empirical mode decomposition on the shrimp acoustic signal to obtain an eigenmode function component, and then perform hilbert spectrum analysis on the eigenmode function component to obtain a normalized time-frequency spectrum;
the complementary ensemble empirical mode decomposition and hilbert spectral analysis comprises the steps of:
s11: let the original signal be x (t), and add a pair of white noise sequences with positive and negative complementation to x (t)
x b1 (t)＝x(t)+(-1) b σn(t)
S12: decomposing x by empirical mode b1 (t) decomposition into n eigenmode function components and the remainder R (t)
S13: repeating the steps S11 and S12m times and using different white noise sequences, the j-th decomposition result is expressed as:
s14: summing and averaging the corresponding intrinsic mode function components in each decomposition result to obtain a final intrinsic mode function:
wherein b is 1, 2, n (t) is an expected 0, a white noise sequence with a variance of 1, n is the number of IMFs, m is the integration frequency, σ is the standard deviation, and the value of σ is generally 0.1-0.2 times of the standard deviation of the signal;
after the complementary set empirical mode decomposition is finished, a Hilbert spectrum analysis step is carried out;
s16: the generated intrinsic mode function component is analyzed by a Hilbert spectrum to obtain a Hilbert amplitude spectrum, and the expression is
Wherein w is the angular frequency,
y i (t) is IMF i (t) a Hilbert transform value,
p is the value of the cauchy principle,
4. the method for extracting the time-frequency characteristics of the shrimp acoustic signals based on the frequency domain convolution and the marginal spectrum feedback as claimed in claim 1, wherein in the step S2, a frequency domain convolution vector is constructed according to actual requirements, and the frequency domain convolution processing is performed on the time-frequency spectrum by using the vector to smooth the time-frequency spectrum; optionally, different frequency domain convolution vectors are constructed according to actual requirements aiming at the time-frequency feature extraction of the shrimp acoustic signal based on frequency domain convolution and marginal spectrum feedback, and different time-frequency spectrum smooth expressions are correspondingly obtained;
the frequency domain convolution process comprises the following steps:
s21: the hilbert amplitude spectrum obtained in step S16 is rewritten into a discrete form, and its expression is:
in the formula, s ij ，i∈[1，k]，j∈[1，l]Representing the amplitude value of the corresponding point in the time-frequency spectrum; give a
And
to construct a frequency domain convolution vector K, denoted as:
s22: according to given
Zero-filling S to obtain matrix S 0 ，S 0 The expression of (a) is:
s23: convolving a matrix S with a frequency domain convolution vector K 0 . Center point of K is from s 11 Slide one by one to s lk (ii) a Is s' pq ，p∈[1，k]，q∈[1，l]Indicating that it was originally at S0
Row and q columns of points whose convolution value is:
the new values of the convolved points are:
the time-frequency spectrum after the frequency domain convolution processing is represented as:
5. the method for extracting time-frequency characteristics of a shrimp acoustic signal based on frequency domain convolution and marginal spectrum feedback as claimed in claim 1, wherein the step S3 obtains a marginal spectrum by calculation, and performs primary feedback and secondary feedback on the time-frequency spectrogram after frequency domain smoothing by using the characteristics of the marginal spectrum reflecting the frequency characteristics of the time-frequency spectrum to obtain an optimized time-frequency spectrogram under different marginal spectrum feedback times, and the feedback process comprises the following specific steps:
s31: calculating a marginal spectrum of the Hilbert time spectrum of the step S16, wherein the expression of the marginal spectrum is as follows:
where w is the signal angular frequency, f is the signal frequency, and T is the signal time span; the marginal spectrum is regarded as a spectrogram of the signal, and the intensities of different frequency components in the signal are reflected;
s32: in order to avoid the loss of strong and weak energy discrimination caused by overlarge feedback amplitude, an average marginal spectrum is used as a feedback factor; the time frequency spectrum expression after the first feedback optimization is as follows:
the discrete expression form is as follows:
in which l is S p K is S p The number of rows of (a) to (b),
is S p (ii) a Hilbert marginal spectrum of;
s33: performing secondary feedback processing on the time frequency spectrum by using the marginal frequency to obtain an optimized time frequency spectrum expression as follows:
whose discrete expression is
In which l is S z K is S z Number of lines of (A) z Is S z The Hilbert marginal spectrum of (A) is,
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210446792.XA CN114882901A (en) | 2022-04-26 | 2022-04-26 | Shrimp acoustic signal time-frequency feature extraction method based on frequency domain convolution and marginal spectrum feedback |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210446792.XA CN114882901A (en) | 2022-04-26 | 2022-04-26 | Shrimp acoustic signal time-frequency feature extraction method based on frequency domain convolution and marginal spectrum feedback |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114882901A true CN114882901A (en) | 2022-08-09 |
Family
ID=82670768
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210446792.XA Pending CN114882901A (en) | 2022-04-26 | 2022-04-26 | Shrimp acoustic signal time-frequency feature extraction method based on frequency domain convolution and marginal spectrum feedback |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114882901A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117939726A (en) * | 2024-03-25 | 2024-04-26 | 深圳市能波达光电科技有限公司 | Air purifying method and system based on LED sterilizing lamp |
-
2022
- 2022-04-26 CN CN202210446792.XA patent/CN114882901A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117939726A (en) * | 2024-03-25 | 2024-04-26 | 深圳市能波达光电科技有限公司 | Air purifying method and system based on LED sterilizing lamp |
| CN117939726B (en) * | 2024-03-25 | 2024-05-24 | 深圳市能波达光电科技有限公司 | Air purifying method and system based on LED sterilizing lamp |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US8041136B2 (en) | System and method for signal processing using fractal dimension analysis | |
| CN115326783B (en) | Raman spectrum preprocessing model generation method, system, terminal and storage medium | |
| CN105279379B (en) | Tera-hertz spectra feature extracting method based on convex combination Kernel principal component analysis | |
| CN116701845B (en) | Aquatic product quality evaluation method and system based on data processing | |
| Torres-Castillo et al. | Neuromuscular disorders detection through time-frequency analysis and classification of multi-muscular EMG signals using Hilbert-Huang transform | |
| Sonavane et al. | Classification and segmentation of brain tumor using Adaboost classifier | |
| CN108567418A (en) | A kind of pulse signal inferior health detection method and detecting system based on PCANet | |
| CN114081508B (en) | Spike detection method based on fusion of deep neural network and CCA (common cancer cell) characteristics | |
| CN114882901A (en) | Shrimp acoustic signal time-frequency feature extraction method based on frequency domain convolution and marginal spectrum feedback | |
| CN117116290B (en) | Method and related equipment for positioning defects of numerical control machine tool parts based on multidimensional characteristics | |
| Naranjo-Alcazar et al. | On the performance of residual block design alternatives in convolutional neural networks for end-to-end audio classification | |
| Meng et al. | Gaussian mixture models of ECoG signal features for improved detection of epileptic seizures | |
| CN116211322A (en) | Depression recognition method and system based on machine learning electroencephalogram signals | |
| Zhang et al. | Slight crack identification of cottonseed using air-coupled ultrasound with sound to image encoding | |
| CN115944305B (en) | Electrocardiogram abnormality detection method, system, equipment and medium without heart beat segmentation | |
| CN112697270A (en) | Fault detection method and device, unmanned equipment and storage medium | |
| Khatar et al. | Advanced detection of cardiac arrhythmias using a three-stage CBD filter and a multi-scale approach in a combined deep learning model | |
| CN113940638B (en) | Pulse wave signal identification and classification method based on frequency domain dual-feature fusion | |
| CN112807000B (en) | Method and device for generating robust electroencephalogram signals | |
| Hua | Improving YANGsaf F0 Estimator with Adaptive Kalman Filter. | |
| Nejad et al. | An adaptive FECG extraction and analysis method using ICA, ICEEMDAN and wavelet shrinkage | |
| Gunjan et al. | An effective user interface image processing model for classification of Brain MRI to provide prolific healthcare | |
| US10539655B1 (en) | Method and apparatus for rapid acoustic analysis | |
| CN115541021A (en) | Method for locating characteristic peak of Raman spectrum, electronic device and storage medium | |
| CN109793511A (en) | Electrocardiosignal noise detection algorithm based on depth learning technology |
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 |