CN103336269B - A kind of cognitive waveform optimization disposal route for multi-emitting GMTI radar system - Google Patents

Abstract

The invention discloses a kind of cognitive waveform optimization disposal route for multi-emitting GMTI radar system, range unit to be detected is set to wanted signal, and the information of its neighbor distance unit is seen as undesired signal.Based on an initial pulse pressure filter weights, obtain initial time domain response Output rusults, then according to prior imformation, Human disturbance is added to its secondary lobe district, power again by controlling Human disturbance carrys out the weights of optimal design pulse pressure wave filter next time, the time domain response figure that mate inverse with environment is obtained after successive ignition like this, namely the place of stronger noise jamming formed low sidelobe, more weak or without noise jamming place formed higher secondary lobe.Wave filter designed by the present invention with the template of environmentally priori design for reference, at strong clutter district design low sidelobe, weak or design high secondary lobe without clutter district.Therefore it has adaptive ability, can clutter reduction better.

Description

A kind of cognitive waveform optimization disposal route for multi-emitting GMTI radar system

Technical field

The invention belongs to Radar Technology field, relate to one in this field based on the adaptive pulse compression filter method for designing of priori environment knowledge, the present invention can be applied to random waveform process of pulse-compression of mating inverse with any environment, is separated the pulse compression information of each waveform under also can be applicable to multi-emitting waveform situation.Wave filter is with reference to the inverse stencil design of mate of environment, therefore, it is possible to clutter reduction thus the detection perform of raising target better.

Background technology

Cognitive system is by the study for target and environment, by the information feed back that extracts from echo to transmitter, form closed-loop system, then transmitter environmentally new the transmitting of information self-adapting optimal design, makes it conform and reaches the object improving radar performance.Environmental information is mainly derived from: the echo after radar emission signal reflects in space environment, the environment distributed intelligence that the knowledge of the enemy radar reconnaissance equipment obtained based on our relevant device and other sensors obtain.In the cognitive radar system of multi-emitting GMTI, suppress the echo from static target or ground clutter more easily could detect ground moving object.Multichannel clutter recognition performance is relevant with the consistance of passage.Therefore, in the cognitive radar system of multi-emitting GMTI, not only require each passage (pulse compression result) and the inverse coupling of environmental information, also require, between the isolated multiple passage of pulse pressure, there is high related coefficient.

Traditional pulse compression filter does not have adaptive ability to environment, and under complex electromagnetic environment background, target detection capabilities often declines.Existing least square method can the equally distributed secondary lobe of optimal design, but due to energy constant, equally distributed sidelobe level has theoretical boundary, therefore not necessarily can obtain desirable result.And actual environment information is often fast changing and be not equally distributed, with time to complicate the adaptive Waveform Design of electromagnetic environment and process more can realistic demand.

Summary of the invention

The object of the invention is to overcome above-mentioned the deficiencies in the prior art, propose a kind of with time changing environment adaptive pulse compression filter method for designing.The method with the template of environmentally priori design for reference, the wave filter that mate inverse with environment is designed by iterative processing, improve the correlativity between hyperchannel, thus the problem that under improving clutter and jamming pattern, target not easily detects, improve the detection perform of target.

Basic ideas of the present invention are: the echo data first receiving multi-emitting waveform; Cognition is carried out to environment, designs the pulse compression filter time domain response figure that mate inverse with environment, it can be used as reference template; Then for the pulse compression filter that random waveform design matches with reference template; With the filter process list designed/multi-emitting waveform echo data, isolate the pulse compression information of each waveform, export from different passages; Finally the output data of each passage are disappeared process mutually, complete the signal transacting such as clutter recognition.

For a cognitive waveform optimization disposal route for multi-emitting GMTI radar system, comprise the steps:

(1) echo data of multi-emitting waveform is received;

(2) cognition is carried out to environment, design the pulse compression reference template that mate inverse with environment;

(3) with reference to the inverse random waveform pulse compression filter mated of reference template design and environment;

3a) initiation parameter: retrain main lobe region, expect binding occurrence g, the initial weighting coefficients matrix W of information matrix C and correspondence, iterations K;

3b) calculate weighting time domain correlation matrix B;

3c) calculate filter weights h;

3d) the pulse pressure result obtained and reference template are contrasted.If judge, whether pulse pressure result mates with template matches, and stop, design of filter is complete; If do not mate, then the minimum peak level Pr finding out secondary lobe, as with reference to level, upgrades weighting coefficient matrix W, turns back to 3b) iteration K time is until complete the design of wave filter after meeting the demands;

(4) when multi-emitting waveform echo data being processed with the pulse compression filter that (3) are designed, obtain each waveform in multi-emitting waveform for multiple separation and carry out pulse pressure, ensure that there is similar main sidelobe performance as far as possible, namely require to adopt same pulse pressure reference template to different transmitted waveforms;

Have M transmitted waveform in multi-emitting waveform, the constraint matrix of each waveform is , use replace the C in random waveform design of filter in (3);

(5) clutter recognition is completed to doing the process that disappears mutually between the multi-emitting waveform pulse pressure data obtained.

Described optimized treatment method, step (3) specifically performs following steps:

First initial parameter is set: given main lobe region Ω, expectation information matrix C=span (x _i) i ∈ Ω and corresponding binding occurrence g; Wherein x _ibe i-th column vector receiving echo data matrix, span () represents the vector opened by bracket interior element; If the main lobe scope of constraint only comprises a main peak point, so C=x ₀and g=1; Initial weighting coefficients matrix W is set; The initial value of the Human disturbance power in secondary lobe district is 1, and the initial value of main lobe district Human disturbance power is 0, i.e. the w (i)=0 as i ∈ Ω; When time w (i)=1, iteration coefficient λ is a constant;

Then initial weighting time domain correlation matrix B is calculated:

B=XWX ^*(1)

Wherein () ^*for transpose of a matrix in bracket;

By solving the optimization problem of belt restraining $\left\{\begin{array}{c}\underset{h}{\min }h*\mathrm{Bh}\\ s.t{C}^{*}h=g\end{array}\right.$ Draw the best initial weights h of wave filter :

h=B ^-1C(C ^*B ^-1C) ^-1g(2)

Wherein, represent with h to be the minimum value that variable gets bracket interior element, s.t represents constraint condition symbol, () ^-1for the inverse matrix of matrix in bracket;

Calculate time domain response y:

y=X ^*h （3）

Then the pulse compression result obtained and reference template are contrasted; If with template matches, stop, design of filter is complete; If do not mate, then find the minimum peak level Pr of secondary lobe, it can be used as datum; The position of Pr is also change in an iterative process, find by the following method: level value adjacent with its left and right for the level value of a certain range unit is compared, if its level value is greater than the level value of its left and right neighbor distance unit, then it is peak level, and one minimum in all peak levels is minimum peak level Pr; Compared with the secondary lobe of reference template, should increase at its weighting coefficient of secondary lobe district higher than reference template secondary lobe, and should decline lower than its weighting coefficient of secondary lobe region with reference to sidelobe level; Upgrade weighting coefficient matrix W,

$w\left(i\right)=\left\{\begin{array}{cc}0& i\∈\mathrm{\Ω}\\ \max \{w\left(i\right)+\frac{\mathrm{\λw}\left(i\right)\left[\right|y\left(i\right)|-\mathrm{Pr\; D}\left(i\right)]}{\mathrm{Pr\; D}\left(i\right)},0\}& i\∉\mathrm{\Ω}\end{array}\right.---\left(4\right)$

Wherein, max () represents the maximal value of getting bracket interior element, | () | represent the absolute value getting bracket interior element, Pr D (i) is the degree reference template linear-scale that Time-Domain Pulse Compression responds being transformed to minimum peak level, is called absolute level;

Finally recalculate time domain correlation matrix B according to the W after upgrading, iteration K time, until the response of wave filter and template matches, completes the design of wave filter; The object of iteration is the maximal value of the difference minimizing actual sidelobe level and expect reference template sidelobe level.

Described optimized treatment method, step (4) specifically performs following steps: establish in multi-emitting waveform and have M transmitted waveform, the input matrix of echoed signal with weighting time domain correlation matrix and the constraint matrix corresponding to m transmitted waveform is expressed as:

$\tilde{X}=\underset{m=1}{\overset{M}{\mathrm{\Σ}}}{X}^m,{\tilde{C}}^m=\mathrm{span}\left(x_i^m\right),i\∈\mathrm{\Ω}---\left(5\right)$

Wherein, represent the summation of bracket interior element, X ^mthe echo data matrix corresponding to m transmitted waveform, for the constraint matrix in m transmitted waveform, x ^mi-th column vector; With x in replacement step 3 design of filter, same C uses replace, just separablely obtain pulse compression result corresponding to each transmitted waveform in multi-emitting waveform, export from different passages.

The present invention compared with prior art tool has the following advantages:

The first, the wave filter designed by the present invention is applicable to any environment and random waveform and places an order the process of pulse-compression of transmitted waveform, and from echo, isolate the process of pulse-compression of each transmitted waveform under being applicable to multi-emitting waveform situation.

The second, the wave filter designed by the present invention with the template of environmentally priori design for reference, at strong clutter district design low sidelobe, weak or design high secondary lobe without clutter district.Therefore it has adaptive ability, can clutter reduction better.

Accompanying drawing explanation

Fig. 1 is process flow diagram of the present invention;

Fig. 2 is the process flow diagram of random waveform pulse compression filter design;

Fig. 3 is the template of the pulse compression filter based on the design of environment priori;

Fig. 4 is the first waveform pulse compression result that the first passage of multi-emitting waveform after receiving end is separated adopting the inventive method to obtain exports;

Fig. 5 is the second waveform pulse compression result that the second channel of multi-emitting waveform after receiving end is separated adopting the inventive method to obtain exports;

Fig. 6 is the 3rd waveform pulse compression result that the third channel of multi-emitting waveform after receiving end is separated adopting the inventive method to obtain exports;

Fig. 7 be the first and second multi-emitting channel pulse packed datas offset result;

Fig. 8 be second and the 3rd multi-emitting channel pulse packed data offset result;

Fig. 9 be first and the 3rd multi-emitting channel pulse packed data offset result;

Embodiment

Below in conjunction with specific embodiment, the present invention is described in detail.

See figures.1.and.2, specific embodiment of the invention step is as follows:

Step 1. receives the echo data X of multi-emitting waveform;

Step 2. pair environment carries out cognition, designs the pulse compression reference template D that mate inverse with environment;

Step 3. is with reference to the inverse random waveform pulse compression filter mated of reference template design and environment;

First initial parameter is set: given main lobe region Ω (comprising L range unit), expectation information matrix C=span (x _i) i ∈ Ω and corresponding binding occurrence g.Wherein x _ibe i-th column vector receiving echo data matrix, span () represents the vector opened by bracket interior element.If the main lobe scope of constraint only comprises a main peak point, so C=x ₀and g=1.Arrange initial weighting coefficients matrix W, in fact weighting coefficient is Human disturbance power.The initial value of the Human disturbance power in secondary lobe district is 1, and the initial value of main lobe district Human disturbance power is 0, i.e. the w (i)=0 as i ∈ Ω; When time w (i)=1.Iteration coefficient λ is a constant.

Then initial weighting time domain correlation matrix B is calculated:

B=XWX ^*(1)

Wherein () ^*for transpose of a matrix in bracket.

By solving the optimization problem of belt restraining $\left\{\begin{array}{c}\underset{h}{\min }h*\mathrm{Bh}\\ s.t{C}^{*}h=g\end{array}\right.$ Draw the best initial weights h of wave filter :

h=B ^-1C(C ^*B ^-1C) ^-1g(2)

Wherein, represent with h to be the minimum value that variable gets bracket interior element, s.t represents constraint condition symbol, () ^-1for the inverse matrix of matrix in bracket.

Calculate time domain response y:

y=X ^*h （3）

Then the pulse compression result obtained and reference template are contrasted.If with template matches, stop, design of filter is complete; If do not mate, then find the minimum peak level Pr of secondary lobe, it can be used as datum.It is to be noted that the position of Pr is also change in an iterative process, can find by the following method: level value adjacent with its left and right for the level value of a certain range unit is compared, if its level value is greater than the level value of its left and right neighbor distance unit, then it is peak level, and one minimum in all peak levels is minimum peak level Pr.Compared with the secondary lobe of reference template, should increase at its weighting coefficient of secondary lobe district higher than reference template secondary lobe, and should decline lower than its weighting coefficient of secondary lobe region with reference to sidelobe level.Upgrade weighting coefficient matrix W,

$w\left(i\right)=\left\{\begin{array}{cc}0& i\∈\mathrm{\Ω}\\ \max \{w\left(i\right)+\frac{\mathrm{\λw}\left(i\right)\left[\right|y\left(i\right)|-\mathrm{Pr\; D}\left(i\right)]}{\mathrm{Pr\; D}\left(i\right)},0\}& i\∉\mathrm{\Ω}\end{array}\right.---\left(4\right)$

Wherein, max () represents the maximal value of getting bracket interior element, | () | represent the absolute value getting bracket interior element, Pr D (i) is the degree reference template linear-scale that Time-Domain Pulse Compression responds being transformed to minimum peak level, is called absolute level.

Finally recalculate time domain correlation matrix B according to the W after upgrading, iteration K time, until the response of wave filter and template matches, completes the design of wave filter.The object of iteration is the maximal value of the difference minimizing actual sidelobe level and expect reference template sidelobe level.

When step 4. processes multi-emitting waveform echo data with the pulse compression filter that step 3 designs, obtain each waveform in multi-emitting waveform for multiple separation and carry out pulse pressure, ensure that there is similar main sidelobe performance as far as possible, namely require to adopt same pulse pressure reference template to different transmitted waveforms;

If have M transmitted waveform in multi-emitting waveform, the input matrix of echoed signal with weighting time domain correlation matrix and the constraint matrix corresponding to m transmitted waveform is expressed as:

$\tilde{X}=\underset{m=1}{\overset{M}{\mathrm{\Σ}}}{X}^m,{\tilde{C}}^m=\mathrm{span}\left(x_i^m\right),i\∈\mathrm{\Ω}---\left(5\right)$

Wherein, represent the summation of bracket interior element, X ^mthe echo data matrix corresponding to m transmitted waveform, for the constraint matrix in m transmitted waveform, x ^mi-th column vector.With x in replacement step 3 design of filter, same C uses replace, just separablely obtain pulse compression result corresponding to each transmitted waveform in multi-emitting waveform, export from different passages.

Step 5. is done to the pulse pressure data that each passage exports each waveform the process that disappears mutually and is completed clutter recognition.

Effect of the present invention can be illustrated by following emulation experiment:

Simulated conditions

The multi-emitting waveform adopted be utilize one of Genetic Algorithm optimized design group of three code length be 150 orthogonal Coded Signals, multiple transmitting is launched by multiple emitting antenna simultaneously.The initial parameter of Optimal Filters Design is as follows: constraint main lobe scope Ω=[0], code length N=P=150, iterations is 30, iteration coefficient λ=0.1.

Simulation result

Fig. 3 is the reference template of the pulse compression time domain response figure of the inverse coupling of a kind of environment.Based on the reference template of Fig. 3, adopt method design pulse compression filter of the present invention, Fig. 4 is the pulse compression result of first passage, and Fig. 5 is the pulse compression result of second passage, and Fig. 6 is the pulse compression result of the 3rd passage.In Fig. 4 to Fig. 6, horizontal ordinate represents range unit sequence number, and ordinate represents the pulse pressure Output rusults of different distance unit, and its unit is dB.As can be seen from simulation result, Output rusults and the template of pulse compression filter match, and demonstrate and adopt the pulse compression filter designed by the present invention that pulse compression result and actual environment can be made against mating.

Fig. 7 is the 1st, the result that disappears mutually of 2 passages, and Fig. 8 is the 2nd, the result that disappears mutually of 3 passages, and Fig. 9 is the 1st, the result that disappears mutually of 3 passages.Horizontal ordinate represents range unit sequence number, and ordinate represents the surplus after disappearing mutually.Can be seen by simulation result, utilize the present invention according to the wave filter of stencil design, offset after pulse compression, surplus is little, has good rejection ability to clutter, and then improves the detection perform of target.

Should be understood that, for those of ordinary skills, can be improved according to the above description or convert, and all these improve and convert the protection domain that all should belong to claims of the present invention.

Claims (1)

1., for a cognitive waveform optimization disposal route for multi-emitting GMTI radar system, it is characterized in that, comprise the steps:

(1) echo data of multi-emitting waveform is received;

(2) cognition is carried out to environment, design the pulse compression reference template that mate inverse with environment;

(3) with reference to the inverse random waveform pulse compression filter mated of reference template design and environment;

3a) initiation parameter: retrain main lobe region, expect binding occurrence g, the initial weighting coefficients matrix W of information matrix C and correspondence, iterations K;

3b) calculate weighting time domain correlation matrix B;

3c) calculate filter weights h;

3d) the pulse pressure result obtained and reference template are contrasted, judge that whether and template matches pulse pressure result, if coupling, stop, design of filter is complete; If do not mate, then the minimum peak level Pr finding out secondary lobe, as with reference to level, upgrades weighting coefficient matrix W, turns back to 3b) iteration K time is until complete the design of wave filter after meeting the demands;

(4) when multi-emitting waveform echo data being processed with the pulse compression filter that (3) are designed, obtain each waveform in multi-emitting waveform for multiple separation and carry out pulse pressure, ensure that there is similar main sidelobe performance as far as possible, namely require to adopt same pulse pressure reference template to different transmitted waveforms;

Have M transmitted waveform in multi-emitting waveform, the constraint matrix of each waveform is with replace the C in random waveform design of filter in (3);

(5) clutter recognition is completed to doing the process that disappears mutually between the multi-emitting waveform pulse pressure data obtained;

Step (3) specifically performs following steps:

First initial parameter is set: given main lobe region Ω, expectation information matrix C=span (x _i) i ∈ Ω and corresponding binding occurrence g; Wherein x _ibe i-th column vector receiving echo data matrix, span () represents the vector opened by bracket interior element; If the main lobe scope of constraint only comprises a main peak point, so C=x ₀and g=1; Initial weighting coefficients matrix W is set; The initial value of the Human disturbance power in secondary lobe district is 1, and the initial value of main lobe district Human disturbance power is 0, i.e. the w (i)=0 as i ∈ Ω; When time w (i)=1, iteration coefficient λ is a constant; W (i) represents the initial value of Human disturbance power;

Then initial weighting time domain correlation matrix B is calculated:

B＝XWX ^*(1)

Wherein () ^*for transpose of a matrix in bracket; X represents echo data;

By solving the optimization problem of belt restraining $\left\{\begin{array}{c}\underset{h}{\min }{h}^{*}\mathrm{Bh}\\ s.t{C}^{*}h=g\end{array}\right.$ Draw the best initial weights h of wave filter:

h＝B ^-1C(C ^*B ^-1C) ^-1g (2)

Wherein, represent with h to be the minimum value that variable gets bracket interior element, s.t represents constraint condition symbol, () ^-1for the inverse matrix of matrix in bracket;

Calculate time domain response y:

y＝X ^*h (3)

Then the pulse compression result obtained and reference template are contrasted; If with template matches, stop, design of filter is complete; If do not mate, then find the minimum peak level Pr of secondary lobe, it can be used as datum; The position of Pr is also change in an iterative process, find by the following method: level value adjacent with its left and right for the level value of a certain range unit is compared, if its level value is greater than the level value of its left and right neighbor distance unit, then it is peak level, and one minimum in all peak levels is minimum peak level Pr; Compared with the secondary lobe of reference template, should increase at its weighting coefficient of secondary lobe district higher than reference template secondary lobe, and should decline lower than its weighting coefficient of secondary lobe region with reference to sidelobe level; Upgrade weighting coefficient matrix W,

$w\left(i\right)=\left\{\begin{array}{cc}0& i\∈\mathrm{\Ω}\\ \max \{w\left(i\right)+\frac{\mathrm{\λw}\left(i\right)\left[\right|y\left(i\right)|-\mathrm{PrD}\left(i\right)]}{\mathrm{PrD}\left(i\right)},0\}& i\∉\mathrm{\Ω}\end{array}\right.---\left(4\right)$

Wherein, max () represents the maximal value of getting bracket interior element, | () | represent the absolute value getting bracket interior element, PrD (i) is the degree reference template linear-scale that Time-Domain Pulse Compression responds being transformed to minimum peak level, is called absolute level;

Finally recalculate time domain correlation matrix B according to the W after upgrading, iteration K time, until the response of wave filter and template matches, completes the design of wave filter; The object of iteration is the maximal value of the difference minimizing actual sidelobe level and expect reference template sidelobe level;

Step (4) specifically performs following steps: establish in multi-emitting waveform and have M transmitted waveform, the input matrix of echoed signal and the constraint matrix corresponding to m transmitted waveform is expressed as:

$\tilde{X}=\underset{m=1}{\overset{M}{\mathrm{\Σ}}}{X}^m,{\tilde{C}}^m=\mathrm{span}\left(x_i^m\right),i\∈\mathrm{\Ω}---\left(5\right)$

Wherein, represent the summation of bracket interior element, X ^mthe echo data matrix corresponding to m transmitted waveform, the constraint matrix corresponding to m transmitted waveform, x ^mi-th column vector; With x in replacement step (3) design of filter, same C uses replace, just separablely obtain pulse compression result corresponding to each transmitted waveform in multi-emitting waveform, export from different passages.