CN104618033A - Multi-layer self-adapting morphological filtering gravity signal noise inhibition method - Google Patents

Abstract

The invention discloses a multi-layer self-adapting morphological filtering gravity signal noise inhibition method. The multi-layer self-adapting morphological filtering gravity signal noise inhibition method includes that after gathering related data information, carrying out multiple-level decomposition on an original gravity signal, calculating the value of relevancy between each layer of component and reference noise, confirming a filter weight of each layer of component for performing self-adapting morphological filtering according to the relevancy value, finishing the self-adapting morphological filtering for each layer of component, and using the filtered result of each layer of component to finish the gravity signal reconstruction. The multi-layer self-adapting morphological filtering gravity signal noise inhibition method overcomes the problem that filtering parameters of a traditional filtering method cannot perform self-adapting regulation; the multi-layer self-adapting morphological filtering gravity signal noise inhibition method is capable of retaining effective gravity signals and realizing high-precision gravity signal extraction.

Description

A kind of multilayer self-adaptive harmonics detection gravitational cue noise suppressing method

Technical field

The present invention relates to terrestrial gravitation Detection Techniques field, specifically a kind of multilayer self-adaptive harmonics detection gravitational cue noise suppressing method, this invention effectively can suppress the noise in gravity measurement signal, extracts useful gravitational cue.

Background technology

Terrestrial gravitation data are grand strategy resources of country, have important effect at numerous areas such as national defense construction, space technology, resource detection and geophysics science.Gravitational cue process utilizes ocean or airborne gravitormeter to carry out gravitational cue collection on the spot exactly, and extracts useful gravitational cue by effective data processing method, for above-mentioned field provides gravimetric data support.In order to the measurement noises in gravitational cue effectively can be eliminated, the methods such as multiplex FIR or the IIR low pass filter of conventional process, Kalman filter.Although these methods achieve certain filter effect, but there is the problems such as the such as Parameter uncertainties such as cut-off frequency, exponent number, these uncertain factors bring adverse effect to the noise suppressed of filter, can not provide effective technical guarantee for gravity detection and data processing.

Summary of the invention

Goal of the invention: in order to overcome the deficiency existed in traditional filtering technique, the invention provides a kind of multilayer self-adaptive harmonics detection gravitational cue noise suppressing method, can effectively eliminate gravity measurement noise and extract effective gravity signal.

Technical scheme: for achieving the above object, the technical solution used in the present invention is:

A kind of multilayer self-adaptive harmonics detection gravitational cue noise suppressing method, comprises the steps:

(1) utilize ocean or airborne gravitormeter to measure according to effective survey line, obtain original gravitational cue sequence g (t)=[g (1) g (2) ... g (N)] of t, N is sequence length; Meanwhile, a high-precision accelerometer installed by gravimeter stabilized platform, measure interference and the noise of vertical direction, obtain noise reference signal sequence n (t) that length is N;

(2) initialization represents that the parameter k of number of repetition be the span of 1, k be 1 to K, K is arbitrary value in 3 ~ 10;

(3) constructing a pair length is N, and amplitude is identical, the Gaussian sequence x of carrier phase shift 180 ° ^(k)(t) and-x ^(k)t (), then by x ^(k)(t) and-x ^(k)t () is superimposed with g (t) respectively obtains two new signal sequences with namely $g_1^{\left(k\right)}\left(t\right)=g\left(t\right)+x^{\left(k\right)}\left(t\right),g_2^{\left(k\right)}\left(t\right)=g\left(t\right)-x^{\left(k\right)}\left(t\right),$ Wherein, Gaussian sequence x ^(k)(t) and-x ^(k)t the amplitude of () selects the standard deviation of 0.2 ~ 0.5 times of original gravitational cue sequence g (t);

(4) to new signal sequence with carry out multilayer decomposition respectively, obtain two groups of decomposition result, often organize results set and be designated as respectively with wherein $F_1^{\left(k\right)}\left(t\right)=\{f_{11}^{\left(k\right)}\left(t\right),f_{12}^{\left(k\right)}\left(t\right),...,f_{1m}^{\left(k\right)}\left(t\right)\},$ $F_2^{\left(k\right)}\left(t\right)=\{f_{21}^{\left(k\right)}\left(t\right),f_{22}^{\left(k\right)}\left(t\right),...,f_{2m}^{\left(k\right)}\left(t\right)\},$ M is Decomposition order;

(5) k adds 1, if k is less than or equal to K, then returns step (3), otherwise enters step (6);

(6) set of computations with the mean value of middle respective components, obtains new set G (t)={ g ₁(t) g ₂(t) ... g _m(t) }, wherein:

$g_i\left(t\right)=\frac{\underset{k=1}{\overset{K}{\mathrm{\Σ}}}(f_{1i}^{\left(k\right)}\left(t\right)+f_{2i}^{\left(k\right)}\left(t\right))}{2K},i=\mathrm{1,2},..,m;$

(7) each layer component g obtained in calculation procedure (6) _ithe relevance degree of noise reference signal sequence n (t) obtained in (t) and step (1), the relatedness computation formula of i-th layer of component is as follows:

${\mathrm{\ρ}}_i=\frac{g_i\left(t\right)n^T\left(t\right)}{{\left|\right|g_i\left(t\right)\left|\right|}_2^2{\left|\right|n\left(t\right)\left|\right|}_2^2},$

Wherein || g _i(t) || ₂represent sequence g _ithe 2-norm of (t), || n (t) || ₂represent the 2-norm of sequence n (t);

(8) calculate each layer component g obtained in step (6) according to the relevance degree of step (7) _it filter weights R that () processes further _i, filter weights calculates as follows by formula:

$R_i=\frac{{\mathrm{\ρ}}_i}{\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{\mathrm{\ρ}}_i};$

(9) by R _ias the filter weights of self-adaptive harmonics detection process, to each layer component g obtained in step (6) _it () carries out self-adaptive harmonics detection process, obtain the filtering output sequence of each layer

(10) the filtering output sequence of each layer obtained by step (9) carry out the reconstruct of gravitational cue sequence, the effective gravity burst after the measurement noises that can be eliminated that is:

$\hat{g}\left(t\right)=\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{\hat{g}}_i\left(t\right).$

Multilayer decomposition in described step (4) specifically comprises the steps:

(41) new signal sequence is asked for all extreme points, determine all maximum points and the minimum point of this sequence;

(42) all maximum points, minimum point are coupled together with cubic spline curve respectively, obtain upper and lower envelope with sequence between upper and lower envelope

(43) the average line of upper and lower envelope is calculated and obtain $h_1^{\left(k\right)}\left(t\right)=g_1^{\left(k\right)}\left(t\right)-m_1^{\left(k\right)}\left(t\right);$

(44) h is judged ₁t whether () meet following two conditions:

(441) signal zero crossing number is equal with limit number or be more or less the same in 1;

(442) average of the envelope formed respectively by maximum point and minimum point is 0;

If do not meet (441) and (442) two conditions, then use substitute new signal sequence repeat step (41) ~ (44), until meet (441) and (442) two conditions, namely obtain first decomposed component, be expressed as

(45) by sequence repeat step (41) ~ (44) as new signal sequence, obtain each layer decomposed component until m layer has decomposed, gathered $F_1^{\left(k\right)}\left(t\right)=\{f_{11}^{\left(k\right)}\left(t\right),f_{12}^{\left(k\right)}\left(t\right),...,f_{1m}^{\left(k\right)}\left(t\right)\};$

(46) to new signal sequence carry out multilayer decomposition according to step (41) ~ (45), obtain multilayer and decompose set $F_2^{\left(k\right)}\left(t\right)=\{f_{21}^{\left(k\right)}\left(t\right),f_{22}^{\left(k\right)}\left(t\right),...,f_{2m}^{\left(k\right)}\left(t\right)\}.$

Self-adaptive harmonics detection process in described step (9) specifically comprises the steps:

(91) designated length is the shape filtering structural element b of L, according to the feature of gravimeter signal, can choose sinusoidal configuration element, such as b={0 7.7071 10 7.7071 0}, now L=5;

(92) to g _it () employing form open-close filter and form are closed-Kai filter and are carried out data filtering, wherein:

Form open-close filter is defined as:

OC(g _i(t))＝g _i(t)ο(R _ib)·(R _ib)，

Form is closed-Kai filter and is defined as:

CO(g _i(t))＝g _i(t)·(R _ib)ο(R _ib)，

Open-close operation definition is:

Close-opening operation is defined as:

Operator Θ is defined as:

$g_i\left(t\right)\mathrm{\Θ}\left(R_{i}b\right)=\underset{n=1,...N}{\min }[g_i(n+l)-R_{i}b\left(n\right)],l=\mathrm{1,2},...,N-L,$

Operator be defined as:

$g_i\left(t\right)\⊕\left(R_{i}b\right)=\underset{n=\mathrm{1,2},...,N}{\max }[g_i(l-n)+R_{i}b\left(n\right)],l=\mathrm{1,2},...,N-L;$

(93) to open-close filter results with close-Kai filter results and average, obtain self-adaptive harmonics detection and export ${\hat{g}}_i\left(t\right)=(\mathrm{OC}\left(g_i\left(t\right)\right)+\mathrm{CO}\left(g_i\left(t\right)\right))/2;$

(94) (92) ~ (93) are repeated, until g _i(t), i=1,2 ..., the whole filtering of m is complete, obtains the filtering output sequence of each layer i=1,2 ..., m.

Beneficial effect: multilayer self-adaptive harmonics detection gravitational cue noise suppressing method provided by the invention, has following advantage relative to prior art: 1, overcoming filtering parameter in traditional filtering method can not the problem of self-adaptative adjustment; 2, effective gravitational cue can be retained, realize high-precision gravitational cue and extract.

Accompanying drawing explanation

Fig. 1 is the realization flow figure of the inventive method;

Fig. 2 is the implementation result figure of the inventive method.

Embodiment

Below in conjunction with drawings and Examples, technical solutions according to the invention are further elaborated.

A kind of multilayer self-adaptive harmonics detection gravitational cue noise suppressing method, comprises the steps:

(1) utilize ocean or airborne gravitormeter to measure according to effective survey line, obtain original gravitational cue sequence g (t)=[g (1) g (2) ... g (N)] of t, N is sequence length; Meanwhile, a high-precision accelerometer installed by gravimeter stabilized platform, measure interference and the noise of vertical direction, obtain noise reference signal sequence n (t) that length is N;

(2) initialization represents that the parameter k of number of repetition be the span of 1, k be 1 to K, K is arbitrary value in 3 ~ 10;

(3) constructing a pair length is N, and amplitude is identical, the Gaussian sequence x of carrier phase shift 180 ° ^(k)(t) and-x ^(k)t (), then by x ^(k)(t) and-x ^(k)t () is superimposed with g (t) respectively obtains two new signal sequences with namely $g_1^{\left(k\right)}\left(t\right)=g\left(t\right)+x^{\left(k\right)}\left(t\right),g_2^{\left(k\right)}\left(t\right)=g\left(t\right)-x^{\left(k\right)}\left(t\right),$ Wherein, Gaussian sequence x ^(k)(t) and-x ^(k)t the amplitude of () selects the standard deviation of 0.2 ~ 0.5 times of original gravitational cue sequence g (t);

(4) to new signal sequence with carry out multilayer decomposition respectively, obtain two groups of decomposition result, often organize results set and be designated as respectively with wherein $F_1^{\left(k\right)}\left(t\right)=\{f_{11}^{\left(k\right)}\left(t\right),f_{12}^{\left(k\right)}\left(t\right),...,f_{1m}^{\left(k\right)}\left(t\right)\},$ $F_2^{\left(k\right)}\left(t\right)=\{f_{21}^{\left(k\right)}\left(t\right),f_{22}^{\left(k\right)}\left(t\right),...,f_{2m}^{\left(k\right)}\left(t\right)\},$ M is Decomposition order;

Multilayer decomposition specifically comprises the steps:

(41) new signal sequence g is asked for ₁ ^(k)t all extreme points of (), determine all maximum points and the minimum point of this sequence;

(42) all maximum points, minimum point are coupled together with cubic spline curve respectively, obtain upper and lower envelope with sequence between upper and lower envelope

(43) the average line of upper and lower envelope is calculated and obtain $h_1^{\left(k\right)}\left(t\right)=g_1^{\left(k\right)}\left(t\right)-m_1^{\left(k\right)}\left(t\right);$

(44) h is judged ₁t whether () meet following two conditions:

(441) signal zero crossing number is equal with limit number or be more or less the same in 1;

(442) average of the envelope formed respectively by maximum point and minimum point is 0;

If do not meet (441) and (442) two conditions, then use substitute new signal sequence repeat step (41) ~ (44), until meet (441) and (442) two conditions, namely obtain first decomposed component, be expressed as

(45) by sequence repeat step (41) ~ (44) as new signal sequence, obtain each layer decomposed component until m layer has decomposed, gathered $F_1^{\left(k\right)}\left(t\right)=\{f_{11}^{\left(k\right)}\left(t\right),f_{12}^{\left(k\right)}\left(t\right),...,f_{1m}^{\left(k\right)}\left(t\right)\};$

(46) to new signal sequence carry out multilayer decomposition according to step (41) ~ (45), obtain multilayer and decompose set $F_2^{\left(k\right)}\left(t\right)=\{f_{21}^{\left(k\right)}\left(t\right),f_{22}^{\left(k\right)}\left(t\right),...,f_{2m}^{\left(k\right)}\left(t\right)\};$

(5) k adds 1, if k is less than or equal to K, then returns step (3), otherwise enters step (6);

(6) set of computations with the mean value of middle respective components, obtains new set G (t)={ g ₁(t) g ₂(t) ... g _m(t) }, wherein:

$g_i\left(t\right)=\frac{\underset{k=1}{\overset{K}{\mathrm{\Σ}}}(f_{1i}^{\left(k\right)}\left(t\right)+f_{2i}^{\left(k\right)}\left(t\right))}{2K},i=\mathrm{1,2},..,m;$

(7) each layer component g obtained in calculation procedure (6) _ithe relevance degree of noise reference signal sequence n (t) obtained in (t) and step (1), the relatedness computation formula of i-th layer of component is as follows:

${\mathrm{\ρ}}_i=\frac{g_i\left(t\right)n^T\left(t\right)}{{\left|\right|g_i\left(t\right)\left|\right|}_2^2{\left|\right|n\left(t\right)\left|\right|}_2^2},$

Wherein || g _i(t) || ₂represent sequence g _ithe 2-norm of (t), || n (t) || ₂represent the 2-norm of sequence n (t);

(8) calculate each layer component g obtained in step (6) according to the relevance degree of step (7) _it filter weights R that () processes further _i, filter weights calculates as follows by formula:

$R_i=\frac{{\mathrm{\ρ}}_i}{\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{\mathrm{\ρ}}_i};$

(9) by R _ias the filter weights of self-adaptive harmonics detection process, to each layer component g obtained in step (6) _it () carries out self-adaptive harmonics detection process, obtain the filtering output sequence of each layer

Self-adaptive harmonics detection process specifically comprises the steps:

(91) designated length is the shape filtering structural element b of L, according to the feature of gravimeter signal, can choose sinusoidal configuration element, such as b={07.7071107.70710}, now L=5;

(92) to g _it () employing form open-close filter and form are closed-Kai filter and are carried out data filtering, wherein:

Form open-close filter is defined as:

OC(g _i(t))＝g _i(t)ο(R _ib)·(R _ib)，

Form is closed-Kai filter and is defined as:

CO(g _i(t))＝g _i(t)·(R _ib)ο(R _ib)，

Open-close operation definition is:

Close-opening operation is defined as:

Operator Θ is defined as:

$g_i\left(t\right)\mathrm{\Θ}\left(R_{i}b\right)=\underset{n=1,...N}{\min }[g_i(n+l)-R_{i}b\left(n\right)],l=\mathrm{1,2},...,N-L,$

Operator be defined as:

$g_i\left(t\right)\⊕\left(R_{i}b\right)=\underset{n=\mathrm{1,2},...,N}{\max }[g_i(l-n)+R_{i}b\left(n\right)],l=\mathrm{1,2},...,N-L;$

(93) to open-close filter results with close-Kai filter results and average, obtain self-adaptive harmonics detection and export ${\hat{g}}_i\left(t\right)=(\mathrm{OC}\left(g_i\left(t\right)\right)+\mathrm{CO}\left(g_i\left(t\right)\right))/2;$

(94) (92) ~ (93) are repeated, until g _i(t), i=1,2 ..., the whole filtering of m is complete, obtains the filtering output sequence of each layer i=1,2 ..., m;

(10) the filtering output sequence of each layer obtained by step (9) carry out the reconstruct of gravitational cue sequence, the effective gravity burst after the measurement noises that can be eliminated that is:

$\hat{g}\left(t\right)=\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{\hat{g}}_i\left(t\right).$

Performance evaluation

The multilayer self-adaptive harmonics detection gravitational cue noise suppressed processing method that the present invention proposes overcomes filtering parameter in traditional filtering method can not the problem of self-adaptative adjustment, each noise like can be suppressed preferably, retain effective gravitational cue, realize high-precision gravitational cue and extract.

Accompanying drawing 2 is gravitational cue noise suppressing method implementation result figure.In figure, solid line represents gravity measurement signal before treatment, and dotted line represents the output gravitational cue after noise suppressing method of the present invention.As can be seen from accompanying drawing 2, high-frequency noise is effectively suppressed, and gravitational cue and variation tendency thereof are extracted efficiently, and reaches the object of noise suppressed.

The above is only main embodiment of the present invention; be noted that for those skilled in the art; under the premise without departing from the principles of the invention, can also make some improvements and modifications, these improvements and modifications also should be considered as protection scope of the present invention.

Claims (3)

1. a multilayer self-adaptive harmonics detection gravitational cue noise suppressing method, is characterized in that: comprise the steps:

(1) utilize ocean or airborne gravitormeter to measure according to effective survey line, obtain original gravitational cue sequence g (t)=[g (1) g (2) of t ... g (N)], N is sequence length; Meanwhile, a high-precision accelerometer installed by gravimeter stabilized platform, measure interference and the noise of vertical direction, obtain noise reference signal sequence n (t) that length is N;

(2) initialization represents that the parameter k of number of repetition be the span of 1, k be 1 to K, K is arbitrary value in 3 ~ 10;

(3) constructing a pair length is N, and amplitude is identical, the Gaussian sequence x of carrier phase shift 180 ° ^(k)(t) and-x ^(k)t (), then by x ^(k)(t) and-x ^(k)t () is superimposed with g (t) respectively obtains two new signal sequences with namely wherein, Gaussian sequence x ^(k)(t) and-x ^(k)t the amplitude of () selects the standard deviation of 0.2 ~ 0.5 times of original gravitational cue sequence g (t);

(4) to new signal sequence with carry out multilayer decomposition respectively, obtain two groups of decomposition result, often organize results set and be designated as respectively with wherein $F_1^{\left(k\right)}\left(t\right)=\{f_{11}^{\left(k\right)}\left(t\right),f_{12}^{\left(k\right)}\left(t\right),...,f_{1m}^{\left(k\right)}\left(t\right)\},$ $F_2^{\left(k\right)}\left(t\right)=\{f_{21}^{\left(k\right)}\left(t\right),f_{22}^{\left(k\right)}\left(t\right),...,f_{2m}^{\left(k\right)}\left(t\right)\},$ M is Decomposition order;

(5) k adds 1, if k is less than or equal to K, then returns step (3), otherwise enters step (6);

(6) set of computations with the mean value of middle respective components, obtains new set G (t)={ g ₁(t) g ₂(t) ... g _m(t) }, wherein:

$g_i\left(t\right)=\frac{\underset{k=1}{\overset{k}{\mathrm{\Σ}}}(f_{1i}^{\left(k\right)}\left(t\right)+f_{2i}^{\left(k\right)}\left(t\right))}{2K},i=\mathrm{1,2},...,m;$

(7) each layer component g obtained in calculation procedure (6) _ithe relevance degree of noise reference signal sequence n (t) obtained in (t) and step (1), the relatedness computation formula of i-th layer of component is as follows:

${\mathrm{\ρ}}_i=\frac{g_i\left(t\right)n^T\left(t\right)}{{\left|\right|g_i\left(t\right)\left|\right|}_2^2{\left|\right|n\left(t\right)\left|\right|}_2^2},$

Wherein || g _i(t) || ₂represent sequence g _ithe 2-norm of (t), || n (t) || ₂represent the 2-norm of sequence n (t);

(8) calculate each layer component g obtained in step (6) according to the relevance degree of step (7) _it filter weights R that () processes further _i, filter weights calculates as follows by formula:

$R_i=\frac{{\mathrm{\ρ}}_i}{\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{\mathrm{\ρ}}_i};$

(9) by R _ias the filter weights of self-adaptive harmonics detection process, to each layer component g obtained in step (6) _it () carries out self-adaptive harmonics detection process, obtain the filtering output sequence of each layer

(10) the filtering output sequence of each layer obtained by step (9) carry out the reconstruct of gravitational cue sequence, the effective gravity burst after the measurement noises that can be eliminated that is:

$\hat{g}\left(t\right)=\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{\hat{g}}_i\left(t\right).$

2. multilayer self-adaptive harmonics detection gravitational cue noise suppressing method according to claim 1, is characterized in that: the multilayer decomposition in described step (4) specifically comprises the steps:

(41) new signal sequence is asked for all extreme points, determine all maximum points and the minimum point of this sequence;

(42) all maximum points, minimum point are coupled together with cubic spline curve respectively, obtain upper and lower envelope with sequence between upper and lower envelope

(43) the average line of upper and lower envelope is calculated and obtain $h_1^{\left(k\right)}\left(t\right)=g_1^{\left(k\right)}\left(t\right)-m_1^{\left(k\right)}\left(t\right);$

(44) h is judged ₁t whether () meet following two conditions:

(441) signal zero crossing number is equal with limit number or be more or less the same in 1;

(442) average of the envelope formed respectively by maximum point and minimum point is 0;

If do not meet (441) and (442) two conditions, then use substitute new signal sequence repeat step (41) ~ (44), until meet (441) and (442) two conditions, namely obtain first decomposed component, be expressed as

(45) by sequence repeat step (41) ~ (44) as new signal sequence, obtain each layer decomposed component until m layer has decomposed, gathered $F_1^{\left(k\right)}\left(t\right)=\{f_{11}^{\left(k\right)}\left(t\right),f_{12}^{\left(k\right)}\left(t\right),...,f_{1m}^{\left(k\right)}\left(t\right)\},$

(46) to new signal sequence carry out multilayer decomposition according to step (41) ~ (45), obtain multilayer and decompose set $F_2^{\left(k\right)}\left(t\right)=\{f_{21}^{\left(k\right)}\left(t\right),f_{22}^{\left(k\right)}\left(t\right),...,f_{2m}^{\left(k\right)}\left(t\right)\},$

3. multilayer self-adaptive harmonics detection gravitational cue noise suppressing method according to claim 1, is characterized in that: the self-adaptive harmonics detection process in described step (9) specifically comprises the steps:

(91) designated length is the shape filtering structural element b of L;

(92) to g _it () employing form open-close filter and form are closed-Kai filter and are carried out data filtering, wherein:

Form open-close filter is defined as:

Form is closed-Kai filter and is defined as:

Open-close operation definition is:

Close-opening operation is defined as:

Operator Θ is defined as:

$g_i\left(t\right)\mathrm{\Θ}\left(R_{i}b\right)=\underset{n=1,...N}{\min }[g_i(n+l)-R_{i}b\left(n\right)],l=\mathrm{1,2},...,N-L,$

Operator ⊕ is defined as:

$g_i\left(t\right)\⊕\left(R_{i}b\right)=\underset{n=1,2,...N}{\min }[g_i(l-n)+R_{i}b\left(n\right)],l=\mathrm{1,2},...,N-L,$

(93) to open-close filter results with close-Kai filter results and average, obtain self-adaptive harmonics detection and export ${\hat{g}}_i\left(t\right)=(\mathrm{OC}\left(g_i\left(t\right)\right)+\mathrm{CO}\left(g_i\left(t\right)\right))/2;$

(94) (92) ~ (93) are repeated, until g _i(t), i=1,2 ..., the whole filtering of m is complete, obtains the filtering output sequence of each layer ${\hat{g}}_i\left(t\right),i=\mathrm{1,2},...,m.$