CN108469602A - A kind of pulse signal type automatic distinguishing method based on spectrum signature extraction - Google Patents
A kind of pulse signal type automatic distinguishing method based on spectrum signature extraction Download PDFInfo
- Publication number
- CN108469602A CN108469602A CN201810276193.1A CN201810276193A CN108469602A CN 108469602 A CN108469602 A CN 108469602A CN 201810276193 A CN201810276193 A CN 201810276193A CN 108469602 A CN108469602 A CN 108469602A
- Authority
- CN
- China
- Prior art keywords
- pulse signal
- amplitude spectrum
- amplitude
- spectrum
- snr
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
- G01S7/02—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00
- G01S7/28—Details of pulse systems
- G01S7/285—Receivers
- G01S7/292—Extracting wanted echo-signals
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Radar Systems Or Details Thereof (AREA)
Abstract
The invention discloses a kind of pulse signal type automatic distinguishing methods based on spectrum signature extraction, and this method comprises the following steps：The first step：Obtain the sample data sequence of pulse signal to be discriminated；Second step：Parameter initialization；Third walks：Calculate sample data sequence amplitude spectrum；4th step：Smoothing processing is iterated to amplitude spectrum and obtains smooth amplitude spectrum；5th step：It searches for the starting corresponding to discrete frequency index and half amplitude of main lobe corresponding to smooth amplitude spectrum maximum value and terminates discrete frequency index；6th step：Calculate separately the peak value main lobe signal-to-noise ratio of original amplitude spectrum and smooth amplitude spectrum；7th step：Differentiate pulse signal type.The characteristic parameter that the decision method of the present invention is adjudicated using the ratio of the forward and backward peak value main lobe signal-to-noise ratio of amplitude spectrum smoothing as single-frequency and chirp signal automatically, the automatic judgement of this high accuracy of two classes pulse signal can be achieved, operand is small, highly practical, is suitble to handle signal in real time.
Description
Technical field
The invention belongs to signal processing technology field more particularly to a kind of pulse signal types based on spectrum signature extraction certainly
Dynamic method of discrimination.
Background technology
Pulse signal type differentiation is to realize signal by signal processing algorithm in the case of no any priori
The automatic discrimination of type.Single-frequency and frequency modulation this two classes pulse signal have obtained most commonly used application in radar and sonar field,
Automatic discrimination is carried out to the signal type of this two classes signal, can be provided for further fining estimation signal parameter and target identification
Foundation has important theory and application value, especially in radar and sonar signal in fields such as sonar, radar and electronic warfares
Shared status is more prominent in processing.
Domestic and foreign scholars propose the automatic decision method of many pulse signal types at present, mainly have：(1) pulse is based on to believe
The interior temporal characteristics of feeling the pulse know method for distinguishing, for example, using the information such as the instantaneous amplitude of signal, instantaneous phase and instantaneous frequency as
The foundation of classification judgement, such method clear physics conception, principle is simple, but big by SNR influence；(2) it is special to be based on transform domain
The method of discrimination of sign, it is more classical at present have Short Time Fourier Transform (Short Time Fourier Transform,
STFT), wavelet transformation, mainly there is a Morlet wavelet transformations, Marr wavelet transformations and Harr wavelet transformations etc., S-transformation,
Wigner-Ville distribution (WVD) and Hilbert-Huang transform (Hilbert-Huang Transform, HHT) etc., wherein S become
Change be Short Time Fourier Transform and continuous wavelet transform thought extension or popularization, the signal-to-noise ratio lower limit that such method is suitable for is excellent
The method that temporal characteristics extract in based on pulse signal arteries and veins, but operand is huge, and classifying quality unobvious；(3) zero crossing
The method of discrimination of detection, classifying quality is good under high s/n ratio, but generally requires data sampling rate relatively high, and with signal-to-noise ratio
Decline and differentiates that effect is drastically deteriorated；(4) method based on maximum likelihood, such method ensure that classification results in shellfish in theory
It is optimal under this criterion of leaf, but disadvantage is that more prior information is classified in the case of no any priori
It is poor to adjudicate effect.
Invention content
Goal of the invention：For problem and shortage existing for above-mentioned existing method, the present invention provides one kind being based on spectrum signature
The pulse signal type automatic distinguishing method of extraction, this method can meet the needs of radar and signal processing, and can be
Compared under low signal-to-noise ratio, with the automatic judgement of very small operand realization single-frequency and the high accuracy of chirp signal, engineering is real
It is strong with property.
Technical solution：In order to achieve the above-mentioned object of the invention, the present invention uses following technical scheme：One kind is carried based on spectrum signature
The pulse signal type automatic distinguishing method taken, includes the following steps：
(1) pulse signal sampling data sequence to be discriminated is obtained：The real-time acquisition number of N number of sampled point is received from sensor
According to as data sequence x (n), n=0,1 ... to be discriminated, N-1, or extraction includes the N of entire pulse signal from memory
A sample point data is to include entire pulse signal as data sequence x (n), n=0,1 ... to be discriminated, N-1, the N
Sampling number, value be 2 integral number power；
(2) iteration smoothly judges that relevant parameter is initialized with pulse signal type：Iteration smooth greatest iteration time is set
Number thresholding I and precision controlling thresholding ε, the long M of iteration smoothing windows, judge single-frequency and frequency modulation on pulse signal-to-noise ratio ratio threshold η；
(3) pulse amplitude spectrum P (k) is calculated according to the data sequence x (n)；
(4) smoothing processing is iterated to P (k) and obtains smooth amplitude spectrum S (k)；
(5) it searches for the discrete frequency corresponding to S (k) maximum value and indexes kMAnd the starting corresponding to half amplitude of main lobe and
It terminates discrete frequency and indexes kLAnd kR；
(6) the peak value main lobe Signal to Noise Ratio (SNR) of original amplitude spectrum and smooth amplitude spectrum is calculated separatelypAnd SNRS；
(7) according to SNRpAnd SNRSPulse signal type is differentiated.
Preferably, in step (2), setting iteration smooth maximum iteration thresholding I and precision controlling thresholding ε, iteration
The long M of smoothing windows judges single-frequency and frequency modulation on pulse signal-to-noise ratio ratio threshold η, and wherein I values are the positive integer more than 2, and ε values are
Positive number less than 1, the odd number of a length of 3≤M of iteration smoothing windows≤N/2-3, the value range of η are the arbitrary number of 1.5≤η≤2.5.
It is 0.01, M values be 5, η values is 2.0 that preferred I values, which are 10, ε values,.
Preferably, in step (3), Fast Fourier Transform (FFT) is done to the data sequence x (n), data sequence is calculated
Discrete Fourier transform X (l) and amplitude spectrum P (k), specifically comprise the following steps：
(3-1) calculate x (n) discrete Fourier transform be
The discrete frequency that wherein I is X (l) indexes, and j indicates imaginary unit, i.e.,The formula can pass through fast Fourier
Transformation is realized.
(3-2) calculates the amplitude spectrum of x (n) according to X (l)
The discrete frequency that wherein k is P (k) indexes, | | represent Modulus of access operation.
Preferably, in step (3), the discrete Fourier transform X (l) that step (3-1) calculates x (n) can be by quick Fu
Leaf transformation is realized；
Preferably, in step (4), smoothing processing is iterated to P (k) and obtains smooth amplitude spectrum S (k), specifically include as
Lower step：
(4-1) initializes iterations i=0, for the first time iteration sharpening result S0(k) it is
S0(k)=P (k), k=0,1,2 ..., N/2-1
(4-2) enables iterations i=i+1, and to last iteration sharpening result Si-1(k) it is smoothed to obtain when secondary
Sharpening result Si(k)
(4-3) judges whether to reach the smooth number of greatest iteration, that is, judges whether i≤I is true, is entered step if setting up
(4-4) otherwise jumps to step (4-6)；
(4-4) calculates separately last smoothed result Si-1(k) and as time sharpening result Si(k) with original amplitude spectrum P's (k)
Residual sum of squares (RSS) Ji-1And Ji, respectively
(4-5) judges | Ji-1-Ji|≤εJiIt is whether true, (4-6) is entered step if setting up, otherwise return to step (4-2)；
(4-6) enables S (k)=Si(k), k=0,1,2 ..., N/2-1 obtain smooth amplitude spectrum S (k).
Preferably, it in step (5), searches for the discrete frequency corresponding to S (k) maximum value and indexes kMAnd half amplitude of main lobe
Corresponding starting and termination discrete frequency indexes kLAnd kR, specifically comprise the following steps：
(5-1) searches for the discrete frequency corresponding to S (k) maximum value and indexes kM
WhereinIndicate within the scope of 1≤k≤N/2-1 search for S (k) maximum value corresponding to from
Dissipate frequency indices；
(5-2) carries out maximum value normalized to S (k) must normalize smooth amplitude spectrum Z (k)：
Z (k)=S (k)/S (kM), k=0,1,2 ..., N/2-1
(5-3) searches for the starting discrete frequency corresponding to half amplitude of S (k) main lobe and indexes kL, search process includes following step
Suddenly：
(5-3-1) initializes initiating searches discrete frequency and indexes kl=1；
(5-3-2) judges kM-kl=0 or Z (kM-kl) whether -0.5 ＜ 0 true, jump to (5-3-4) if setting up, otherwise into
Enter (5-3-3)；
(5-3-3) enables kl=kl+ 1, and return to (5-3-2)；
(5-3-4) enables kL=kM-kl, obtain the starting discrete frequency index k corresponding to half amplitude of main lobeL。
(5-4) searches for the termination discrete frequency corresponding to half amplitude of S (k) main lobe and indexes kR, search process includes following step
Suddenly：
(5-4-1) initialization terminates search discrete frequency and indexes kr=1；
(5-4-2) judges kM+kr=N/2-1 or Z (kM+kr) whether -0.5 ＜ 0 true, it is no if establishment is jumped to (5-4-4)
Then enter (5-4-3)；
(5-4-3) enables kr=kr+ 1, and return to (5-4-2)；
(5-4-4) enables kR=kM-kr, obtain the termination discrete frequency index k corresponding to half amplitude of main lobeR。
Preferably, in step (6), the peak value main lobe Signal to Noise Ratio (SNR) of original amplitude spectrum and smooth amplitude spectrum is calculated separatelypWith
SNRS：
Preferably, in step (7), according to SNRpAnd SNRSPulse signal type is differentiated, following steps are divided into：
(7-1) calculates the ratio SNRR of the forward and backward peak value main lobe signal-to-noise ratio of amplitude spectrum smoothing
(7-2) judges whether SNRR ＜ η are true, is judged to chirp signal if setting up, is otherwise judged to pure-tone pulse signal.
Advantageous effect：The present invention compared with the existing methods, there is following advantageous effect：
(1) whole pulse signal is utilized in the amplitude spectrum for seeking signal in method of estimation of the invention, can get approximate
Signal processing gain with filtering, signal-to-noise ratio lower limit applicatory is low, and is realized using Fast Fourier Transform (FFT), operand
It is small；
(2) method of estimation of the invention is after the amplitude spectrum for obtaining pulse signal using Fourier transformation, to signal amplitude
Spectrum is iterated smoothing processing, can reduce requirement of the half amplitude bandwidth estimation of pulse signal to signal-to-noise ratio, further accurately to estimate
Meter peak value main lobe signal-to-noise ratio provides precondition；
(3) method of estimation of the invention defines the ratio of the forward and backward peak value main lobe signal-to-noise ratio of amplitude spectrum smoothing as signal
The characteristic quantity parameter that type is adjudicated automatically, this feature parameter have preferable discrimination to single-frequency and chirp signal, can be
Compared under low signal-to-noise ratio, realize that the high accuracy of single-frequency and chirp signal is adjudicated automatically with very small operand.
Description of the drawings
Fig. 1 is the flow diagram of the method for the present invention；
Fig. 2 is 1 exemplary pulse signal amplitude spectrogram of embodiment；
Fig. 3 is 1 exemplary pulse signal smoothing amplitude spectrogram of embodiment；
Fig. 4 is 2 exemplary pulse signal amplitude spectrogram of embodiment；
Fig. 5 is 2 exemplary pulse signal smoothing amplitude spectrogram of embodiment.
Specific implementation mode
The present invention is described further with reference to the accompanying drawings and examples：
As shown in Figure 1, a kind of pulse signal type automatic distinguishing method based on spectrum signature extraction, includes the following steps：
(1) pulse signal sampling data sequence to be discriminated is obtained：The real-time acquisition number of N number of sampled point is received from sensor
According to as data sequence x (n), n=0,1 ... to be discriminated, N-1, or extraction includes the N of entire pulse signal from memory
A sample point data is to include entire pulse signal as data sequence x (n), n=0,1 ... to be discriminated, N-1, the N
Sampling number, value be 2 integral number power.
(2) iteration smoothly judges that relevant parameter is initialized with pulse signal type：Iteration smooth greatest iteration time is set
Number thresholding I and precision controlling thresholding ε, the long M of iteration smoothing windows, judge single-frequency and frequency modulation on pulse signal-to-noise ratio ratio threshold η, wherein I
Value is the positive integer more than 2, and ε values are the positive number less than 1, and the odd number of a length of 3≤M of iteration smoothing windows≤N/2-3, η's takes
The arbitrary number of value ranging from 1.5≤η≤2.5.
In (2) step, operand of the invention and estimated accuracy, preferred I values are that 10, ε values are in order to balance
0.01, M value is that 5, η values are 2.0.
(3) pulse amplitude spectrum P (k) is calculated according to the data sequence x (n)：The data sequence x (n) is done soon
Fast Fourier transformation is calculated the discrete Fourier transform X (l) and amplitude spectrum P (k) of data sequence, specifically includes following two
Step：
(3-1) calculate x (n) discrete Fourier transform be
The discrete frequency that wherein l is X (l) indexes, and j indicates imaginary unit, i.e.,The formula can be by quick Fu
Leaf transformation is realized.
(3-2) calculates the amplitude spectrum of x (n) according to X (l)
The discrete frequency that wherein k is P (k) indexes, | | represent Modulus of access operation.
In (3) step, the discrete Fourier transform X (l) that step (3-1) calculates x (n) can pass through Fast Fourier Transform (FFT)
It realizes, improves the computational efficiency of algorithm.
(4) smoothing processing is iterated to P (k) and obtains smooth amplitude spectrum S (k), specifically comprised the following steps：
(4-1) initializes iterations i=0, for the first time iteration sharpening result S0(k) it is
S0(k)=P (k), k=0,1,2 ..., N/2-1
(4-2) enables iterations i=i+1, and to last iteration sharpening result Si-1(k) it is smoothed to obtain when secondary
Sharpening result Si(k)
(4-3) judges whether to reach the smooth number of greatest iteration, that is, judges whether i≤I is true, is entered step if setting up
(4-4) otherwise jumps to step (4-6)；
(4-4) calculates separately last smoothed result Si-1(k) and as time sharpening result Si(k) with original amplitude spectrum P's (k)
Residual sum of squares (RSS) Ji-1And Ji, respectively
(4-5) judges | Ji-1-Ji|≤εJiIt is whether true, (4-6) is entered step if setting up, otherwise return to step (4-2)；
(4-6) enables S (k)=Si(k), k=0,1,2 ..., N/2-1 obtain smooth amplitude spectrum S (k).
(5) it searches for the discrete frequency corresponding to S (k) maximum value and indexes kMAnd the starting corresponding to half amplitude of main lobe and
It terminates discrete frequency and indexes kLAnd kR, specifically comprise the following steps：
(5-1) searches for the discrete frequency corresponding to S (k) maximum value and indexes kM
WhereinIndicate within the scope of 1≤k≤N/2-1 search for S (k) maximum value corresponding to from
Dissipate frequency indices；
(5-2) carries out maximum value normalized to S (k) must normalize smooth amplitude spectrum Z (k)：
Z (k)=S (k)/S (kM), k=0,1,2 ..., N/2-1
(5-3) searches for the starting discrete frequency corresponding to half amplitude of S (k) main lobe and indexes kL, search process includes following step
Suddenly：
(5-3-1) initializes initiating searches discrete frequency and indexes kl=1；
(5-3-2) judges kM-kl=0 or Z (kM-kl) whether -0.5 ＜ 0 true, jump to (5-3-4) if setting up, otherwise into
Enter (5-3-3)；
(5-3-3) enables kl=kl+ 1, and return to (5-3-2)；
(5-3-4) enables kL=kM-kl, obtain the starting discrete frequency index k corresponding to half amplitude of main lobeL。
(5-4) searches for the termination discrete frequency corresponding to half amplitude of S (k) main lobe and indexes kR, search process includes following step
Suddenly：
(5-4-1) initialization terminates search discrete frequency and indexes kr=1；
(5-4-2) judges kM+kr=N/2-1 or Z (kM+kr) whether -0.5 ＜ 0 true, it is no if establishment is jumped to (5-4-4)
Then enter (5-4-3)：
(5-4-3) enables kr=kr+ 1, and return to (5-4-2)；
(5-4-4) enables kR=kM-kr, obtain the termination discrete frequency index k corresponding to half amplitude of main lobeR。
(6) the peak value main lobe Signal to Noise Ratio (SNR) of original amplitude spectrum and smooth amplitude spectrum is calculated separatelypAnd SNRS：
(7) according to SNRpAnd SNRSPulse signal type is differentiated, following steps are divided into：
(7-1) calculates the ratio SNRRR of the forward and backward peak value main lobe signal-to-noise ratio of amplitude spectrum smoothing
(7-2) judges whether SNRR ＜ η are true, is judged to chirp signal if setting up, is otherwise judged to pure-tone pulse signal.
In the embodiment of the present invention, emulation return pulse signal model is：
Wherein A is signal amplitude,For initial phase, τ is pulse width, f1For signal initial frequency, μ=(f2-f1)/τ
For frequency modulation rate, f2For signal terminating frequency, work as f2=f1When, i.e. it is pure-tone pulse signal when μ=0, is otherwise frequency modulation on pulse letter
Number.W (t) is that mean value is 0, variance σ2White Gaussian noise, variances sigma2Size determined by Signal to Noise Ratio (SNR)：SNR=10log (A2/
2σ2)。
With sample frequency fsDiscrete sampling is carried out to above-mentioned pulse signal, pulse signal sampling data sequence can be obtained：
Wherein Nτ=int (fsτ), int () represents the operation that rounds up.
Embodiment 1：
Emulation signal parameter is respectively set to：Signal amplitude A=2, initial phasePulsewidth τ=0.512s, signal
Initial frequency f1=300Hz, signal terminating frequency f2=300Hz, frequency modulation rate μ=0Hz/s, i.e. the emulation signal are pure-tone pulse
Signal, sample frequency fs=2000Hz, observation data sequence points N=1024, Signal to Noise Ratio (SNR)=0dB.
In (2) step, maximum iteration thresholding I=10, precision controlling thresholding ε=0.01, iteration smoothing windows are set
Long M=5 judges single-frequency and frequency modulation on pulse signal-to-noise ratio ratio threshold η=2.
According to (3) step, the amplitude spectrum P (k) of the data sequence x (n) is calculated, as shown in Figure 2.
According to (4) step, smoothing processing is iterated to P (k) and obtains smooth amplitude spectrum S (k), as shown in figure 3, comparison diagram
2 and Fig. 3 can be seen that：After iteration smoothing processing, the main lobe of pure-tone pulse signal obviously broadens, and entire band spectrum shape is more
Smoothly.
According to (5) step, search for corresponding to discrete frequency index and half amplitude of main lobe corresponding to S (k) maximum value
Starting and termination discrete frequency index are respectively
kM=155, kL=148, kR=162.
According to (6) step, the ratio of the peak value main lobe signal-to-noise ratio of original amplitude spectrum and smooth amplitude spectrum is calculated separately
SNRp=8.1592, SNRS=1.3991.
According to (7) step, the ratio of the forward and backward peak value main lobe signal-to-noise ratio of amplitude spectrum smoothing is calculated
SNRR=8.9152/1.3991=6.3719.
It can thus be appreciated that SNRR ＞ η, pure-tone pulse signal is judged to by the pulse signal, it is consistent with the signal type of setting.
Embodiment 2：
Emulation signal parameter is respectively set to：Signal amplitude A=1, initial phasePulsewidth τ=0.512s, letter
Number initial frequency f1=400Hz, signal terminating frequency f2=460Hz, frequency modulation rate μ=117.19Hz/s, i.e. the emulation signal are to adjust
Frequency pulse signal, sample frequency fs=2000Hz, observation data sequence points N=1024, Signal to Noise Ratio (SNR)=3dB.
In (2) step, maximum iteration thresholding I=8, precision controlling thresholding ε=0.005, iteration smoothing windows are set
Long M=7 judges single-frequency and frequency modulation on pulse signal-to-noise ratio ratio threshold η=1.9.
According to (3) step, the amplitude spectrum P (k) of the data sequence x (n) is calculated, as shown in Figure 4.
According to (4) step, smoothing processing is iterated to P (k) and obtains smooth amplitude spectrum S (k), as shown in figure 5, comparison diagram
4 and Fig. 5 can be seen that：After iteration smoothing processing, the main lobe width of chirp signal is basically unchanged, entire band spectrum shape
It is more smooth.
According to (5) step, search for corresponding to discrete frequency index and half amplitude of main lobe corresponding to S (k) maximum value
Starting and termination discrete frequency index are respectively
kM=217, kL=204, kR=237.
According to (6) step, the ratio of the peak value main lobe signal-to-noise ratio of original amplitude spectrum and smooth amplitude spectrum is calculated separately
SNRp=1.2851, SNRS=1.218.
According to (7) step, the ratio of the forward and backward peak value main lobe signal-to-noise ratio of amplitude spectrum smoothing is calculated
SNRR=1.2851/1.218=1.055.
It can thus be appreciated that SNRR ＜ η, chirp signal is judged to by the pulse signal, it is consistent with the signal type of setting.
Claims (8)
1. a kind of pulse signal type automatic distinguishing method based on spectrum signature extraction, which is characterized in that include the following steps：
(1) pulse signal sampling data sequence x (n), n=0,1 ... to be discriminated are obtained, N-1, the N are the arteries and veins detected
Rush the sampled point number corresponding to signal pulsewidth length, the integral number power that value is 2；
(2) parameter smoothly judged with pulse signal type iteration initializes；
(3) pulse amplitude spectrum P (k) is calculated according to the data sequence x (n)；
(4) smoothing processing is iterated to pulse amplitude spectrum P (k) and obtains smooth amplitude spectrum S (k)；
(5) starting corresponding to discrete frequency index and half amplitude of main lobe corresponding to smooth Amplitude maxima and end are searched for
Only discrete frequency indexes；
(6) the peak value main lobe signal-to-noise ratio of original amplitude spectrum and smooth amplitude spectrum is calculated separately；
(7) differentiate pulse signal type.
2. the pulse signal type automatic distinguishing method according to claim 1 based on spectrum signature extraction, which is characterized in that
In step (1), pulse signal sampling data sequence x (n) to be discriminated is obtained with the following method：It is received from sensor N number of
The real-time data collection of sampled point is as data sequence x (n) to be discriminated；Or extraction includes entire pulse signal from memory
N number of sample point data as data sequence x (n) to be discriminated.
3. the pulse signal type automatic distinguishing method according to claim 1 based on spectrum signature extraction, which is characterized in that
In step (2), parameter, which initializes, smoothly to be judged with pulse signal type to iteration with the following method：Setting initialization
Iteration smooth maximum iteration thresholding I and precision controlling thresholding ε, the long M of iteration smoothing windows judge single-frequency and frequency modulation on pulse letter
It makes an uproar than ratio threshold η, wherein I values are the positive integer more than 2, and ε values are the positive number less than 1, a length of 3≤M of iteration smoothing windows
The odd number of≤N/2-3, the value range of η are the arbitrary number of 1.5≤η≤2.5.
4. the pulse signal type automatic distinguishing method according to claim 1 based on spectrum signature extraction, which is characterized in that
In step (3), pulse amplitude spectrum P (k) is calculated according to the data sequence x (n) with the following method：To the data
Sequence x (n) does Fast Fourier Transform (FFT), and the discrete Fourier transform X (l) and pulse amplitude spectrum of data sequence is calculated
P (k) includes the following steps：
(3-1) calculate x (n) discrete Fourier transform be
The discrete frequency that wherein l is X (l) indexes, and j indicates imaginary unit, i.e.,The formula passes through Fast Fourier Transform (FFT) reality
It is existing；
(3-2) calculates the pulse amplitude spectrum of x (n) according to X (l)
The discrete frequency that wherein k is P (k) indexes, | | represent Modulus of access operation.
5. the pulse signal type automatic distinguishing method according to claim 1 based on spectrum signature extraction, which is characterized in that
In step (4), smoothing processing is iterated to pulse amplitude spectrum P (k) with the following method and obtains smooth amplitude spectrum S
(k), specifically comprise the following steps：
(4-1) initializes iterations i=0, for the first time iteration sharpening result S0(k) it is
S0(k)=P (k), k=0,1,2 ..., N/2-1,
(4-2) enables iterations i=i+1, and to last iteration sharpening result Si-1(k) it is smoothed to obtain current smooth
As a result Si(k)
Wherein, M is that iteration smoothing windows are long；
(4-3) judges whether to reach the smooth number of greatest iteration, that is, judges whether i≤I is true, and (4-4) is entered step if setting up,
Otherwise step (4-6) is jumped to, I is the smooth maximum iteration thresholding of iteration；
(4-4) calculates separately last smoothed result Si-1(k) and as time sharpening result Si(k) flat with the residual error of original amplitude spectrum P (k)
Side and Ji-1And Ji, respectively
(4-5) judges | Ji-1-Ji|≤εJiIt is whether true, (4-6) is entered step if setting up, otherwise return to step (4-2), wherein
ε is precision controlling thresholding；
(4-6) enables S (k)=Si(k), k=0,1,2 ..., N/2-1 obtain smooth amplitude spectrum S (k).
6. the pulse signal type automatic distinguishing method according to claim 1 based on spectrum signature extraction, which is characterized in that
In step (5), the discrete frequency index k corresponding to smooth amplitude spectrum S (k) maximum value is searched for the following methodM, Yi Jizhu
Starting and termination discrete frequency corresponding to half amplitude of valve index kLAnd kR, include the following steps：
(5-1) searches for the discrete frequency corresponding to S (k) maximum value and indexes kM
WhereinIndicate to search for the discrete frequency corresponding to the maximum value of S (k) within the scope of 1≤k≤N/2-1
Index；
(5-2) carries out maximum value normalized to S (k) must normalize smooth amplitude spectrum Z (k)：
Z (k)=S (k)/S (kM), k=0,1,2 ..., N/2-1
(5-3) searches for the starting discrete frequency corresponding to half amplitude of S (k) main lobe and indexes kL, search process includes the following steps：
(5-3-1) initializes initiating searches discrete frequency and indexes kl=1；
(5-3-2) judges kM-kl=0 or Z (kM-kl) whether -0.5 ＜ 0 true, if (5-3-4) is jumped in establishment, otherwise enter (5-
3-3)；
(5-3-3) enables kl=kl+ 1, and return to (5-3-2)；
(5-3-4) enables kL=kM-kl, obtain the starting discrete frequency index k corresponding to half amplitude of main lobeL；
(5-4.) searches for the termination discrete frequency corresponding to half amplitude of S (k) main lobe and indexes kR, search process includes the following steps：
(5-4-1) initialization terminates search discrete frequency and indexes kr=1；
(5-4-2) judges kM+kr=N/2-1 or Z (kM+kr) whether -0.5 ＜ 0 true, (5-4-4) is jumped to if setting up, is otherwise entered
(5-4-3)；
(5-4-3) enables kr=kr+ 1, and return to (5-4-2)；
(5-4-4) enables kR=kM-kr, obtain the termination discrete frequency index k corresponding to half amplitude of main lobeR。
7. the pulse signal type automatic distinguishing method according to claim 1 based on spectrum signature extraction, which is characterized in that
In step (6), the peak value main lobe Signal to Noise Ratio (SNR) of original amplitude spectrum and smooth amplitude spectrum is calculated separately with the following methodpWith
SNRS：
Wherein, S (k) smooth amplitude spectrum, kMFor discrete frequency index, the k corresponding to S (k) maximum valuesLAnd kRRespectively main lobe half
Starting and termination discrete frequency index, P (k) corresponding to amplitude are pulse amplitude spectrum, S (kM) it is that smooth amplitude is maximum
Value, P (kM) it is pulse amplitude maximum value.
8. the pulse signal type automatic distinguishing method according to claim 1 based on spectrum signature extraction, which is characterized in that
In step (7), pulse signal type is differentiated with the following method, according to SNRpAnd SNRSPulse signal type is differentiated,
Steps are as follows：
(7-1) calculates the ratio SNRR of the forward and backward peak value main lobe signal-to-noise ratio of amplitude spectrum smoothing
Wherein, SNRpAnd SNRSFor the peak value main lobe signal-to-noise ratio of original amplitude spectrum and smooth amplitude spectrum, max (SNRP, SNRS) indicate
For SNRpAnd SNRSMaximum value, min (SNRP, SNRS) it is expressed as SNRpAnd SNRSMinimum value；
(7-2) judges whether SNRR ＜ η are true, is judged to chirp signal if setting up, is otherwise judged to pure-tone pulse signal；Its
In, n is single-frequency and frequency modulation on pulse signal-to-noise ratio ratio threshold.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810276193.1A CN108469602B (en) | 2018-03-29 | 2018-03-29 | Pulse signal type automatic discrimination method based on spectral feature extraction |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810276193.1A CN108469602B (en) | 2018-03-29 | 2018-03-29 | Pulse signal type automatic discrimination method based on spectral feature extraction |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108469602A true CN108469602A (en) | 2018-08-31 |
| CN108469602B CN108469602B (en) | 2021-09-07 |
Family
ID=63262094
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810276193.1A Active CN108469602B (en) | 2018-03-29 | 2018-03-29 | Pulse signal type automatic discrimination method based on spectral feature extraction |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108469602B (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108985282A (en) * | 2018-09-21 | 2018-12-11 | 中国铁道科学研究院集团有限公司铁道科学技术研究发展中心 | A kind of signal processing method for identifying the low taper snake of high-speed EMUs and shaking vehicle feature |
| CN109655832A (en) * | 2019-03-13 | 2019-04-19 | 上海赫千电子科技有限公司 | A kind of blind zone detection method |
| CN110068727A (en) * | 2019-04-09 | 2019-07-30 | 东南大学 | A kind of simple signal frequency estimating methods based on the comprehensive interpolation of Candan-Rife |
| CN110852201A (en) * | 2019-10-28 | 2020-02-28 | 东南大学 | Pulse signal detection method based on multi-pulse envelope spectrum matching |
| CN111273233A (en) * | 2020-03-04 | 2020-06-12 | 北京环境特性研究所 | Asynchronous pulse detection method and device for electronic corner reflector |
| CN112099081A (en) * | 2020-08-26 | 2020-12-18 | 武汉大学 | Single-pulse and multi-pulse seismic oscillation identification and extraction method |
| CN113884997A (en) * | 2021-09-18 | 2022-01-04 | 北京遥感设备研究所 | Anti-interference shielding pulse main lobe broadening test method and system for radar equipment |
| CN114896554A (en) * | 2022-05-10 | 2022-08-12 | 东南大学 | Frequency modulation signal frequency range and bandwidth estimation method based on spectral feature extraction |
| CN115659162A (en) * | 2022-09-15 | 2023-01-31 | 云南财经大学 | Method, system and equipment for extracting features in radar radiation source signal pulse |
| CN118075069A (en) * | 2024-04-17 | 2024-05-24 | 成都玖锦科技有限公司 | Frequency modulation pulse parameter calculation method and device, cloud end equipment and computer device |
Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102707157A (en) * | 2012-05-18 | 2012-10-03 | 天津理工大学 | Single-frequency impulse signal parameter estimating method based on power spectrum |
| CN102721950A (en) * | 2011-03-29 | 2012-10-10 | 中国科学院电子学研究所 | Processing method for m-sequential code pulse modulation ultra-wide-band radar echo |
| US20120313809A1 (en) * | 2011-06-10 | 2012-12-13 | Sony Corporation | Signal processing unit and method |
| CN103207389A (en) * | 2013-03-14 | 2013-07-17 | 中国人民解放军63801部队 | Radar antenna side lobe identification algorithm based on signal magnitude-phase characteristics |
| CN103235296A (en) * | 2013-05-05 | 2013-08-07 | 西安电子科技大学 | Power spectral feature correction-based radar target identification method under noise background |
| CN104198999A (en) * | 2014-08-07 | 2014-12-10 | 北京理工大学 | Pulse sorting method based on isomorphic sequence |
| CN104468001A (en) * | 2014-11-26 | 2015-03-25 | 北京邮电大学 | Signal identification method and system based on radio signal frequency spectrum feature template |
| CN105072067A (en) * | 2015-08-27 | 2015-11-18 | 西安电子科技大学 | Method for detecting burst-mode signal in low signal to noise ratio |
| CN105093207A (en) * | 2015-07-10 | 2015-11-25 | 南京大学 | Optimized-LM-algorithm-based laser radar waveform decomposition method |
| US20160209497A1 (en) * | 2015-01-20 | 2016-07-21 | Raytheon Bbn Technologies Corp. | System and method for authenticated interrogation of a target with quantum entanglement |
| CN106896348A (en) * | 2017-01-16 | 2017-06-27 | 哈尔滨工程大学 | A kind of Radar Signal Sorting Method based on probabilistic data association |
| CN107479036A (en) * | 2017-07-10 | 2017-12-15 | 上海电机学院 | A kind of radar signal feature method |
| CN107748356A (en) * | 2017-11-06 | 2018-03-02 | 西安电子科技大学 | Phase-coded signal deception jamming recognition methods based on multiple resemblance Coefficient |
-
2018
- 2018-03-29 CN CN201810276193.1A patent/CN108469602B/en active Active
Patent Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102721950A (en) * | 2011-03-29 | 2012-10-10 | 中国科学院电子学研究所 | Processing method for m-sequential code pulse modulation ultra-wide-band radar echo |
| US20120313809A1 (en) * | 2011-06-10 | 2012-12-13 | Sony Corporation | Signal processing unit and method |
| CN102707157A (en) * | 2012-05-18 | 2012-10-03 | 天津理工大学 | Single-frequency impulse signal parameter estimating method based on power spectrum |
| CN103207389A (en) * | 2013-03-14 | 2013-07-17 | 中国人民解放军63801部队 | Radar antenna side lobe identification algorithm based on signal magnitude-phase characteristics |
| CN103235296A (en) * | 2013-05-05 | 2013-08-07 | 西安电子科技大学 | Power spectral feature correction-based radar target identification method under noise background |
| CN104198999A (en) * | 2014-08-07 | 2014-12-10 | 北京理工大学 | Pulse sorting method based on isomorphic sequence |
| CN104468001A (en) * | 2014-11-26 | 2015-03-25 | 北京邮电大学 | Signal identification method and system based on radio signal frequency spectrum feature template |
| US20160209497A1 (en) * | 2015-01-20 | 2016-07-21 | Raytheon Bbn Technologies Corp. | System and method for authenticated interrogation of a target with quantum entanglement |
| CN105093207A (en) * | 2015-07-10 | 2015-11-25 | 南京大学 | Optimized-LM-algorithm-based laser radar waveform decomposition method |
| CN105072067A (en) * | 2015-08-27 | 2015-11-18 | 西安电子科技大学 | Method for detecting burst-mode signal in low signal to noise ratio |
| CN106896348A (en) * | 2017-01-16 | 2017-06-27 | 哈尔滨工程大学 | A kind of Radar Signal Sorting Method based on probabilistic data association |
| CN107479036A (en) * | 2017-07-10 | 2017-12-15 | 上海电机学院 | A kind of radar signal feature method |
| CN107748356A (en) * | 2017-11-06 | 2018-03-02 | 西安电子科技大学 | Phase-coded signal deception jamming recognition methods based on multiple resemblance Coefficient |
Non-Patent Citations (3)
| Title |
|---|
| RUAIRÍ DE FRÉIN等: "The Synchronized Short-Time-Fourier-Transform: Properties and Definitions for Multichannel Source Separation", 《IEEE TRANSACTIONS ON SIGNAL PROCESSING》 * |
| 吴振兴等: "基于谱特征的雷达信号脉内编码调制样式识别", 《电子信息对抗技术》 * |
| 胡嬿等: "相参脉冲信号识别方法和性能分析", 《数据采集与处理》 * |
Cited By (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108985282A (en) * | 2018-09-21 | 2018-12-11 | 中国铁道科学研究院集团有限公司铁道科学技术研究发展中心 | A kind of signal processing method for identifying the low taper snake of high-speed EMUs and shaking vehicle feature |
| CN108985282B (en) * | 2018-09-21 | 2021-04-02 | 中国铁道科学研究院集团有限公司铁道科学技术研究发展中心 | Signal processing method for identifying low-taper crawling and shaking characteristics of high-speed motor train unit |
| CN109655832A (en) * | 2019-03-13 | 2019-04-19 | 上海赫千电子科技有限公司 | A kind of blind zone detection method |
| CN109655832B (en) * | 2019-03-13 | 2019-06-21 | 上海赫千电子科技有限公司 | A kind of blind zone detection method |
| CN110068727B (en) * | 2019-04-09 | 2021-03-30 | 东南大学 | Single-frequency signal frequency estimation method based on Candan-Rife comprehensive interpolation |
| CN110068727A (en) * | 2019-04-09 | 2019-07-30 | 东南大学 | A kind of simple signal frequency estimating methods based on the comprehensive interpolation of Candan-Rife |
| CN110852201A (en) * | 2019-10-28 | 2020-02-28 | 东南大学 | Pulse signal detection method based on multi-pulse envelope spectrum matching |
| CN110852201B (en) * | 2019-10-28 | 2023-08-11 | 东南大学 | Pulse signal detection method based on multi-pulse envelope spectrum matching |
| CN111273233A (en) * | 2020-03-04 | 2020-06-12 | 北京环境特性研究所 | Asynchronous pulse detection method and device for electronic corner reflector |
| CN111273233B (en) * | 2020-03-04 | 2022-05-03 | 北京环境特性研究所 | Asynchronous pulse detection method and device for electronic corner reflector |
| CN112099081A (en) * | 2020-08-26 | 2020-12-18 | 武汉大学 | Single-pulse and multi-pulse seismic oscillation identification and extraction method |
| CN113884997A (en) * | 2021-09-18 | 2022-01-04 | 北京遥感设备研究所 | Anti-interference shielding pulse main lobe broadening test method and system for radar equipment |
| CN113884997B (en) * | 2021-09-18 | 2024-06-04 | 北京遥感设备研究所 | Method and system for testing main lobe broadening of anti-interference shield pulse of radar equipment |
| CN114896554A (en) * | 2022-05-10 | 2022-08-12 | 东南大学 | Frequency modulation signal frequency range and bandwidth estimation method based on spectral feature extraction |
| CN114896554B (en) * | 2022-05-10 | 2024-06-21 | 东南大学 | Frequency modulation signal frequency range and bandwidth estimation method based on spectral feature extraction |
| CN115659162A (en) * | 2022-09-15 | 2023-01-31 | 云南财经大学 | Method, system and equipment for extracting features in radar radiation source signal pulse |
| CN115659162B (en) * | 2022-09-15 | 2023-10-03 | 云南财经大学 | Method, system and equipment for extracting intra-pulse characteristics of radar radiation source signals |
| CN118075069A (en) * | 2024-04-17 | 2024-05-24 | 成都玖锦科技有限公司 | Frequency modulation pulse parameter calculation method and device, cloud end equipment and computer device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108469602B (en) | 2021-09-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108469602A (en) | A kind of pulse signal type automatic distinguishing method based on spectrum signature extraction | |
| CN106130942B (en) | A kind of wireless communication signals Modulation Identification and method for parameter estimation based on Cyclic Spectrum | |
| CN108680786A (en) | A kind of pulse signal frequency domain adaptive filtering envelope extraction method | |
| CN111175718B (en) | Automatic target recognition method and system for ground radar combining time-frequency domains | |
| CN106443178B (en) | A kind of sinusoidal signal frequency estimation method comprehensive based on IQuinn-Rife | |
| CN108509377A (en) | A kind of pulse signal arrival time based on along feature extraction and pulsewidth method of estimation | |
| CN109165678A (en) | Emitter Recognition and device based on bispectrum 3-D image textural characteristics | |
| CN105785324B (en) | Linear frequency-modulated parameter estimating method based on MGCSTFT | |
| CN105071830B (en) | A kind of detection recognition method of direct sequence signal | |
| CN110690931B (en) | Digital signal adaptive code rate estimation method and device based on multi-wavelet-base combination | |
| CN105429719B (en) | Based on power spectrum and multi-scale wavelet transformation analysis high reject signal detection method | |
| CN114363128B (en) | Broadband signal detection method based on deep learning | |
| CN109034087B (en) | PCA (principal component analysis) dimension reduction-based hybrid machine learning signal classification method | |
| CN114896554B (en) | Frequency modulation signal frequency range and bandwidth estimation method based on spectral feature extraction | |
| CN113780521B (en) | Radiation source individual identification method based on deep learning | |
| CN109584256A (en) | A kind of pulsar DM algorithm for estimating based on Hough straight-line detection | |
| CN116166936A (en) | Radar signal modulation type identification method based on wavelet transformation phase mutation detection | |
| CN111540381A (en) | Voice simulation modulation characteristic recognition method based on random forest | |
| CN105959035B (en) | A kind of direct sequence signal intercepts and captures detection method | |
| CN112910813B (en) | LDA-KNN-based underwater sound signal automatic modulation identification method | |
| CN112202696B (en) | Underwater sound signal automatic modulation identification method based on fuzzy self-encoder | |
| Zhang et al. | A speech endpoint detection method based on wavelet coefficient variance and sub-band amplitude variance | |
| CN113988121A (en) | Weighted voting multi-classifier ground radar target automatic identification method based on feature extraction | |
| CN102624418B (en) | Hydroacoustic biphase modulation direct sequence spread spectrum signal carrier frequency estimation method | |
| Kuo et al. | Extraction of radar target length based on high resolution range profile |
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 |