CN105974400A - Robust space-time detection method based on symmetric spectral characteristic - Google Patents

Abstract

The invention relates to the signal detection field and especially relates to a robust space-time detection method based on a symmetric spectral characteristic. In one embodiment, the method comprises the following steps of using a sonar to receive echo data and target detection under a reverberation background condition is concluded into a binary hypothesis test; based on the symmetric spectral characteristic, converting the binary hypothesis test into an equivalent actual value binary composite hypothesis test; based on the actual value binary composite hypothesis test, acquiring a joint probability density function of main data and auxiliary data in the echo data under the condition that there is a target; and based on the joint probability density function, acquiring a spectrum symmetry RAO detection statistics amount so as to complete detection of the target. In the method of the invention, a symmetry characteristic of a reverberation power spectrum is considered during designing and is merged into the detection statistics amount so that effective utilization to prior information is realized and detection performance of space-time adaptive detection (STAD) under a non-homogeneous environment is greatly increased.

Description

A kind of sane space-time detecting method based on symmetric spectral characteristic

Technical field

The present invention relates to signal detection field, especially design a kind of sane space-time detecting method based on symmetric spectral characteristic.

Background technology

Reverberation is that the inhomogeneities scattering of random distribution in seabed, the irregularities on sea and sea water causes, and is torpedo One of the main reasons Deng the interference of motion sonar.Be different from seanoise, reverberation by launch signal cause, its spectrum structure with Launching signal and have certain similarity, and motion sonar wave beams touches interface, reverberation, by incident from different cone angles, causes The extension of Doppler, it is impossible to use platform displacement compensation technique to be completely eliminated.For the reverberation of this characteristic, Ying Lian Close and consider spatial domain and the suppressing method on frequency domain, and can adjust according to environment, here it is space-time adaptive processing method. Brennan in 1973 proposes space-time adaptive first and processes (STAP) concept, and proves that STAP can combine spatial domain well Advantage with Time Domain Processing, it is thus achieved that ideal clutter recognition performance.Reed etc. propose sample covariance matrix and ask subsequently Inverse (SMI) detector, thus in theory STAP is developed into a kind of filtering and the method for detection combination, when being referred to as empty Self-adapting detecting (STAD) method.It is integrated with target detection that STAD achieves Reverberation Rejection, detects after first Reverberation Rejection STAP method compare, STAD more can effectively utilize observation data, and detection performance is more excellent.

In recent years, the STAD research in motion sonar field is the most active, especially puts mesh under Gauss distribution Reverberation Target STAD test problems, many classical detection methods can be applied, including maximum likelihood ratio inspection (GLRT), certainly Adapt to matched filtering device (AMF) and adaptive coherent estimator (ACE) etc..What deserves to be explained is, these traditional methods are based on one Individual critically important hypotheses, it is simply that can obtain abundant uniform assistance data, in order to estimate to-be-measured cell (master data) Reverberation covariance matrix, construct self-adapting detecting statistic.The uniformity of assistance data refers to have identical with master data Reverberation covariance matrix, for guaranteeing this characteristic, assistance data is generally from the distance unit selection closed on master data.

For single base sonar, the power spectral density of its reverberation is generally of the symmetry characteristic centered by zero Doppler, But this priori is not the most used by existing detection method.And in actual applications, motion sonar is often Being operated in non-homogeneous environment, the most obtainable uniform assistance data length is by very limited.Both the above reason causes existing The detection performance of method will decline to a great extent, and how improve the robustness of STAD under non-homogeneous environment, always motion sonar work The difficult problem that person is urgently to be resolved hurrily.

For solving this problem, the present invention proposes a kind of sane space-time detecting method utilizing priori, the elder generation utilized Test the symmetry that knowledge is exactly reverberation power spectrum.The inventive method the most just considers this characteristic, is incorporated detection Statistic, it is achieved the effective utilization to this prior information, is greatly improved the detection performance of STAD under non-homogeneous environment.

Summary of the invention

It is an object of the invention to realize the efficient utilization of this prior information of symmetry to reverberation power spectrum, be effectively improved Motion sonar performance under non-homogeneous background.

For achieving the above object, a kind of sane space-time detection side utilizing symmetric spectral characteristic is embodiments provided Method, step is as follows:

It is received back to wave datum by sonar, and the target detection under the conditions of Reverberation is summarized as binary hypothesis test asks Topic；Based on symmetric spectral characteristic, described binary hypothesis test is converted into the real-valued binary composite hypothesis inspection of equivalence；Based on described Real-valued binary composite hypothesis is checked, it is thus achieved that having under target conditions, in described echo data, the associating of master data and assistance data is general Rate density function；Based on described joint probability density function, obtain composing symmetrical RAO detection statistic, in order to complete target Detection.

Preferably, described it is received back to wave datum by sonar, and the target detection under the conditions of Reverberation is summarized as binary Hypothesis testing, including: the linear array being made up of N number of array element receives the echo data of K length, and by the mesh under the conditions of Reverberation Mark detection is summarized as binary hypothesis test:

$\left\{\begin{array}{cc}{H}_0:\left\{\begin{array}{c}z=n,\\ z_k=n_k,\end{array}\right.& k=1,\mathrm{...},K\\ H_1:\left\{\begin{array}{c}z=\mathrm{\α}p+n,\\ z_k=n_k,\end{array}\right.& k=1,\mathrm{...},K\end{array}\right.$

Wherein, H₀And H₁Represent driftlessness respectively to assume and assume with the presence of target；P=p₁+jp₂It it is the target nominal of N-dimensional Guiding vector, p₁And p₂It is respectively its real part and imaginary part；α=α₁+jα₂It is the echo signal amplitude received, determines ginseng for the unknown Number, α₁And α₂It is its real part and imaginary part respectively；N=n₁+jn₂, n_k=n_1k+jn_2kIt is that independent identically distributed, zero-mean N-dimensional is combined Gaussian reverberation vector, n_1kAnd n_2kIt is respectively real part and imaginary part；Z represents cell data to be detected, also known as master data；z_kRepresent length Uniform assistance data for K.

Preferably, described based on symmetric spectral characteristic, the real-valued binary that described binary hypothesis test is converted into equivalence is combined Hypothesis testing, including: according to the symmetry characteristic of reverberation power spectrum, described binary hypothesis test is converted into the real-valued binary of equivalence Composite hypothesis is checked:

$\left\{\begin{array}{cc}{H}_0:\left\{\begin{array}{c}{z}_1=n_1,z_2=n_2,\\ z_{1k}=n_{1k},z_{2k}=n_{2k}\end{array}\right.& k=1,\mathrm{...},K\\ H_1:\left\{\begin{array}{c}{z}_1=({\mathrm{\α}}_1{p}_1-{\mathrm{\α}}_2{p}_2)+n_1,\\ z_{1k}=n_{1k},z_{2k}=n_{2k},\end{array}\right.& \begin{array}{c}{z}_2=({\mathrm{\α}}_1{p}_2-{\mathrm{\α}}_2{p}_2)+n_2,\\ k=1,\mathrm{...},K\end{array}\end{array}\right.$

Wherein, H₀And H₁Represent driftlessness respectively to assume and assume with the presence of target；p₁And p₂It is respectively target nominal to guide The real part of vector and imaginary part；α₁And α₂It is real part and the imaginary part of the echo signal amplitude received respectively；n₁、n₂、n_1kAnd n_2kIt is independent With distribution, zero-mean N-dimensional complex Gaussian reverberation vector；z₁And z₂Represent cell data to be detected, also known as master data；z_1kAnd z_2k Represent the uniform assistance data of a length of K.

Preferably, described based on the inspection of described real-valued binary composite hypothesis, it is thus achieved that to have under target conditions, described echo data Middle master data and the joint probability density function of assistance data, including:

Check based on described real-valued binary composite hypothesis, under having target conditions, master data Z ≡ [z₁,z₂] and assistance data Z_K=[z_l1..., z_1k, z₂₁..., z_2K] joint probability density function be:

$f(Z,Z_K|\mathrm{\θ},H_1)={\left(2\mathrm{\π}\right)}^{-N(K+1)}\det {\left(M\right)}^{-(K+1)}\exp \{-tr\[M^{-1}(u_1{u}_1^T+u_1{u}_1^T+S)\]\}$

Wherein,^TRepresent the transposition operation of matrix or vector；u₁=z₁-α₁p₁+α₂p₂, u₂=z₂-α₁p₂-α₂p₁； For sample covariance matrix based on assistance data；Det () and tr () represents determinant of a matrix and trace of a matrix respectively；θ It is the parameter column vector of N (N-1)/2+2 dimension, i.e.Wherein signal parameter vector θ_A=[α₁,α₂]^T, nuisance parameter arranges Vector θ_BBeing made up of the real-valued element of covariance matrix M, dimension is N (N-1)/2.

Preferably, described based on described joint probability density function, it is thus achieved that the symmetrical RAO method of spectrum, in order to complete target Detection, including:

According to described joint probability density function and Fisher information matrix, it is thus achieved that the symmetrical RAO detection statistic of spectrum；Utilize The symmetrical RAO detection statistic of described spectrum, completes the detection to target.

Preferably, according to below equation, complete the detection to target:

Wherein, η is detection threshold value,^TRepresent the transposition operation of matrix or vector,For H₁In the case of covariance matrix M Maximal possibility estimation, p₁And p₂It is respectively real part and imaginary part, the α of target nominal guiding vector₁And α₂It is the target letter received respectively The real part of number amplitude and imaginary part, H₀Represent driftlessness situation, H₁Indicate target conditions, z₁And z₂Represent cell data to be detected, Also known as master data.

The invention have the advantages that

(1) a kind of new space-time adaptive detection method based on priori, the priori utilized is reverberation merit The symmetry characteristic of rate spectrum；

(2) in the design process, the symmetry characteristic to reverberation power spectrum utilizes is by the inspection of binary composite hypothesis being asked Inscribe and be transformed into what real number field realized from complex field.Assistance data length is doubled by this conversion, is effectively increased reverberation association side The precision of difference Matrix Estimation.

(3) present invention assumes that and can obtain one group of uniform assistance data, in order to estimate the reverberation covariance square of to-be-measured cell Battle array, thus construct self-adapting detecting method；

(4) inventive method assumes that target direction is known, in order to calculate nominal guiding vector.

Accompanying drawing explanation

In order to be illustrated more clearly that the embodiment of the present invention or technical scheme of the prior art, below will be to embodiment or existing In having technology to describe, the required accompanying drawing used is introduced briefly.It should be evident that reflect in accompanying drawings below is only this A part of embodiment of invention, for those of ordinary skill in the art, on the premise of not paying creative work, also Other embodiments of the present invention can be obtained according to these accompanying drawings.And all these embodiment or embodiment are all the present invention's Within protection domain.

Based on symmetric spectral characteristic the sane space-time detection flow chart that Fig. 1 provides for the embodiment of the present invention；

Fig. 2 be the embodiment of the present invention by Meng Te-Caro emulation mode, in false-alarm probability P_fa=10^-4, N=8, K=16 In the case of, the present invention and tradition GLRT, AMF and ACE detector detection probability P_dRelation curve with SRR；

Fig. 3 be the embodiment of the present invention by Meng Te-Caro emulation mode, in false-alarm probability P_fa=10^-4, N=8, K=12 In the case of, the present invention and tradition GLRT, AMF and ACE detector detection probability P_dRelation curve with SRR；

Fig. 4 be the embodiment of the present invention by Meng Te-Caro emulation mode, in false-alarm probability P_fa=10^-4, N=8, K ＜ N feelings Under condition, the detection performance of the present invention.

Detailed description of the invention

Below by drawings and Examples, technical scheme is described in further detail.

During motion sonar target empty, target detection is actually a binary hypothesis test problem, and it is false that it includes target If with then driftlessness it is assumed that solve this problem according to corresponding test criterion (such as GLRT criterion and RAO criterion), being detected Statistic.The symmetry characteristic hint reverberation covariance matrix of reverberation power spectrum is a real-valued matrix, and this special structure permits Being permitted us and from complex field, binary hypothesis test problem is transformed into real number field, this conversion is equivalent to increase the length of assistance data One times, the precision of reverberation covariance matrix will be effectively improved.For the hypothesis problem after conversion, based on RAO test criterion, The present invention proposes a kind of symmetrical RAO method of new many bright spots object detection method, referred to as spectrum.The reason selecting RAO criterion has 2 points: one is compared to GLRT criterion, and RAO criterion needs the unknown parameter estimated less；Two is design simplicity, computation complexity The least.

Based on symmetric spectral characteristic the sane space-time detection flow chart that Fig. 1 provides for the embodiment of the present invention, as it is shown in figure 1, A kind of sane space-time detecting method based on symmetric spectral characteristic, concretely comprises the following steps:

Step S10, is received back to wave datum by sonar, and the target detection under the conditions of Reverberation is summarized as dualism hypothesis Inspection.

Specifically, it is assumed that the linear array that echo data is made up of N number of array element receives, in one embodiment, by N number of array element The linear array of composition receives the echo data of K length, and the Point Target Detection under the conditions of Reverberation is summarized as dualism hypothesis Inspection:

$\left\{\begin{array}{cc}{H}_0:\left\{\begin{array}{c}z=n,\\ z_k=n_k,\end{array}\right.& k=1,\mathrm{...},K\\ H_1:\left\{\begin{array}{c}z=\mathrm{\α}p+n,\\ z_k=n_k,\end{array}\right.& k=1,\mathrm{...},K\end{array}\right.---\left(1\right)$

Wherein, H₀And H₁Represent driftlessness respectively to assume and assume with the presence of target；P=p₁+jp₂It it is the target nominal of N-dimensional Guiding vector, p₁And p₂It is respectively its real part and imaginary part；α=α₁+jα₂It is the echo signal amplitude received, determines ginseng for the unknown Number, α₁And α₂It is its real part and imaginary part respectively；N=n₁+jn₂, n_k=n_1k+jn_2kIt is that independent identically distributed, zero-mean N-dimensional is combined Gaussian reverberation vector (n_1kAnd n_2kIt is respectively real part and imaginary part),^HRepresent the conjugate transposition operation of matrix or vector,；Z represents to be checked Survey cell data, also known as master data；z_kRepresenting the uniform assistance data of a length of K, it has identical reverberation association with master data Variance matrix

Step S20, based on symmetric spectral characteristic, is converted into the real-valued binary of equivalence by the binary hypothesis test in step S10 Composite hypothesis is checked.

It is worthy of note, owing to reverberation has the symmetric power spectrum centered by zero Doppler, then main number in step S10 According to the reverberation covariance matrix M with assistance data₀For real-valued.I.e. in (1) formula, reverberation is modeled as average and is zero, mixes by us Ringing covariance matrix is real-valued complex Gaussian reverberation vector, which imply that n and n_kReal part and imaginary part between covariance be Zero.Thus it is concluded that n₁、n₂、n_1kAnd n_2kBeing independent identically distributed Gaussian vectors, its average is zero, and now, reverberation is assisted Variance matrix becomes M, andAnd (1) formula can be converted into following equivalence real-valued binary composite hypothesis inspection ask Topic:

$\left\{\begin{array}{cc}{H}_0:\left\{\begin{array}{c}{z}_1=n_1,z_2=n_2,\\ z_{1k}=n_{1k},z_{2k}=n_{2k}\end{array}\right.& k=1,\mathrm{...},K\\ H_1:\left\{\begin{array}{c}{z}_1=({\mathrm{\α}}_1{p}_1-{\mathrm{\α}}_2{p}_2)+n_1,\\ z_{1k}=n_{1k},z_{2k}=n_{2k},\end{array}\right.& \begin{array}{c}{z}_2=({\mathrm{\α}}_1{p}_2-{\mathrm{\α}}_2{p}_2)+n_2,\\ k=1,\mathrm{...},K\end{array}\end{array}\right.---\left(2\right)$

Obviously, compared with (1) formula, in (2) formula, the length of assistance data doubles, and this is to utilize reverberation power spectrum symmetrical The result of this priori of characteristic.Correspondingly, solve the detection method obtained by (2) formula and will have more preferable robustness.

Step S30, checks based on the real-valued binary composite hypothesis in step S20, it is thus achieved that have under target conditions, echo data Middle master data and the joint probability density function of assistance data.

Specifically, check based on the real-valued binary composite hypothesis in step S20, under having target conditions, master data Z ≡ [z₁,z₂] and assistance data Z_K≡[z₁₁,...,z_1K,z₂₁...,z_2K] joint probability density function be:

$f(Z,Z_K|\mathrm{\θ},H_1)={\left(2\mathrm{\π}\right)}^{-N(K+1)}\det {\left(M\right)}^{-(K+1)}\exp \{-tr\[M^{-1}(u_1{u}_1^T+u_1{u}_1^T+S)\]\}---\left(3\right)$

Wherein,^TRepresent the transposition operation of matrix or vector；u₁=z₁-α₁p₁+α₂p₂, u₂=z₂-α₁p₂-α₂p₁； For sample covariance matrix based on assistance data；Det () and tr () represents determinant of a matrix and trace of a matrix respectively；θ It is the parameter column vector of N (N-1)/2+2 dimension, i.e.Wherein signal parameter vector θ_A=[α₁,α₂]^T, nuisance parameter arranges Vector θ_BBeing made up of the real-valued element of reverberation covariance matrix M, dimension is N (N-1)/2, owing to its concrete arrangement mode is to derivation Process is without impact, so not discussing here.

Step S40, based on the joint probability density function in step S30, obtains composing symmetrical RAO detection statistic, in order to Complete the detection to target.

Specifically include following steps:

(1) according to step S30 joint probability density function and Fisher information matrix, it is thus achieved that the symmetrical RAO detection statistics of spectrum Amount.

For the ease of following derivation, we provide Fisher information matrix, represent with J (θ), and its inverse matrix can represent For:

$J{\left(\mathrm{\θ}\right)}^{-1}={\left[\begin{array}{cc}{J}_{AA}\left(\mathrm{\θ}\right)& J_{AB}\left(\mathrm{\θ}\right)\\ J_{BA}\left(\mathrm{\θ}\right)& J_{BB}\left(\mathrm{\θ}\right)\end{array}\right]}^{-1}=\left[\begin{array}{cc}{C}_{AA}\left(\mathrm{\θ}\right)& C_{AB}\left(\mathrm{\θ}\right)\\ C_{BA}\left(\mathrm{\θ}\right)& C_{BB}\left(\mathrm{\θ}\right)\end{array}\right]---\left(4\right)$

Wherein, block matrixθ be the parameter of N (N-1)/2+2 dimension arrange to Amount, i.e.Wherein signal parameter vector θ_A=[α₁,α₂]^T, nuisance parameter column vector θ_BReality by covariance matrix M Value element is constituted, and dimension is N (N-1)/2.

For the test problems in (2), RAO statistic of test is represented by:

Wherein lnf (Z, Z_K|θ,H₁) it is f (Z, Z_K|θ,H₁) natural logrithm, Represent H₀In the case of θ_BMaximal possibility estimation, η is by false-alarm probability P_faThe detection threshold value determined.Easily try to achieve:

$\begin{array}{c}\frac{\∂\ln f(Z,Z_K|\mathrm{\θ},H_1)}{\∂{\mathrm{\θ}}_A}={\[\frac{\∂\ln f(Z,Z_K|\mathrm{\θ},H_1)}{\∂{\mathrm{\α}}_1},\frac{\∂\ln f(Z,Z_K|\mathrm{\θ},H_1)}{\∂{\mathrm{\α}}_2}\]}^T\\ ={\[p_1^T{M}^{-1}{u}_1+p_2^T{M}^{-1}{u}_2,p_1^T{M}^{-1}{u}_2-p_2^T{M}^{-1}{u}_1\]}^T\end{array}---\left(6\right)$

Wherein,WithRepresent respectively relative to α₁And α₂Partial derivative.

Further derive and need to calculate the block matrix of Fisher information matrix:

Because,

$J_{AA}\left(\mathrm{\θ}\right)=(p_1^T{M}^{-1}{p}_1+p_2^T{M}^{-1}{p}_2)I_2,J_{AA}\left(\mathrm{\θ}\right)=0_{2,N(N-1)/2},---\left(7\right)$

Wherein, I₂The unit square formation of expression 2 × 2,0_2,N(N-1)/2Representation dimension is the null matrix of 2 × N (N-1)/2.Through deriving Available block matrix:

$C_{AA}\left(\mathrm{\θ}\right)=J_{AA}^{-1}\left(\mathrm{\θ}\right)=\frac{1}{p_1^T{M}^{-1}{p}_1+p_2^T{M}^{-1}{p}_2}{I}_2---\left(8\right)$

It addition, be apparent from H₁In the case of the maximal possibility estimation of M beTo sum up, by (6), (7) and (8) Substituting into (5) can obtain, the final expression formula of the symmetrical RAO method of spectrum is:

Wherein, η is detection threshold value,^TRepresent the transposition operation of matrix or vector,For H₁In the case of covariance matrix M Maximal possibility estimation, p₁And p₂It is respectively real part and imaginary part, the α of target nominal guiding vector₁And α₂It is the target letter received respectively The real part of number amplitude and imaginary part, H₀Represent driftlessness situation, H₁Indicate target conditions, z₁And z₂Represent cell data to be detected, Also known as master data.

Above-mentioned formula (9) is the symmetrical RAO detection statistic of spectrum.

(2) utilize the symmetrical RAO detection statistic of described spectrum, complete the detection to target.

Specifically, according to formula (9), complete the detection to target, i.e. under Reverberation, complete target data and training The automatic screening of data.

The present invention comes the detection performance of analytical spectra symmetry RAO method, and and traditional detection by Meng Te-Caro emulation mode Device GLRT, AMF and ACE compare.Wherein, false-alarm probability P_faWith detection probability P_dSimulation times be respectively 100/P_faWith 10⁴, it is sufficient to reliable simulation result is provided.Reverberation model uses common correlation of indices complex Gaussian model, real-valued covariance Matrix M=0.9^|i-j|, wherein, (i j) is the coordinate of matrix element.False-alarm probability P_fa=10^-4, the mixed ratio of letter is defined as SRR= p^HM^-1P, nominal target steering vector p=[1 ..., 1]^T/ N, wherein []^TRepresent transposition operation.

Four kinds of detector detection probabilities P in the case of two kinds of K values when Fig. 2 and Fig. 3 sets forth N=8_dRelation with SRR Curve.By this two width figure it can be seen that the performance of the symmetrical RAO method of spectrum is substantially better than other three kinds of traditional detectors, and Assistance data length is the least, and performance advantage is the most obvious.Such as K=16, P_d=0.9 symmetrical RAO method of spectrum is relative to GLRT The detection gain of method is 2.8dB, and when K is reduced to 12, this gain increases to 5.1dB.Visible, reasonably utilize reverberation The symmetry of spectrum can be effectively improved the detection performance of high-resolution sonar system, the especially situation of small sample assistance data.

Fig. 4 gives the detection performance in the case of the symmetrical RAO method K ＜ N of spectrum, it can be seen that along with the increasing of K value Greatly, the symmetrical RAO method of spectrum has better performance.Need exist for it is emphasized that the method for the present invention is by examining dualism hypothesis Problem of testing is transformed into real number field from complex field, is equivalent to double assistance data length, thus relaxes assistance data The requirement of length, i.e. K >=N/2.And for conventional detector, then require K >=N, it is impossible to work in this situation.

For motion sonar system, traditional STAD method does not accounts for the symmetrical special of reverberation power spectrum in the design process Property, ignoring the utilization to this priori, its robustness needs to be improved further, with the demand of satisfied actual application.For Solving this problem, the inventive method the most just considers this characteristic, is incorporated detection statistic, it is achieved right Effective utilization of this priori, is greatly improved the detection performance of STAD under non-homogeneous environment.

Professional should further appreciate that, each example described in conjunction with the embodiments described herein Unit and algorithm steps, it is possible to electronic hardware, computer software or the two be implemented in combination in, hard in order to clearly demonstrate Part and the interchangeability of software, the most generally describe composition and the step of each example according to function. These functions perform with hardware or software mode actually, depend on application-specific and the design constraint of technical scheme. Professional and technical personnel can use different methods to realize described function to each specifically should being used for, but this realization It is not considered that it is beyond the scope of this invention.

The method described in conjunction with the embodiments described herein or the step of algorithm can use hardware, processor to perform Software module, or the combination of the two implements.Software module can be placed in random access memory (RAM), internal memory, read only memory (ROM), electrically programmable ROM, electrically erasable ROM, depositor, hard disk, moveable magnetic disc, CD-ROM or technical field In any other form of storage medium well known in.

Above-described detailed description of the invention, has been carried out the purpose of the present invention, technical scheme and beneficial effect further Describe in detail, be it should be understood that the detailed description of the invention that the foregoing is only the present invention, be not intended to limit the present invention Protection domain, all within the spirit and principles in the present invention, any modification, equivalent substitution and improvement etc. done, all should comprise Within protection scope of the present invention.

Claims (6)

1. a sane space-time detecting method based on symmetric spectral characteristic, it is characterised in that described method includes:

It is received back to wave datum by sonar, and the target detection under the conditions of Reverberation is summarized as binary hypothesis test；

Based on symmetric spectral characteristic, described binary hypothesis test is converted into the real-valued binary composite hypothesis inspection of equivalence；

Check based on described real-valued binary composite hypothesis, it is thus achieved that have under target conditions, master data and auxiliary in described echo data The joint probability density function of data；

Based on described joint probability density function, obtain composing symmetrical RAO detection statistic, in order to complete the detection to target.

Method the most according to claim 1, it is characterised in that described be received back to wave datum by sonar, and by Reverberation Under the conditions of target detection be summarized as binary hypothesis test problem, including:

The linear array being made up of N number of array element receives the echo data of K length, and the target detection under the conditions of Reverberation is concluded For binary hypothesis test:

Wherein,

H₀And H₁Represent driftlessness respectively to assume and assume with the presence of target；

P=p₁+jp₂It is the target nominal guiding vector of N-dimensional, p₁And p₂It is respectively its real part and imaginary part；

α=α₁+jα₂It is the echo signal amplitude received, α₁And α₂It is its real part and imaginary part respectively；

N=n₁+jn₂, n_k=n_1k+jn_2kIt is independent identically distributed, zero-mean N-dimensional complex Gaussian reverberation vector, n_1kAnd n_2kIt is respectively Real part and imaginary part；

Z represents cell data to be detected, also known as master data；z_kRepresent the uniform assistance data of a length of K.

Method the most according to claim 1, it is characterised in that described based on symmetric spectral characteristic, examines described dualism hypothesis Test the real-valued binary composite hypothesis inspection being converted into equivalence, including:

According to the symmetry characteristic of reverberation power spectrum, described binary hypothesis test is converted into the real-valued binary composite hypothesis inspection of equivalence Test:

Wherein,

H₀And H₁Represent driftlessness respectively to assume and assume with the presence of target；

p₁And p₂It is respectively real part and the imaginary part of target nominal guiding vector；

α₁And α₂It is real part and the imaginary part of the echo signal amplitude received respectively；

n₁、n₂、n_1kAnd n_2kIt it is independent identically distributed, zero-mean N-dimensional complex Gaussian reverberation vector；

z₁And z₂Represent cell data to be detected, also known as master data；z_1kAnd z_2kRepresent the uniform assistance data of a length of K.

Method the most according to claim 1, it is characterised in that described based on the inspection of described real-valued binary composite hypothesis, obtains Must have under target conditions, master data and the joint probability density function of assistance data in described echo data, including:

Check based on described real-valued binary composite hypothesis, under having target conditions, master data Z ≡ [z₁,z₂] and assistance data Z_K≡ [z₁₁,...,z_1K,z₂₁...,z_2K] joint probability density function be:

Wherein,

T represents the transposition operation of matrix or vector；

u₁=z₁-α₁p₁+α₂p₂, u₂=z₂-α₁p₂-α₂p₁；

For sample covariance matrix based on assistance data；

Det () and tr () represents determinant of a matrix and trace of a matrix respectively；

θ is the parameter column vector of N (N-1)/2+2 dimension, i.e.Wherein signal parameter vector θ_A=[α₁,α₂]^T, redundancy Parameter column vector θ_BBeing made up of the real-valued element of reverberation covariance matrix M, dimension is N (N-1)/2.

Method the most according to claim 1, it is characterised in that described based on described joint probability density function, it is thus achieved that spectrum Symmetrical RAO detection statistic, in order to complete the detection to target, including:

According to described joint probability density function and Fisher information matrix, it is thus achieved that the symmetrical RAO detection statistic of spectrum；

Utilize the symmetrical RAO detection statistic of described spectrum, complete the detection to target.

6., according to the method described in claim 1 or 4, it is characterised in that according to below equation, complete the detection to target:

Wherein, η is detection threshold value, and T represents the transposition operation of matrix or vector,For H₁In the case of the maximum of covariance matrix M Possibility predication, p₁And p₂It is respectively real part and imaginary part, the α of target nominal guiding vector₁And α₂It is the echo signal width received respectively The real part of degree and imaginary part, H₀Represent driftlessness situation, H₁Indicate target conditions, z₁And z₂Represent cell data to be detected, also known as Master data.