CN103412287A - Linear frequency modulation signal parameter evaluation method based on LVD (Lv's distribution) - Google Patents

Abstract

The invention discloses a method for estimating parameters of a linear frequency modulation signal based on LVD, which utilizes the energy impulse characteristic of the linear frequency modulation signal in the LVD domain, realizes relative suppression of cross items, and accurately estimates the center frequency and frequency modulation slope parameter values of the linear frequency modulation signal. Specific steps include: 1. Acquire signal, 2. Determine whether it contains impulse noise, 3. Reduce order preprocessing, 4. Extract autocorrelation features, 5. Phase scale transformation, 6. Extract LVD spectrum features, 7. Search for LVD spectrum peaks . The present invention overcomes the defect that the prior art cannot take into account cross-term suppression and resolution enhancement, converts the decoupled autocorrelation signal into an independently distributed LVD spectrum, improves the parameter estimation accuracy of the signal in a complex noise environment, and contributes to future signal phase characteristics. The design of extraction technology provides a new way.

Description

Parameter Estimation Method of LFM Signal Based on LVD

technical field

The invention belongs to the technical field of communication, and further relates to a method for estimating parameters of linear frequency modulation signals based on LVD (Lv's Distribution LVD) in the technical field of radar signal processing. The invention uses a low-order preprocessing method to suppress pulse noise, and LVD transforms to estimate the center frequency and frequency modulation slope parameters of the linear frequency modulation signal, so as to realize phase feature extraction of the linear frequency modulation signal in a complex noise environment.

Background technique

The chirp signal has a large time-width-bandwidth product, and the use of pulse compression technology can significantly reduce the peak transmit power of the radar, thereby lowering the sensitivity of the intercepting receiver and achieving the purpose of low interception of the transmitted signal. As one of the most mature low probability of intercept radar signals, chirp signal is widely used in pulse compression radar. Therefore, in the case of digital reception, accurate estimation of the center frequency and FM slope parameters of the chirp signal can realize target detection and recognition in electronic reconnaissance systems. At present, the parameter estimation methods of LFM signal mainly include bilinear time-frequency analysis method represented by Wigner-Villi distribution (WVD) and linear time-frequency analysis method represented by short-time Fourier transform.

The bilinear time-frequency analysis method represented by WVD transforms the chirp signal into the time-frequency domain through a quadratic function. This type of method has good energy accumulation for single-component chirp signals, but when the signal components increase , the transformation process inevitably has serious cross terms, which leads to the inability to correctly extract the characteristic parameters of the signal term. In order to suppress the cross term, many improved types of WVD have been proposed, such as signal decomposition, fixed kernel function design, adaptive kernel function and so on.

A patented technology owned by Chongqing University of Posts and Telecommunications "A method for suppressing cross terms in the time-frequency distribution of multi-component linear frequency modulation signals" (application number 201010550784.7, application date 2010.11.19, authorization number CN102158443A, authorization date 2011.08.17) proposes a An improved method based on subspace decomposition, which decomposes the time-frequency distribution matrix containing noise and cross-terms into signal subspace and noise subspace through eigendecomposition, and separates the signal subspace, which can reduce cross-term interference to a certain extent . The disadvantage of this patented technology is that in the process of subspace decomposition, the signals are separated from each other and it is difficult to use relevant information, resulting in the use of this patented technology cannot fully reflect the time-frequency characteristics of the chirp signal, and when the amount of data is large, the calculation rate is not ideal. .

The linear time-frequency analysis method represented by the short-time Fourier transform performs window shift on the observed signal, and then obtains the Fourier transform of the windowed signal.

The patented technology "multi-target detection method based on short-time Fourier transform and fractional Fourier transform" owned by North University of China (application number 201210335020.5, application date 2012.09.05, authorization number CN102866391A, authorization date 2013.01.09) A chirp radar signal detection method based on the combination of short-time Fourier transform and fractional order Fourier transform is proposed. This method uses a single variable to represent time-frequency information, which can effectively avoid the occurrence of cross terms to a certain extent and improve the signal-to-noise ratio of the signal to be detected. However, the disadvantage of this patented technology is that because the narrow observation window of the short-time Fourier transform reduces the resolution of the time-frequency domain, the estimation performance of the signal parameters is reduced under the condition of low signal-to-noise ratio, and the fraction of each component order Fourier spectra mask each other.

To sum up, for the parameter estimation method of chirp signal, the existing time-frequency analysis method only considers the time-frequency characteristics of chirp signal changing with time, and does not consider the unique certainty of the center frequency and FM slope on the instantaneous frequency of the signal. The extracted time-frequency features cannot directly reflect the changing characteristics of the chirp signal, and the estimation accuracy is not high by using this method for parameter estimation.

Contents of the invention

The purpose of the present invention is to overcome the above-mentioned deficiencies in the prior art, and propose a method for estimating parameters of linear frequency modulation signals based on LVD. The present invention fully considers the information that the linear frequency modulation signal is uniquely determined by the central frequency and the frequency modulation slope on the LVD plane, and converts the decoupled autocorrelation signal into an independently distributed LVD spectrum through two-dimensional Fourier transform, so as to obtain higher parameter estimation precision.

The concrete train of thought that realizes the object of the present invention is: collect the chirp signal that contains actual noise in the radar antenna, judge whether to contain pulse noise in the collected signal, if there is pulse noise, realize the suppression of pulse noise by step reduction preprocessing; The correlation function extracts the autocorrelation information of the collected signal, and then uses Fourier transform to convert it into an LVD spectrum; the LVD spectrum of a multi-component chirp signal is represented as independent pulse peaks, and the position coordinates of the peaks are searched, and the coordinate values are used as linear The center frequency and FM slope parameter values of the FM signal.

According to above-mentioned main train of thought, concrete realization steps of the present invention are as follows:

(1) Acquisition signal:

The signal acquisition system collects any chirp signal containing actual noise in the radar antenna through the receiver equipment of the pulse pressure radar.

(2) Determine whether the collected chirp signal contains impulse noise:

2a) Using the local amplitude feature method to obtain a local threshold, and use this threshold as the discrimination threshold;

2b) The amplitude statistics module compares the local amplitude of the collected chirp signal with the discrimination threshold, and judges whether the collected chirp signal contains impulse noise; if there is impulse noise, the amplitude statistics module sends a pulse indication signal, and executes Step 3: If there is no impulse noise, then the amplitude statistics module sends a collection signal, and executes step 4.

(3) According to the pulse indication signal sent by the amplitude statistics module, set the order p of the reduced-order preprocessing, and the range of p satisfies the characteristic parameters greater than 0 and smaller than the pulse noise. operation to obtain the acquisition signal after the order reduction preprocessing.

(4) Extract autocorrelation features:

According to the following formula, the autocorrelation characteristic signal of the acquired signal after the order reduction preprocessing:

$\mathrm{<span\; class="notranslate">\; R</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; +</span>\frac{\mathrm{<span\; class="notranslate">\; \&tau;</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; x</span>}}^{<span\; class="notranslate">\; *</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; -</span>\frac{\mathrm{<span\; class="notranslate">\; \&tau;</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; )</span>$

Among them, R represents the autocorrelation characteristic signal of the collected signal; x represents the collected signal; t represents the sampling time of the collected signal; τ represents the delay time of the phase of the collected signal; * represents the conjugate symbol.

(5) Phase scale transformation:

5a) Using the discrete Fourier transform method and taking the sampling time as the conversion factor, the autocorrelation characteristic signal is transformed into the frequency domain to obtain the instantaneous autocorrelation spectrum sequence;

5b) Using the Singer function interpolation method, the sampling time in the spectrum sequence is scale-transformed to obtain the interpolated spectrum sequence;

5c) Using the inverse discrete Fourier transform method, using the scale-transformed sampling time as the conversion factor, the spectrum sequence is transformed into a time-domain signal to obtain a decoupled autocorrelation signal.

(6) LVD spectrum feature extraction:

Using the two-dimensional discrete Fourier transform method, the phase delay and sampling time in the decoupled autocorrelation signal are used as conversion factors in turn, and the time-frequency domain conversion is performed to obtain the LVD spectrum.

(7) Estimate the parameter values of center frequency and FM slope:

Use the peak detection method to search for the peak value of the LVD spectrum, search the peak value of the LVD spectrum, obtain the coordinates corresponding to the point where the peak value of the spectrum is located, and use the coordinates as the parameter values of the center frequency of the linear frequency modulation signal and the frequency modulation slope.

Compared with the prior art, the present invention has the following advantages:

First, because the present invention fully considers the autocorrelation information of the coupling relationship between the chirp signal phase time and the phase delay, it overcomes the limitation that the time-frequency characteristics of the chirp signal cannot be fully reflected in the prior art, making the present invention The cross-term can be effectively suppressed, and the time-frequency feature resolution of the self-term of the LVD planar signal can be improved.

Second, because the present invention directly extracts parameter information by using two-dimensional discrete Fourier transform, it overcomes the shortcoming of signal fractional Fourier spectrum mutual shielding in the prior art, so that the present invention can realize the independent distribution of multi-component chirp signals, Effectively improve parameter estimation accuracy and operation speed.

Description of drawings

Fig. 1 is a flow chart of the present invention;

Fig. 2 is the sub-flow chart of judging whether to contain impulse noise in the acquisition signal among the present invention;

Fig. 3 is the sub-flow chart of phase scale conversion among the present invention;

Fig. 4 is the LVD spectrum simulation diagram of chirp signal in the white noise environment;

Figure 5 is a comparison diagram of LVD spectrum simulation effects before and after order reduction preprocessing in an impulsive noise environment.

Detailed ways

The present invention will be further described below in conjunction with the accompanying drawings.

With reference to Fig. 1, concrete implementation steps of the present invention are as follows:

Step 1, collect the signal:

The signal acquisition system collects the chirp signal containing actual noise in any section of the radar antenna through the receiver equipment of the pulse pressure radar, and its mixed signal model can be expressed as follows:

$\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; K</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\mathrm{<span\; class="notranslate">\; A</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}{\mathrm{<span\; class="notranslate">\; e</span>}}^{{\mathrm{<span\; class="notranslate">\; jf</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; j</span>}\frac{{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}}{\mathrm{<span\; class="notranslate">\; 2</span>}}{\mathrm{<span\; class="notranslate">\; no</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; w</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span>$

Among them, x(·) represents the collected signal, n represents the sampling time, i represents the i-th signal component, j represents the imaginary number unit, K represents the total number of signal components, A _i , f _i , r _i represent the linear frequency modulation signal of each component Amplitude, center frequency, FM slope, w(n) represents the noise term, including Gaussian noise and impulse noise.

Step 2, to determine whether the collected chirp signal contains impulse noise:

Referring to Fig. 2, the detailed steps of judging whether the collected linear frequency modulation signal contains impulse noise are as follows.

In the first step, the collected signal is input into the discrimination module as a discrimination signal.

In the second step, a detection window with a fixed length N is set, and the value range of the length N is greater than 0 and less than the total number of sampling points of the chirp signal.

The third step is to use the time-domain smoothing of the detection window to truncate the chirp signal according to the time period, divide it into multiple sub-signals of equal length and non-overlapping each other, and use the local amplitude characteristic method to calculate the amplitude mean value of the sub-signals. Get the local threshold.

In the fourth step, the threshold is used as a discrimination threshold, and the amplitude statistics module compares the local amplitude of the collected chirp signal with the discrimination threshold to determine whether the collected chirp signal contains impulse noise;

The fifth step is to judge that there is pulse noise in the collected signal, then the amplitude statistics module sends a pulse indication signal, and then execute step 3;

In the sixth step, it is judged that there is no impulse noise in the collected signal, and then the amplitude statistics module sends out the collected signal, and then step 4 is performed.

Step 3, order reduction preprocessing:

According to the pulse indication signal sent by the amplitude statistics module, set the order p of the reduced-order preprocessing, and the range of p satisfies the characteristic parameters greater than 0 and smaller than the impulse noise. Use the reduced-order preprocessing formula to perform low-order calculations on the collected signals, according to The following formula is carried out:

x ^<p> =|x| ^p+1 /x ^* , x ^-<p> = (x ^* ) ^<p> = (x ^<p> ) ^*

Among them, x ^<p> represents the collected signal after the p-order reduction preprocessing, p represents the order of the reduction preprocessing, <·> represents the symbol of the reduction preprocessing, |·| represents the symbol of the modulo function, * represents the common The conjugate symbol, x ^-<p> represents the conjugate signal after the p-order preprocessing of the acquired signal.

The low-order signal is obtained after order reduction preprocessing, which reduces the singular value amplitude in the original acquisition signal, but completely retains the phase information of the signal, which provides an effective basis for the estimation of the instantaneous frequency parameter of the signal.

Step 4, extract autocorrelation features:

According to the following formula, the autocorrelation characteristic signal of the acquired signal after the order reduction preprocessing is extracted:

$\mathrm{<span\; class="notranslate">\; R</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; +</span>\frac{\mathrm{<span\; class="notranslate">\; \&tau;</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; x</span>}}^{<span\; class="notranslate">\; *</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; -</span>\frac{\mathrm{<span\; class="notranslate">\; \&tau;</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; )</span>$

Among them, R represents the autocorrelation characteristic signal of the collected signal, x represents the collected signal, t represents the sampling time of the collected signal, τ represents the delay time of the phase of the collected signal, and * represents the conjugate symbol.

Step 5, phase scale transformation:

Referring to FIG. 3 , the detailed steps of phase scale transformation are as follows.

In the first step, the autocorrelation feature signal is used as the input signal for scale transformation.

In the second step, the fast discrete Fourier transform method is used, and the sampling time is used as the conversion factor to transform the autocorrelation characteristic signal into the frequency domain to obtain the instantaneous autocorrelation spectrum sequence.

In the third step, the Singer function interpolation method is used to scale the sampling time in the spectrum sequence to obtain the interpolated spectrum sequence. After scaling and transforming the time variable in the spectrum sequence by using the Singer function factor, there is no coupling relationship between the sampling time and the delay time.

The use of scale transformation refers to the following formula:

t=(τ+1)T

Among them, t represents the sampling time in the spectrum sequence, τ represents the delay time of the acquisition signal phase, and T represents the sampling time after scale transformation.

In the fourth step, the inverse discrete Fourier transform method is used, and the sampling time after scale transformation is used as the conversion factor to transform the spectrum sequence into a time domain signal to obtain a decoupled autocorrelation signal.

In the fifth step, the decoupled autocorrelation signal is obtained after scale transformation, and the chirp signal information after the time variable coupling relationship is eliminated, which can suppress the time-frequency ambiguity between different components of the signal in the parameter domain.

Step 6, LVD spectrum feature extraction:

Using two-dimensional discrete Fourier transform, the phase delay and sampling time in the decoupled autocorrelation signal are used as conversion factors in turn, and the time-frequency domain conversion is performed to obtain the LVD spectrum; the autocorrelation signal R is subjected to two-dimensional Fourier transform , each signal self-term can be modeled as an ideal impulse spike function:

$\mathrm{<span\; class="notranslate">\; L</span>}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; \&tau;</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{{\mathrm{<span\; class="notranslate">\; t</span>}}_{\mathrm{<span\; class="notranslate">\; no</span>}}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; R</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{\mathrm{<span\; class="notranslate">\; K</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\mathrm{<span\; class="notranslate">\; A</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}{\mathrm{<span\; class="notranslate">\; e</span>}}^{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; 2</span>}\mathrm{<span\; class="notranslate">\; \&pi;</span>}{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}}\mathrm{<span\; class="notranslate">\; \&delta;</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span>\mathrm{<span\; class="notranslate">\; \&delta;</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; r</span>}{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{\mathrm{<span\; class="notranslate">\; K</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 2</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; K</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\mathrm{<span\; class="notranslate">\; L</span>}}_{{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}}$

(·)分别表示关于τ、

的快速傅里叶变换，R表示采集信号的自相关特征信号，i表示第i个信号分量，K表示信号分量总数，A_i、f_i、r_i分别表示各分量线性调频信号的幅度、中心频率、调频斜率，j表示虚数单位，δ（·）表示冲激函数，x_i表示第i个采集信号，x_j表示第j个采集信号

表示第i个采集信号和第j个采集信号的LVD交叉项谱。Among them, L represents the LVD spectrum, F _τ (·),

(·) respectively represent about τ,

R represents the autocorrelation characteristic signal of the collected signal, i represents the i-th signal component, K represents the total number of signal components, A _i , f _i , r _i represent the amplitude and center of the linear frequency modulation signal of each component Frequency, FM slope, j represents the imaginary number unit, δ(·) represents the impulse function, x _i represents the i-th acquisition signal, x _j represents the j-th acquisition signal

Indicates the LVD cross-term spectrum of the i-th acquisition signal and the j-th acquisition signal.

For a single-component chirp signal in a noise-free real environment, only the self-term of the signal modeled as a single-frequency function exists in the LVD plane. For multi-component LFM signals, LVD transformation makes the self-term of the signal have energy impulse characteristics, makes the cross-term negligible, and has approximately linear characteristics.

Step 7, search for LVD spectrum peak coordinates:

Use the peak detection method to search for the peak of the LVD spectrum to obtain the coordinates corresponding to the point where the peak of the spectrum is located, and use the coordinates as the parameter values of the center frequency and the frequency modulation slope of the linear frequency modulation signal.

The present invention will be further described below in conjunction with the simulation diagram.

1. Simulation conditions:

The operating system of the simulation experiment of the present invention is Intel(R) Core(TM) i5CPU6503.20GHz, 32-bit Windows operating system, and the simulation software adopts MATLAB R (2011a).

The simulation parameter settings are as follows.

Select the center frequency and initial frequency of the three-component LFM signal as follows: f ₁ =-15.95Hz, r ₁ =9.44Hz/s; f ₂ =6.34Hz, r ₂ =9.44Hz/s; f ₃ =6.34Hz, r ₃ =-20.41Hz/s; the sampling rate is f _s =256Hz, and the number of sampling points N=256. Signal amplitude A ₃ =1, A ₁ =A ₂ =0.8. The additive noise was set as white noise and impulse noise respectively, and the signal-to-noise ratio was taken as -3dB.

2. Simulation results:

The LVD transform maps the chirp signal from the time domain to the parameter space, so that the energy of the chirp signal gathers at the peak point uniquely determined by the center frequency and the chirp slope, presenting a peak single-frequency signal, and the peak of the single-frequency signal represents the chirp signal energy accumulation value.

Fig. 4 shows the LVD spectrum simulation diagram of the chirp signal in the white noise environment. In Figure 4, the x coordinate represents the center frequency/Hz parameter value of the chirp signal in LVD transformation, the y coordinate represents the frequency modulation frequency/Hz/s parameter value of the chirp signal in LVD transformation, and the z coordinate represents the LVD spectrum energy of the chirp signal Values, "Signal 1", "Signal 2", and "Signal 3" represent the LVD spectrum peaks of the first, second, and third chirp components, respectively.

It can be seen from Figure 4 that there are three peak single-frequency signals with high energy accumulation in the parameter space, search for the peak peak points, find the coordinate values corresponding to the three peak points, and use them in turn as the three components of the chirp signal Center frequency and FM slope parameter estimates.

Under the condition of white noise interference, the Monte Carlo method of the prior art is used for simulation, and 100 LVD transformations are simulated under different input SNRs, and the arithmetic mean value of the estimated values of the chirp signal parameters is obtained as shown in Table 1. The actual values in Table 1 represent the simulation signal parameter values set in the simulation conditions.

Table 1-3dB white noise condition three-component chirp signal parameter estimation

As can be seen from the estimated values shown in Table 1, in the presence of white noise interference sources, the estimated values of the present invention are compared with the actual values of the signal parameters, and the error is small, indicating that the present invention can be used to accurately estimate the signal parameters estimate.

Referring to Figure 5, it is a comparison diagram of LVD spectrum simulation effects before and after order reduction preprocessing in an impulse noise environment. In Fig. 5(a), the x coordinate represents the center frequency/Hz parameter value of the chirp signal in the LVD transform, the y coordinate represents the frequency modulation frequency/Hz/s parameter value of the chirp signal in the LVD transform, and the z coordinate represents the chirp signal’s LVD spectrum energy value. In Figure 5(b), the x coordinate represents the center frequency/Hz parameter value of the chirp signal in the LVD transform, the y coordinate represents the frequency modulation frequency/Hz/s parameter value of the chirp signal in the LVD transform, and the z coordinate represents the chirp signal’s LVD spectrum energy value, "Signal 1", "Signal 2", and "Signal 3" represent the LVD spectrum peaks of the first, second and third chirp signal components respectively.

When the LVD transform processes the chirp signal in the impulsive noise environment, if the low-order preprocessing technology is not used to suppress the impact of the impulsive noise, it is impossible to accurately extract the LVD spectral features in the parameter space. It can be seen from Figure 5(a) that the LVD spectrum of the three-component LFM signal is completely submerged by the LVD spectrum of the impulse noise in the parameter space, and the three peak points of the LVD spectrum cannot be identified. After the low-order preprocessing method is used to suppress the influence of impulse noise, the linear frequency modulation signal is mapped to the parameter space through LVD transformation, and the LVD spectrum is obtained. It can be seen from Figure 5(b) that there are three LVD spectrum peak points in the parameter space, search for the peak points, find the coordinate values corresponding to the three LVD spectrum peak points, and use them as the centers of the three components of the chirp signal in turn Frequency and FM slope parameter estimates.

Under the condition of impulse noise interference, the Monte Carlo method of the prior art is used for simulation, and 100 LVD transformations are simulated under different input signal-to-noise ratios, and the arithmetic mean value of the estimated value of the chirp signal parameters is obtained as shown in Table 2. The actual values in Table 2 represent the simulation signal parameter values set in the simulation conditions.

Table 2-Estimated values of three-component linear frequency modulation signal parameters under the condition of 3dB impulse noise

As can be seen from the estimated values shown in Table 2, in the presence of impulse noise sources of interference, the estimated values of the present invention are compared with the actual values of the signal parameters, and the error is relatively small, indicating that the method of the present invention can be used for signal parameters. accurate estimate.

In summary, the results obtained from the above simulation experiments show that the present invention can suppress the influence of impulse noise, make the chirp signal have the characteristics of high energy concentration in the parameter space, improve the resolution of the parameter space, and accurately estimate the linearity FM signal center frequency and FM slope parameter values to achieve accurate extraction of signal phase features. On the premise of meeting the requirement of parameter estimation accuracy, the present invention can accurately estimate the phase parameters of the linear frequency modulation signal in real time.

Claims (4)

1. The linear frequency modulation signal parameter estimation method based on LVD comprises the steps: (1) Acquisition signal: The signal acquisition system collects any chirp signal containing actual noise in the radar antenna through the receiver equipment of the pulse pressure radar; (2) Determine whether the collected chirp signal contains impulse noise: 2a) Using the local amplitude feature method to obtain a local threshold, and use this threshold as the discrimination threshold; 2b) The amplitude statistics module compares the local amplitude of the collected chirp signal with the discrimination threshold, and judges whether the collected chirp signal contains impulse noise; if there is impulse noise, the amplitude statistics module sends a pulse indication signal, and executes the steps 3; If there is no impulse noise, the amplitude statistics module sends out a collection signal, and executes step 4; (3) Reduced order preprocessing: According to the pulse indication signal sent by the amplitude statistics module, set the order p of the reduced-order preprocessing, p is the characteristic parameter value satisfying that it is greater than 0 and smaller than the pulse noise; using the reduced-order preprocessing formula to perform low-order operations on the collected signal, we get Acquisition signal after order reduction preprocessing; (4) Extract autocorrelation features: According to the following formula, the autocorrelation characteristic signal of the acquired signal after the order reduction preprocessing is extracted: $\mathrm{<span\; class="notranslate">\; R</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; +</span>\frac{\mathrm{<span\; class="notranslate">\; \&tau;</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; x</span>}}^{<span\; class="notranslate">\; *</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; -</span>\frac{\mathrm{<span\; class="notranslate">\; \&tau;</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; )</span>$ Among them, R represents the autocorrelation characteristic signal of the collected signal, x represents the collected signal, t represents the sampling time of the collected signal, τ represents the delay time of the phase of the collected signal, and * represents the conjugate symbol; (5) Phase scale transformation: 5a) Using the discrete Fourier transform method and taking the sampling time as the conversion factor, the autocorrelation characteristic signal is transformed into the frequency domain to obtain the instantaneous autocorrelation spectrum sequence; 5b) Using the Singer function interpolation method, the sampling time in the spectrum sequence is scale-transformed to obtain the interpolated spectrum sequence; 5c) Using the inverse discrete Fourier transform method, using the scale-transformed sampling time as the conversion factor, transforming the spectrum sequence into a time-domain signal to obtain a decoupled autocorrelation signal; (6) LVD spectrum feature extraction: Using the two-dimensional discrete Fourier transform method, the phase delay and sampling time in the decoupled autocorrelation signal are used as conversion factors in turn, and the time-frequency domain conversion is performed to obtain the LVD spectrum; (7) Search for the peak value of the LVD spectrum: use the peak detection method to search for the peak value of the LVD spectrum, and obtain the coordinates corresponding to the point where the peak value of the spectrum is located, and use the coordinates as the parameter values of the center frequency and FM slope of the chirp signal. 2. The LVD-based linear frequency modulation signal parameter estimation method according to claim 1, characterized in that: the steps of the local amplitude characteristic method described in step 2a) are as follows: The first step is to set a detection window with a fixed length of N, and the value range of the length N is greater than 0 and less than the value of the total number of chirp signal sampling points; In the second step, using the time-domain smoothing of the detection window, the chirp signal is truncated according to the time period, divided into multiple sub-signals of equal length and non-overlapping each other, and the amplitude mean value of the sub-signals is calculated to obtain the local threshold. 3. the LVD-based linear frequency modulation signal parameter estimation method according to claim 1, is characterized in that: the step-down pretreatment formula described in step 3 is as follows: x ^<p> =|x| ^p+1 /x ^* , x ^-<p> = (x ^* ) ^<p> = (x ^<p> ) ^* Among them, x ^<p> represents the collected signal after the p-order reduction preprocessing, p represents the order of the reduction preprocessing, <·> represents the symbol of the reduction preprocessing, |·| represents the symbol of the modulo function, * represents the common The conjugate symbol, x ^-<p> represents the conjugate signal after the p-order preprocessing of the acquired signal. 4. The LVD-based linear frequency modulation signal parameter estimation method according to claim 1, characterized in that: the scale transformation described in step 5b) refers to performing according to the following formula: t=(τ+1)T Among them, t represents the sampling time in the spectrum sequence, τ represents the delay time of the acquisition signal phase, and T represents the sampling time after scale transformation.

Images