CN106249205A - A kind of cognitive blind balance method for radar channel amplitude Yu phase only pupil filter - Google Patents

Abstract

The invention discloses a kind of cognitive blind balance method for radar channel amplitude Yu phase only pupil filter, including: high-ranking officers' calibration signal is sent to blind equalizer；The output signal of blind equalizer is sent to select switch, monitoring and multiple relevant parameters of blind equalizer convergence property, when the balancing procedure of blind equalizer is normal, exports the signal through equilibrium treatment；When the balancing procedure of blind equalizer is abnormal, do not carry out equilibrium treatment, will export after signal time delay.Blind equalizer uses multiple cognitive techniques to be monitored, and the most whether it is by multiple parameter monitorings, thus ensures the reliability that blind equalizer uses.The present invention realizes digital array radar closed loop equilibrium in a digital transmitting-receiving subassembly, it is not necessary to data down transmission and equalizing coefficient passback；Meanwhile, introduce the cognitive approach monitored for mean square error and weight coefficient vector, improve the convergence reliability of constant mould blind balance method, thus ensure the reliability of digital array radar hundreds of radar channel equalizer.

Description

A kind of cognitive blind balance method for radar channel amplitude Yu phase only pupil filter

Technical field

The invention belongs to digital array radar field, be specifically related to a kind of cognition for radar channel amplitude Yu phase only pupil filter Blind balance method.

Background technology

There is amplitude in digital array radar multiple passage band inconsistent with phase place, even if for narrowband systems, conventional is logical Road calibration typically uses simple signal, it is achieved the consistent correction of certain frequency.And for other frequency in several million bandwidth Point, amplitude and phase place yet suffer from difference.Traditional method use adaptive equalization technique realize band in all frequencies amplitude with Phase only pupil filter.But adaptive equalization technique, needs to refer to passage.During the realization of digital array radar, needs are by reference Channel signal and each channel signal reach below front；After completing adaptive polo placement, also need equalizer coefficients reaches numeral receipts Send out assembly, it is achieved equilibrium calculation.This process needs complicated data transmission link, and system realizes complexity.

Summary of the invention

For the problems referred to above, the present invention proposes a kind of cognitive blind equalization side for radar channel amplitude Yu phase only pupil filter Method, is applied to radar channel-equalization by constant mould Blind Equalization Technique, it is achieved digital array radar is in a digital transmitting-receiving subassembly Closed loop equalizes, it is not necessary to data down transmission and equalizing coefficient passback；Meanwhile, introduce for recognizing that mean square error and weight coefficient vector are monitored Know technology, promote the convergence reliability of constant mould blind equalization algorithm.

Realizing above-mentioned technical purpose, reach above-mentioned technique effect, the present invention is achieved through the following technical solutions:

A kind of cognitive blind balance method for radar channel amplitude Yu phase only pupil filter, comprises the following steps:

Each digital transmitting and receiving group that step one, high-ranking officers' calibration signal are fed in digital array radar through power divider and feedback net The front end of part, and be sent in cognitive blind equalizer as input signal；

Step 2, the output signal of cognitive blind equalizer is sent to select switch, monitoring and blind equalizer convergence property Relevant parameter, it is judged that the convergence process of blind equalizer, and perform following operation:

A: when the convergence process of cognitive blind equalizer is normal, selects switch this cognitive blind equalization output signal to be transferred to Nonlinear transformation module carries out nonlinear transformation, it is thus achieved that adaptive error, and it is blind all to utilize this adaptive error to feed back to cognition The input of weighing apparatus, it is achieved the self-adaptative adjustment of blind equalizer equalizing coefficient vector, the equilibrium system of the cognitive blind equalization after convergence Number vector is for realizing the equilibrium treatment to input signal, and final output is through the signal of equilibrium treatment；

B: when the output of cognitive blind equalizer cannot restrain, selects switch this cognitive blind equalization output signal to be transferred to Time delay module, time delay module is delayed output when input signal being carried out, is not carried out equilibrium treatment.

The parameter that described and cognitive blind equalizer convergence property is relevant, including: instantaneous error, equalizer coefficients delay inequality, Equalizer coefficients ratio and equalizer output signal modulus value.

In the present invention, setting calibration signal as p (k), equalizer input is x_i(k), it is assumed that there are M passage needs blind Equilibrium treatment, then the equalizer output signal y of i-th passage_i(k) be:

$y_i\left(k\right)=f_i\left(k\right)\⊗x_i\left(k\right)---\left(1\right)$

Wherein: f_iK () represents i-th N_f× 1 weight coefficient vector, N_fRepresent equalizer weight coefficient vector length,Represent volume Long-pending computing；Definition coefficient of kurtosis γ:

$\mathrm{\γ}=\frac{E\left\{\right|p\left(k\right){|}^4\}}{E\left\{\right|p\left(k\right){|}^2\}}---\left(2\right)$

E{ } represent and ask for mathematic expectaion, | | representing magnitude calculation, output signal is:

e_i(k)=y_i(k)[γ-|y_i(k)|²] (3)

The input recursive vector of definition i-th equalizer is:

X_i(k)=[x_i(k) x_i(k-1) … x_i(k-N_f+1)]^T (4)

Then weight coefficient vector is iterative is expressed as:

f_i(k+1)=f_i(k)+μ₁·X_i ^*(k)e(k) (5)

In formula, μ₁Step factor, []^*Represent complex conjugate；

Described nonlinear block gained adaptive error is:

e_s(k)=(γ-| y_i(k)|²)(γ-|y_i(k)|²) (6)

Definition N_h× 1 dimension wave filter, N_hFor the length of this wave filter, y_iK () is the equalizer output signal of i-th passage； γ is coefficient of kurtosis；

$\mathrm{\γ}=\frac{E\left\{\right|p\left(k\right){|}^4\}}{E\left\{\right|p\left(k\right){|}^2\}}---\left(2\right)$

H=[1 1 ... 1]^T (7)

Instantaneous error ξ₀K () is used for monitoring whether the convergence of constant mould blind balance method enters stable state, is defined as follows:

${\mathrm{\ξ}}_0\left(k\right)=h\⊗e_s\left(k\right)---\left(8\right)$

Set up thresholding ρ_M, when following formula is set up, then claim constant mould blind balance method to be introduced into stable state；Otherwise, then constant mould is claimed Blind balance method entrance stable state:

ξ₀(k) ＞ ρ_M (9)。

Described equalizer coefficients delay inequality f_d(k) be:

$f_d\left(k\right)=\frac{1}{N_p}\underset{k=1}{\overset{N_p}{\Σ}}\left|\right|f_{\frac{N_f}{2}-1}\left(k\right)|-|f_{\frac{N_f}{2}-1}(k-d)\left|\right|$

Wherein, d express time postpones, | | represent that modulus value is asked for, N_fFor weight vector length, N_pRepresent iterations；

Monitoring weight coefficientWithDifference, if following formula set up, then it is assumed that constant mould is blind Equalization methods entrance stable state:

$f_d\left(k\right)=\frac{1}{N_p}\underset{k=1}{\overset{N_p}{\&Sigma;}}\left|\right|f_{\frac{N_f}{2}-1}\left(k\right)|-|f_{\frac{N_f}{2}-1}(k-d)\left|\right|<{\mathrm{\&rho;}}_f---\left(10\right)$

Wherein, ρ_fRepresent that equalizer coefficients delay inequality judges thresholding.

Described equalizer coefficients ratio is:

$f_r\left(k\right)=\frac{\left|f_{\frac{N_f}{2}}\left(k\right)\right|}{\left|f_0\left(k\right)\right|}$

Wherein, f_rThe weight coefficient ratio in k moment in (k) expression, | f₀(k) | represent first of equalizer coefficient vector Coefficient

Work as equalizer convergence, then weight vector both sides broad in the middle are little, and the initialization of equalizer is to use centre cap to initialize Method, the so-called initialized method of centre cap, refer to the weight vector of initial assignment, at the coefficient near center be 1, remaining coefficient is all zero, relativelys close to the weight vector coefficient at centerWith the coefficient near side | f₀(k) |, If set up with lower inequality, then it is assumed that big by paracentral weight coefficient, and the weight coefficient near edge is little:

$f_r\left(k\right)=\frac{\left|f_{\frac{N_f}{2}}\left(k\right)\right|}{\left|f_0\left(k\right)\right|}>{\mathrm{\&rho;}}_T---\left(11\right)$

Wherein, ρ_TRepresent that equalizer coefficients ratio judges thresholding.

Described equalizer output signal modulus value is:

$s_o=\frac{1}{N_p}\underset{k=1}{\overset{N_p}{\&Sigma;}}\left|y_i\left(k\right)\right|$

Wherein, s_oRepresent signal modulus value average result, N_pRepresent iterations；

Parameter ρ is set₀As i-th channel equalizer output signal y_iK the amplitude of () judges, take N_pPoint data calculates defeated Go out the average of signal modulus value:

$s_o=\frac{1}{N_p}\underset{k=1}{\overset{N_p}{\&Sigma;}}\left|y_i\left(k\right)\right|>{\mathrm{\&rho;}}_0---\left(12\right)$

Wherein, ρ₀Represent that modulus value judges thresholding, if above formula is set up, then it is assumed that output signal is qualified.

In described step 2, it is judged that the convergence process of blind equalizer, particularly as follows: by instantaneous error, equalizer coefficients time delay Difference, equalizer coefficients ratio and equalizer output signal modulus value build cognitive describing word RDW, this cognition describing word determine equilibrium Device coefficient vector use constant mould blind balance method gained weight coefficient vector still use delay equalizer coefficient vector, described in recognize Know that describing word RDW is:

RDW=[ξ₀(k),f_d(k),s_o,f_r(k)] (13)

Work as ξ₀(k)≤ρ_M, quantifying decision value is 1, otherwise is 0；

Work as f_d(k) ＜ ρ_f, quantifying decision value is 1, otherwise is 0；

Work as s_o＞ ρ_o, quantifying decision value is 1, otherwise is 0；

Work as f_r(k) ＞ ρ_T, quantifying decision value is 1, otherwise is 0.

Being all 1 when quantifying decision value, then it is assumed that equalizer can be used, equalizer coefficient vector uses constant mould blind balance method Gained weight coefficient vector；Otherwise, one or more 0 is had to exist, then it is assumed that equalizer is unavailable, when equalizer coefficient vector uses Prolong equalizer coefficient vector.

Beneficial effects of the present invention:

The present invention proposes a kind of cognitive blind balance method for radar channel amplitude Yu phase only pupil filter, by blind for constant mould all Weighing apparatus technology is applied to radar channel-equalization, it is achieved digital array radar closed loop equilibrium in a digital transmitting-receiving subassembly, it is not necessary to number According to passing down and equalizing coefficient passback；Meanwhile, introduce the cognitive techniques monitored for mean square error and weight vector, promote constant mould blind The convergence reliability of equalization algorithm, thus ensure the reliability of digital array radar hundreds of radar channel equalizer.

Accompanying drawing explanation

Fig. 1 is the digital transmitting and receiving modular construction schematic diagram that the present invention applies.

Fig. 2 is the cognitive blind balance method schematic diagram of the present invention.

Fig. 3 (a)-(b) is respectively amplitude difference and the phase contrast schematic diagram of two passages of a kind of embodiment of the present invention.

Fig. 4 (a)-(b) is respectively after equilibrium two interchannel amplitude differences and phase contrast schematic diagram in Fig. 3.

The parameter convergence situation schematic diagram that Fig. 5 (a)-(d) is respectively in RDW.

Fig. 6 (a)-(d) is respectively when inputting data and there is error code mistake, the parameter convergence situation schematic diagram in RDW.

Fig. 7 is when inputting data and there is error code mistake, two channel amplitude error schematic diagrams after equilibrium.

Detailed description of the invention

In order to make the purpose of the present invention, technical scheme and advantage clearer, below in conjunction with embodiment, to the present invention It is further elaborated.Should be appreciated that specific embodiment described herein, only in order to explain the present invention, is not used to Limit the present invention.

Below in conjunction with the accompanying drawings the application principle of the present invention is explained in detail.

As it is shown in figure 1, a kind of digital transmitting and receiving assembly includes: down going channel module and data feedback channel module；Described descending logical Low noise amplification module that road module includes being sequentially connected, the first frequency mixer, the first band filter, the first amplifying circuit, A/D Transducer, Digital Down Converter Module, blind equalizer and data packing block；Described data feedback channel module includes the number being sequentially connected Word synthesis module, the second amplifying circuit, the second band filter, the first frequency mixer and power amplifier；Described digit synthesis mould Block electrically connects with data packing block.

As in figure 2 it is shown, a kind of cognitive blind balance method for radar channel amplitude Yu phase only pupil filter, including following step Rapid:

Each digital transmitting and receiving group that step one, high-ranking officers' calibration signal are fed in digital array radar through power divider and feedback net The front end of part, and be sent in cognitive blind equalizer as input signal；

Step 2, the output signal of cognitive blind equalizer is sent to select switch, monitoring and blind equalizer convergence property Relevant parameter, it is judged that the convergence process of blind equalizer, and perform following operation:

A: when the convergence process of cognitive blind equalizer is normal, selects switch this cognitive blind equalization output signal to be transferred to Nonlinear transformation module carries out nonlinear transformation, it is thus achieved that adaptive error, and it is blind all to utilize this adaptive error to feed back to cognition The input of weighing apparatus, it is achieved the self-adaptative adjustment of blind equalizer equalizing coefficient vector, the equilibrium system of the cognitive blind equalization after convergence Number vector is for realizing the equilibrium treatment to input signal, and final output is through the signal of equilibrium treatment；

B: when the output of cognitive blind equalizer cannot restrain, selects switch this cognitive blind equalization output signal to be transferred to Time delay module, time delay module is delayed output when input signal being carried out, is not carried out equilibrium treatment.In the present invention, time delay module Being exactly Time-Delay Filter coefficient vector, because if having employed the equalizer coefficient vector of blind equalization, then will produce time delay δ, because of This, when being switched to time delay module, if time delay to be kept is constant, then time delay module also to arrange corresponding time delay δ, namely The δ the coefficient saying time delay module (i.e. time delay filter coefficient) is 1, and remaining coefficient is 0.

The parameter that described and cognitive blind equalizer convergence property is relevant, including: instantaneous error, equalizer coefficients delay inequality, Equalizer coefficients ratio and equalizer output signal modulus value.

In the present invention, setting calibration signal as p (k), equalizer input is x_i(k), it is assumed that there are M passage needs blind Equilibrium treatment, then the equalizer output signal y of i-th passage_i(k) be:

$y_i\left(k\right)=f_i\left(k\right)\&CircleTimes;x_i\left(k\right)---\left(1\right)$

Wherein: f_iK () represents i-th N_f× 1 weight coefficient vector, N_fRepresent equalizer weight coefficient vector length,Represent volume Long-pending computing；Definition coefficient of kurtosis γ:

$\mathrm{\&gamma;}=\frac{E\left\{\right|p\left(k\right){|}^4\}}{E\left\{\right|p\left(k\right){|}^2\}}---\left(2\right)$

E{ } represent and ask for mathematic expectaion, | | representing magnitude calculation, output signal is:

e_i(k)=y_i(k)[γ-|y_i(k)|²] (3)

The input recursive vector of definition i-th equalizer is:

X_i(k)=[x_i(k) x_i(k-1) … x_i(k-N_f+1)]^T (4)

Then weight coefficient vector is iterative is expressed as:

f_i(k+1)=f_i(k)+μ₁·X_i ^*(k)e(k) (5)

In formula, μ₁Step factor, []^*Represent complex conjugate；

Described nonlinear block gained adaptive error is:

e_s(k)=(γ-| y_i(k)²)(γ-|y_i(k)|²) (6)

Define N simultaneously_h× 1 dimension wave filter:

H=[1 1 ... 1]^T (7)

Instantaneous error ξ₀K () is used for monitoring whether the convergence of constant mould blind balance method enters stable state, is defined as follows:

${\mathrm{\&xi;}}_0\left(k\right)=h\&CircleTimes;e_s\left(k\right)---\left(8\right)$

Set up thresholding ρ_M, when following formula is set up, then claim algorithm to be introduced into stable state；Otherwise, then title algorithm entrance stable state:

ξ₀(k) ＞ ρ_M (9)。

Described equalizer coefficients delay inequality f_d(k) be:

$f_d\left(k\right)=\frac{1}{N_p}\underset{k=1}{\overset{N_p}{\&Sigma;}}\left|\right|f_{\frac{N_f}{2}-1}\left(k\right)|-|f_{\frac{N_f}{2}-1}(k-d)\left|\right|$

Wherein, d express time postpones, | | represent that modulus value is asked for, N_fFor weight vector length, N_pRepresent iterations；

Monitoring weight coefficientWithDifference, if following formula set up, then it is assumed that constant mould is blind Equalization methods entrance stable state:

$f_d\left(k\right)=\frac{1}{N_p}\underset{k=1}{\overset{N_p}{\&Sigma;}}\left|\right|f_{\frac{N_f}{2}-1}\left(k\right)|-|f_{\frac{N_f}{2}-1}(k-d)\left|\right|<{\mathrm{\&rho;}}_f,---\left(10\right).$

Wherein, ρ_fRepresent that equalizer coefficients delay inequality judges thresholding.

Described equalizer coefficients ratio is:

$f_r\left(k\right)=\frac{\left|f_{\frac{N_f}{2}}\left(k\right)\right|}{\left|f_0\left(k\right)\right|}$

Wherein, f_rThe weight coefficient ratio in k moment in (k) expression, | f₀(k) | represent first of equalizer coefficient vector Coefficient；

Work as equalizer convergence, then weight vector both sides broad in the middle are little, and the initialization of equalizer is to use centre cap to initialize Method, the so-called initialized method of centre cap, refer to the weight vector of initial assignment, at the coefficient near center be 1, remaining coefficient is all zero, relativelys close to the weight vector coefficient at centerWith the coefficient near side | f₀(k) |, If set up with lower inequality, then it is assumed that big by paracentral weight coefficient, and the weight coefficient near edge is little:

$f_r\left(k\right)=\frac{\left|f_{\frac{N_f}{2}}\left(k\right)\right|}{\left|f_0\left(k\right)\right|}>{\mathrm{\&rho;}}_T---\left(11\right)$

Wherein, ρ_TRepresent that equalizer coefficients ratio judges thresholding.

Described equalizer output signal modulus value is:

$s_o=\frac{1}{N_p}\underset{k=1}{\overset{N_p}{\&Sigma;}}\left|y_i\left(k\right)\right|$

Wherein, s_oRepresent signal modulus value average result, N_pRepresent iterations；

Parameter ρ is set₀As i-th channel equalizer output signal y_iK the amplitude of () judges, take N_pPoint data calculates defeated Go out the average of signal modulus value:

$s_o=\frac{1}{N_p}\underset{k=1}{\overset{N_p}{\&Sigma;}}\left|y_i\left(k\right)\right|>{\mathrm{\&rho;}}_0---\left(12\right)$

Wherein, ρ₀Represent that modulus value judges thresholding, if above formula is set up, then it is assumed that output signal is qualified.

In described step 2, it is judged that the convergence process of blind equalizer, particularly as follows: by instantaneous error, equalizer coefficients time delay Difference, equalizer coefficients ratio and equalizer output signal modulus value build cognitive describing word RDW, this cognition describing word determine equilibrium Device coefficient vector use constant mould blind balance method gained weight coefficient vector still use delay equalizer coefficient vector, described in recognize Know that describing word RDW is:

RDW=[ξ₀(k),f_d(k),s_o,f_r(k)] (13)

Work as ξ₀(k)≤ρ_M, quantifying decision value is 1, otherwise is 0；

Work as f_d(k) ＜ ρ_f, quantifying decision value is 1, otherwise is 0；

Work as s_o＞ ρ_o, quantifying decision value is 1, otherwise is 0；

Work as f_r(k) ＞ ρ_T, quantifying decision value is 1, otherwise is 0.

Being all 1 when quantifying decision value, then it is assumed that equalizer can be used, equalizer coefficient vector uses constant mould blind balance method Gained weight coefficient vector；Otherwise, one or more 0 is had to exist, then it is assumed that equalizer is unavailable, when equalizer coefficient vector uses Prolong equalizer coefficient vector.

Embodiment 1

First be defined as follows phantom produce passage constant mould blind equalization checking need data:

The input signal of cognitive equalizer uses following formula to represent:

$x_i\left(k\right)=c_i\&CircleTimes;p\left(k\right)+{\mathrm{\&xi;}}_i\left(k\right)---\left(14\right)$

In formula: k is discrete time sampling, x_iK () is that i-th channel equalizer inputs data；P (k) is defeated for calibration signal Enter, for random phase modulation signal；For i-th channel discrete channel response, Q is that passage rings Answer the number of coefficient, ξ_iK () is the Gaussian noise of i-th passage,Represent convolution；

In emulation, the simulated conditions of constant blind equalization algorithm is as shown in table 1:

Table 1

As shown in Fig. 3 (a)-(b), two amplitudes and phase place is used all to there is the channel authentication blind balance method of bigger gap Passage capability for correcting, inter-channel level error 2dB (peak-to-peak value), phase error is about 10 ° (peak-to-peak values), the coefficient of two passages Setting is shown in Table 2, and the setting of cognitive parameter is shown in Table 3:

Table 2

Table 3

After equilibrium, two interchannel amplitudes and phase error are shown in shown in Fig. 4 (a)-(b).

Parameter convergence situation in cognitive describing word RDW is shown in shown in Fig. 5 (a)-(d), in the simulation RDW1=[1,1,1, 1], show that cognitive constant mould equalizer convergence is normal, can serve as equilibrium.

Embodiment 2

When input data exist the mistakes such as error code, then can affect the convergence capabilities of blind equalizer, thus cause equalizer not Available.The present embodiment uses the cognitive competence of the data verification blind equalization that there is error code, emulates in data the 4000th existence One error code.

Fig. 6 (a)-(d) gives the situation of change of parameter in cognitive describing word RDW.It can be seen that in the position that error code occurs Putting, dissipating occurs in equalizer, thus causes the sampled point moment at ED, and equalizer does not converges to stable state yet.Now RDW =[0,1,0,1].Therefore, this equalizer is defective.Equalizer incision pure time delay mode.Otherwise, the equalizer obtained is used Equilibrium, will obtain more severe channel characteristic, as it is shown in fig. 7, compared with the situation before equalizing with Fig. 3 (a), performance is more severe.

The ultimate principle of the present invention and principal character and advantages of the present invention have more than been shown and described.The technology of the industry Personnel, it should be appreciated that the present invention is not restricted to the described embodiments, simply illustrating this described in above-described embodiment and description The principle of invention, without departing from the spirit and scope of the present invention, the present invention also has various changes and modifications, and these become Change and improvement both falls within scope of the claimed invention.Claimed scope by appending claims and Equivalent defines.

Claims (7)

1. the cognitive blind balance method for radar channel amplitude with phase only pupil filter, it is characterised in that comprise the following steps:

Step one, high-ranking officers' calibration signal are through power divider and feedback each digital transmitting and receiving assembly of being fed in digital array radar of net Front end, and be sent in cognitive blind equalizer as input signal；

Step 2, it is sent to select switch by the output signal of cognitive blind equalizer, monitors relevant to blind equalizer convergence property Parameter, it is judged that the convergence process of blind equalizer, and perform following operation:

A: when the convergence process of cognitive blind equalizer is normal, selects switch that this cognitive blind equalization output signal is transferred to non-thread Property conversion module carry out nonlinear transformation, it is thus achieved that adaptive error, and utilize this adaptive error to feed back to cognitive blind equalizer Input, it is achieved blind equalizer equalizing coefficient vector self-adaptative adjustment, the equalizing coefficient of the cognitive blind equalization after convergence to Amount is for realizing the equilibrium treatment to input signal, and final output is through the signal of equilibrium treatment；

B: when the output of cognitive blind equalizer cannot restrain, selects switch that this cognitive blind equalization output signal is transferred to time delay Module, time delay module is delayed output when input signal being carried out, is not carried out equilibrium treatment.

A kind of cognitive blind balance method for radar channel amplitude Yu phase only pupil filter the most according to claim 1, it is special Levy and be: the parameter that described and cognitive blind equalizer convergence property is relevant, including: instantaneous error, equalizer coefficients delay inequality, all Weighing apparatus coefficient ratio and equalizer output signal modulus value.

A kind of cognitive blind balance method for radar channel amplitude Yu phase only pupil filter the most according to claim 2, it is special Levy and be: described nonlinear block gained adaptive error is:

e_s(k)=(γ-| y_i(k)|²)(γ-|y_i(k)|²) (6)

Definition N_h× 1 dimension wave filter, N_hFor the length of this wave filter, y_iK () is the equalizer output signal of i-th passage；γ is Coefficient of kurtosis；

$\mathrm{\&gamma;}=\frac{E\left\{\right|p\left(k\right){|}^4\}}{E\left\{\right|p\left(k\right){|}^2\}}---\left(2\right)$

H=[1 1 ... 1]^T (7)

Instantaneous error ξ₀K () is used for monitoring whether the convergence of constant mould blind balance method enters stable state, is defined as follows:

${\mathrm{\&xi;}}_0\left(k\right)=h\&CircleTimes;e_s\left(k\right)---\left(8\right)$

Set up thresholding ρ_M, when following formula is set up, then claim constant mould blind balance method to be introduced into stable state；Otherwise, then constant mould blind equalization is claimed Method entrance stable state:

ξ₀(k) ＞ ρ_M (9)。

A kind of cognitive blind balance method for radar channel amplitude Yu phase only pupil filter the most according to claim 3, it is special Levy and be: described equalizer coefficients delay inequality f_d(k) be:

$f_d\left(k\right)=\frac{1}{N_p}\underset{k=1}{\overset{N_p}{\&Sigma;}}\left|\right|f_{\frac{N_f}{2}-1}\left(k\right)|-|f_{\frac{N_f}{2}-1}(k-d)\left|\right|$

Wherein, d express time postpones, | | represent that modulus value is asked for, N_fFor weight vector length, N_pRepresent iterations；Monitoring power is NumberWithDifference, if following formula set up, then it is assumed that constant mould blind balance method enter stable state:

$f_d\left(k\right)=\frac{1}{N_p}\underset{k=1}{\overset{N_p}{\&Sigma;}}\left|\right|f_{\frac{N_f}{2}-1}\left(k\right)|-|f_{\frac{N_f}{2}-1}(k-d)\left|\right|<{\mathrm{\&rho;}}_f---\left(10\right)$

Wherein, ρ_fRepresent that equalizer coefficients delay inequality judges thresholding.

A kind of cognitive blind balance method for radar channel amplitude Yu phase only pupil filter the most according to claim 4, it is special Levy and be: described equalizer coefficients ratio is:

$f_r\left(k\right)=\frac{\left|f_{\frac{N_f}{2}}\left(k\right)\right|}{\left|f_0\left(k\right)\right|}$

Wherein, f_rThe weight coefficient ratio in k moment in (k) expression, | f₀(k) | represent first coefficient of equalizer coefficient vector

Work as equalizer convergence, then weight vector both sides broad in the middle are little, and the initialization of equalizer is to use the initialized side of centre cap Method, the so-called initialized method of centre cap, refer to the weight vector of initial assignment, be 1 at the coefficient near center, its Remaining coefficient is all zero, relativelys close to the weight vector coefficient at centerWith the coefficient near side | f₀(k) |, if Set up with lower inequality, then it is assumed that big by paracentral weight coefficient, and the weight coefficient near edge be little:

$f_r\left(k\right)=\frac{\left|f_{\frac{N_f}{2}}\left(k\right)\right|}{\left|f_0\left(k\right)\right|}>{\mathrm{\&rho;}}_T---\left(11\right)$

Wherein, ρ_TRepresent that equalizer coefficients ratio judges thresholding.

A kind of cognitive blind balance method for radar channel amplitude Yu phase only pupil filter the most according to claim 5, it is special Levy and be: described equalizer output signal modulus value is:

$s_o=\frac{1}{N_p}\underset{k=1}{\overset{N_p}{\&Sigma;}}\left|y_i\left(k\right)\right|$

Wherein, s_oRepresent signal modulus value average result, N_pRepresent iterations；

Parameter ρ is set₀As i-th channel equalizer output signal y_iK the amplitude of () judges, take N_pPoint data calculates output letter The average of number modulus value:

$s_o=\frac{1}{N_p}\underset{k=1}{\overset{N_p}{\&Sigma;}}\left|y_i\left(k\right)\right|>{\mathrm{\&rho;}}_0---\left(12\right)$

Wherein, ρ₀Represent that modulus value judges thresholding, if above formula is set up, then it is assumed that output signal is qualified.

A kind of cognitive blind balance method for radar channel amplitude Yu phase only pupil filter the most according to claim 6, it is special Levy and be: in described step 2, it is judged that the convergence process of blind equalizer, particularly as follows: by instantaneous error, equalizer coefficients time delay Difference, equalizer coefficients ratio and equalizer output signal modulus value build cognitive describing word RDW, this cognition describing word determine equilibrium Device coefficient vector use constant mould blind balance method gained weight coefficient vector still use delay equalizer coefficient vector, described in recognize Know that describing word RDW is:

RDW=[ξ₀(k),f_d(k),s_o,f_r(k)] (13)

Work as ξ₀(k)≤ρ_M, quantifying decision value is 1, otherwise is 0；

Work as f_d(k) ＜ ρ_f, quantifying decision value is 1, otherwise is 0；

Work as s_o＞ ρ_o, quantifying decision value is 1, otherwise is 0；

Work as f_r(k) ＞ ρ_T, quantifying decision value is 1, otherwise is 0.

Being all 1 when quantifying decision value, then it is assumed that equalizer can be used, equalizer coefficient vector uses constant mould blind balance method gained Weight coefficient vector；Otherwise, having one or more 0 to exist, then it is assumed that equalizer is unavailable, equalizer coefficient vector uses time delay equal Weighing apparatus coefficient vector.