CN103716013B - Variable element ratio sef-adapting filter - Google Patents
Variable element ratio sef-adapting filter Download PDFInfo
- Publication number
- CN103716013B CN103716013B CN201410015022.5A CN201410015022A CN103716013B CN 103716013 B CN103716013 B CN 103716013B CN 201410015022 A CN201410015022 A CN 201410015022A CN 103716013 B CN103716013 B CN 103716013B
- Authority
- CN
- China
- Prior art keywords
- sef
- power
- ratio
- adapting filter
- coefficient
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
Landscapes
- Filters That Use Time-Delay Elements (AREA)
Abstract
The invention discloses a kind of variable element ratio sef-adapting filter, belong to digital filter design field. Variable element ratio sef-adapting filter is adjusted the value of the coefficient gain of sef-adapting filter with a time-varying parameter. This time-varying parameter can be expressed as the monotonically increasing function of error signal power and system noise power ratio. In the starting stage of adaptive-filtering, error signal power and system noise power ratio are larger, thereby the large coefficient of sef-adapting filter has larger yield value; In the converged state of adaptive-filtering, error signal power and system noise power ratio are less, thereby the large coefficient of sef-adapting filter has less yield value. Therefore, this variable element method can keep the fast convergence rate of ratio sef-adapting filter, can obtain again the stable state imbalance fluctuation that ratio sef-adapting filter is low.
Description
Technical field
The invention belongs to digital filter design field, relate to a kind of coefficient update side of sef-adapting filterMethod, is specifically related to the self-adjusting ratio sef-adapting filter of a kind of parameter.
Background technology
The coefficient vector of traditional digital filter is fixed. The main task of traditional digital filterBe spectrum component useless in filtering input signal, and retain the spectrum component needing, thereby its operationMode is to obtain output signal according to the coefficient vector of input signal and wave filter. With traditional coefficient vectorFixing wave filter difference, sef-adapting filter can, according to the input of unknown system, output signal, comeApproach this unknown system. Due to resolution system identification, echo elimination, active noise controlling, channel equalization,The name of the games such as Interference Cancellation are to try to achieve this unknown system according to the input and output signal of unknown systemSystem, thereby sef-adapting filter is at hands-free phone, video conference, audiphone, channel equalizer, electronicsIn the equipment such as scalpel, be applied widely.
The leading indicator of weighing sef-adapting filter performance has convergence rate and stable state imbalance. Convergence rate certainlyDetermined sef-adapting filter and approached the time that unknown system needs, and stable state imbalance determines to approach unknown systemThe precision that system can reach. A principal element that affects the convergence rate of sef-adapting filter is unknown systemThe degree of rarefication of system. A unknown system, it approaches or to equal 0 coefficient more, and its degree of rarefication is higher;Otherwise its coefficient degree is lower. In the time that the degree of rarefication of unknown system is very high, traditional LMS and NLMSThe convergence rate of sef-adapting filter is very slow. In hands-free phone, video conference, audiphone, need to forceNear unknown system degree of rarefication is higher, in order to obtain convergence rate faster, need to design more effective fromAdaptive filter.
DonaldL.Duttweiler has proposed a kind of ratio sef-adapting filter in 2000, famousPNLMS sef-adapting filter. Different from traditional NLMS sef-adapting filter, PNLMS certainlyAdaptive filter is that each coefficient has distributed different gains, the sef-adapting filter that this gain is corresponding with itCoefficient relation in direct ratio. This sef-adapting filter is convergence rate in the time approaching the unknown system of high degree of rareficationVery fast, but along with the decline of coefficient of combination degree, constringency performance also with under will. Rare in order to improve estimationThe performance of thin unknown system, JacobBenesty and StevenL.Gay have proposed a kind of improved ratioExample sef-adapting filter, i.e. IPNLMS sef-adapting filter. Due to its superior performance, simple in structure,This ratio sef-adapting filter is widely applied.
Two important performance indications of sef-adapting filter are convergence rate and stable state imbalance. AlthoughWhen IPNLMS sef-adapting filter is used for estimating sparse unknown system, its convergence rate is very fast, but itsThe performance of stable state imbalance is weakened, i.e. stable state imbalance has very large fluctuation. The fluctuation of stable state imbalanceGreatly, illustrate in a lot of time point stable states imbalances very large, and non-in other a lot of time point stable states imbalancesOften little. The stable state very large time point of lacking of proper care, the non-constant of its precision. Therefore, in order to obtain high precision,Need to find an effective solution.
Summary of the invention
The object of the invention is to provide a kind of variable element ratio sef-adapting filter, has solved IPNLMS certainlyThe fluctuation of adaptive filter a lot of time point stable state imbalances because convergence rate causes very is soon very largeProblem.
In order to solve these problems of the prior art, technical scheme provided by the invention is as follows:
A kind of variable element ratio sef-adapting filter, is characterized in that described wave filter comprises:
Noise power estimation module, for when sef-adapting filter estimating system noise in static timePower;
Error power estimation module, for carrying out time smoothing to the output error signal of sef-adapting filterThe power of estimation error signal;
Intermediate variable generation module, in the middle of producing for the power of the power by error signal and system noiseVariable, described intermediate variable is by the power of error signal and the power ratio of system noise, through asking logarithm to obtain;
Time-varying parameter generation module, for intermediate variable is changed by Sigmoid function, obtainsFor the time-varying parameter of ratio sef-adapting filter;
Ratio matrix builds module, for asked for the gain of each coefficient by the time-varying parameter obtaining, then byCoefficient gain structure ratio matrix;
Filter coefficient update module, for carrying out sef-adapting filter according to the ratio matrix buildingCoefficient update, and calculate new error signal value.
Preferred technical scheme is: described noise power estimation module is carried out the power of estimating system noiseSchilling input signal u (n)=0, output error e (n) is system noise v (n); Put down by the timeEqual method, tries to achieve the power of system noise
Preferred technical scheme is: described error power estimation module is carried out error power estimation according to as followsStep is carried out:
1) pass through input signal u (n) and desired signal d (n) according to e (n)=d (n)-wT(n) u (n) calculates mistakeThe value of difference signal, wherein w (n)=[w1(n),w2(n),…,wM(n)] be sef-adapting filter the coefficient in n moment toAmount; U (n)=[u (n), u (n-1) ..., u (n-M+1)]TFor sef-adapting filter is at the input signal vector in n moment,This vector is made up of the current sample value of input signal and M-1 sampling value before it;
2) according toEstimate the power of output error signalWhereinλ is smoothing factor.
Preferred technical scheme is: described intermediate variable generation module is according to system noise power and error letterNumber power according toObtain intermediate variable x (n).
Preferred technical scheme is: described time-varying parameter generation module according to intermediate variable x (n) according toα (n)=(2 α+2)/{ 1+exp[-β x (n)] }-(α+2) obtain the value of time-varying parameter α (n), and wherein α is compromise ginsengNumber; β is the form parameter of Sigmoid function.
Preferred technical scheme is: described ratio matrix build module first according to time-varying parameter according togm(n)＝[1-α(n)]/2M+[1+α(n)]|wm(n)|/[2||w(n)||1+ ε] obtain ratio entry of a matrix element, whereinm＝1,2,…，M-1，wm(n) be m coefficient of sef-adapting filter in the value in n moment, || ||1RepresentL1Norm, ε is the little positive number of introducing; Then the M obtaining a ratio entry of a matrix element formed to diagonal angleMatrix G (n)=diag[g1(n),g2(n),…,gM(n)], wherein in G (n) each diagonal element corresponding to each filterThe gain g of ripple device coefficientm(n)。
Preferred technical scheme is: described filter coefficient update module according to the ratio matrix generating according toMore new formula w (n+1)=w (n)+μ G (n) u (n) e (n)/[uT(n) G (n) u (n)+δ] upgrade the coefficient of sef-adapting filterVector, wherein δ is with the regularization parameter that solves numerical computations difficulty.
Another object of the present invention is to provide a kind of variable element ratio adaptive filter coefficient vector moreNew method, is characterized in that said method comprising the steps of:
(1) when the power of sef-adapting filter estimating system noise in static time;
(2) output error signal of sef-adapting filter is carried out to the merit of time smoothing estimation error signalRate;
(3) produce intermediate variable by the power of error signal and the power of system noise, described middle changeAmount is by the power of error signal and the power ratio of system noise, through asking logarithm to obtain;
(4) intermediate variable is changed by Sigmoid function, obtained filtering for ratio self adaptationThe time-varying parameter of ripple device;
(5) asked for the gain of each coefficient by the time-varying parameter obtaining, then by coefficient gain structure ratioMatrix;
(6) carry out the coefficient update of sef-adapting filter according to the ratio matrix building, and calculateNew error signal value.
Preferred technical scheme is: described method is specifically carried out in accordance with the following steps:
(1) make input signal u (n)=0, output error e (n) is system noise v (n); Pass through the timeAverage method, tries to achieve the power of system noise
(2) carrying out error power estimation carries out in accordance with the following steps:
1) pass through input signal u (n) and desired signal d (n) according to e (n)=d (n)-wT(n) u (n) calculates mistakeThe value of difference signal, wherein w (n)=[w1(n),w2(n),…,wM(n)] be sef-adapting filter the coefficient in n moment toAmount; U (n)=[u (n), u (n-1) ..., u (n-M+1)]TFor sef-adapting filter is at the input signal vector in n moment,This vector is made up of the current sample value of input signal and M-1 sampling value before it;
2) according toEstimate the power of output error signalWhereinλ is smoothing factor;
(3) according to system noise power and error signal power according toIn the middle of obtainingVariable x (n);
(4) while acquisition according to α (n)=(2 α+2)/{ 1+exp[-β x (n)] }-(α+2) according to intermediate variable x (n), becomeThe value of parameter alpha (n), wherein α is compromise parameter; β is the form parameter of Sigmoid function;
(5) first according to time-varying parameter according to gm(n)＝[1-α(n)]/2M+[1+α(n)]|wm(n)|/[2||w(n)||1+ε]Obtain ratio entry of a matrix element, wherein m=1,2 ..., M-1, wm(n) be m system of sef-adapting filterCount the value in the n moment, || ||1Represent L1Norm, ε is the little positive number of introducing; Then by the M obtainingRatio entry of a matrix element forms diagonal matrix G (n)=diag[g1(n),g2(n),…,gM(n)], wherein every in G (n)Individual diagonal element is corresponding to the gain g of each filter coefficientm(n)；
(6) according to the ratio matrix generating according to new formula morew(n+1)＝w(n)+μG(n)u(n)e(n)/[uT(n) G (n) u (n)+δ] upgrade the coefficient vector of sef-adapting filter, itsMiddle δ is with the regularization parameter that solves numerical computations difficulty.
The principle of technical solution of the present invention is:
The present invention adopts variable element ratio method to carry out the adjustment of adaptive filter coefficient vector, usesA time-varying parameter is adjusted the value of the coefficient gain of sef-adapting filter, and this time-varying parameter can be expressed as mistakeThe monotonically increasing function of difference signal power and system noise power ratio.
In the starting stage of sef-adapting filter operation, due to error signal power and system noise power ratioBe worth greatlyr, thereby sef-adapting filter be automatically that large coefficient distributes larger yield value, thereby keeps soonConvergence rate; In the converged state of sef-adapting filter, due to error signal power and system noise meritRate ratio is less, thereby sef-adapting filter be automatically the large less yield value of coefficient distribution, thereby fallsThe fluctuation of low stable state imbalance.
With respect to scheme of the prior art, advantage of the present invention is:
Technical solution of the present invention variable element ratio sef-adapting filter, belongs to digital filter design field,Adopt variable element method to adjust the value of the coefficient gain of sef-adapting filter with a time-varying parameter,The fast convergence rate of ratio sef-adapting filter can be kept, ratio sef-adapting filter can be obtained again lowStable state imbalance fluctuation. It is equal that the present invention can be applied to hands-free phone, video conference, audiphone, channelIn the equipment such as weighing apparatus, electronic lancet.
Brief description of the drawings
Fig. 1 is variable element ratio sef-adapting filter structure principle chart;
Fig. 2 is the unknown system impulse response that comprises 100 coefficients;
Fig. 3 is the unknown system impulse response that comprises 512 coefficients;
Fig. 4 is that sef-adapting filter is under 20dB signal to noise ratio condition when the unknown system shown in drawing for estimate 2The comparison of normalization imbalance curve, the value of its parameter is μ=0.3, β=3;
Fig. 5 is that sef-adapting filter is under 30dB signal to noise ratio condition when the unknown system shown in drawing for estimate 2The comparison of normalization imbalance curve, the value of its parameter is μ=0.5, β=1.5;
Fig. 6 is that sef-adapting filter is under 40dB signal to noise ratio condition when the unknown system shown in drawing for estimate 2The comparison of normalization imbalance curve, the value of its parameter is μ=0.7, β=1;
Fig. 7 is that sef-adapting filter is under 20dB signal to noise ratio condition when the unknown system shown in drawing for estimate 3The comparison of normalization imbalance curve, the value of its parameter is μ=0.3, β=7;
Fig. 8 is that sef-adapting filter is under 30dB signal to noise ratio condition when the unknown system shown in drawing for estimate 3The comparison of normalization imbalance curve, the value of its parameter is μ=0.5, β=5;
Fig. 9 is that sef-adapting filter is under 40dB signal to noise ratio condition when the unknown system shown in drawing for estimate 3The comparison of normalization imbalance curve, the value of its parameter is μ=0.7, β=3.
Detailed description of the invention
Below in conjunction with specific embodiment, such scheme is described further. Should be understood that these embodiment areBe used for the present invention is described and do not limit the scope of the invention. The implementation condition adopting in embodiment can basisThe condition of concrete system is done further adjustment, and not marked implementation condition is generally the bar in normal experimentPart.
Embodiment variable element ratio sef-adapting filter example
As shown in Figure 1, this variable element ratio sef-adapting filter, comprising:
Noise power estimation module: the effect of this module is when sef-adapting filter is in static timeThe power of estimating system noise;
Error power estimation module: the effect of this module is the logical output error signal to sef-adapting filterCarry out time smoothing and carry out the power of estimation error signal;
Intermediate variable generation module: the effect of this module be produce an intermediate variable, this intermediate variable byThe power ratio of error power and system noise, then ask logarithm to obtain;
Time-varying parameter generation module: the effect of this module is that intermediate variable is undertaken by Sigmoid functionConversion, obtains having the variable element that the ratio adaptive-filtering of good filter effect requires;
Ratio matrix builds module: the effect of this module is first to ask for each coefficient by the variable element obtainingGain, so then by coefficient gain structure ratio matrix;
Filter coefficient update module: the effect of this module is to carry out adaptive according to the ratio matrix being divided intoAnswer the coefficient update of wave filter, and calculate new error signal value.
Variable element ratio sef-adapting filter modules is concrete to be moved in accordance with the following steps:
Step 1. is before sef-adapting filter iteration is upgraded, and " noise power estimation module " estimates systemThe variance of system noise. While estimating this variance, make input signal u (n)=0, output error e (n) is and isSystem noise v (n). By time averaging method, try to achieve the power of system noiseEnter afterwards adaptiveAnswer the wave filter self adaptation stage.
Step 2. " sef-adapting filter " is by input signal u (n) and desired signal d (n) error of calculationThe value of signal, its computing formula is e (n)=d (n)-wT(n) u (n), wherein w (n)=[w1(n),w2(n),…,wM(n) be]Sef-adapting filter is at the coefficient vector in n moment; U (n)=[u (n), u (n-1) ..., u (n-M+1)]TFor self adaptationWave filter is at the input signal vector in n moment, and this vector is by the current sample value of input signal and before itM-1 sampling value forms.
Step 3. " error power estimation module " is estimated the power of output error signalIts estimationMethod is to use following computing formula:Wherein λ is smoothing factor,Generally value between 0.9 to 0.999, unknown system length is longer, and λ value is larger; Otherwise λ getsBe worth less.
Step 4. " intermediate variable generation module ", by the system noise obtaining in step 1 and step 2Acoustical power and error signal power calculation obtain intermediate variable, and its computing formula is
Step 5. " time-varying parameter generation module " becomes while using the intermediate variable obtaining in step 3 to calculateThe value of parameter alpha (n). This module is selected monotonic increase Sigmoid function, and has carried out Pan and Zoom.Through the function expression after Pan and Zoom be: α (n)=(2 α+2)/1+exp[-β x (n)] }-(α+2), itsMiddle α is compromise parameter, and its good value is 0 or-0.5; β is the shape ginseng of Sigmoid functionNumber, this parameter has been determined the slope of Sigmoid function waveform, its good span is 1 to 10Between. When signal to noise ratio is lower, or the coefficient vector of unknown system is longer, or signal correlation is higherTime, β should get higher value; Otherwise β should get smaller value.
Step 6. " ratio matrix builds module " is first used the time-varying parameter obtaining in step 4 to calculateRatio entry of a matrix element,gm(n)＝[1-α(n)]/2M+[1+α(n)]|wm(n)|/[2||w(n)||1+ ε], m=1,2 ..., M-1, wherein wm(n) be adaptiveAnswer m coefficient of wave filter in the value in n moment, || ||1Represent L1 norm, ε is for being used for overcoming numerical valueDyscalculia and the little positive number introduced. Then, this module is by the M being calculated by above-mentioned computing formulaIndividual element forms diagonal matrix G (n)=diag[g1(n),g2(n),…,gM(n)], each diagonal element in G (n) whereinElement is corresponding to the ratio step-length g of each filter coefficientm(n)。
Step 7. " filter coefficient update module " is upgraded by the ratio matrix generating in step 6The coefficient vector of sef-adapting filter, its more new formula bew(n+1)＝w(n)+μG(n)u(n)e(n)/[uT(n) G (n) u (n)+δ], wherein δ is with solving numerical computations difficultyRegularization parameter.
Application examples adopts variable element ratio sef-adapting filter to carry out System Discrimination application
(variable element ratio of the present invention is adaptive to use the disclosed variable element ratio of embodiment sef-adapting filterAnswer wave filter referred to as VIPNLMS) sef-adapting filter distinguishes respectively two sparse unknown systems,And the performance of its performance and NLMS and IPNLMS sef-adapting filter is compared.
As shown in Figure 2, its coefficient vector length is 100 to first unknown system; Second unknown systemAs shown in Figure 3, its coefficient vector length is 512. In the time of the unknown system shown in identification Fig. 2, this realityExecute example and adopt the autoregression model on 2 rank as input, i.e. this input byU (n)=0.40u (n-1)-0.40u (n-2)+θ (n) obtains, and wherein θ (n) is Gaussian sequence;
In the time of the unknown system shown in identification Fig. 3, adopt the autoregression model on 1 rank as input, shouldInput is obtained by u (n)=0.9u (n-1)+η (n), and wherein η (n) is Gaussian sequence. By one and defeatedEnter the input that the incoherent white Gaussian noise of signal is added to adaptive filter system, as system noiseSound, thereby the signal to noise ratio of formation 20dB, 30dB or 40dB. The canonical of NLMS sef-adapting filterChanging parameter is taken asAnd the regularization parameter of IPNLMS and VIPNLMS sef-adapting filter is gotForUse normalization imbalance (NormalizedMisalignment) with respect to iterationThe function of number of times (IterationNumber) comes the performance of three kinds of sef-adapting filters of comparison, its definitionFormula is 20log10||w0-w(n)||/||w0||, unit is decibel (dB).
Experimental result as shown in Fig. 4 to Fig. 9, Fig. 4 to Fig. 6 be sef-adapting filter respectively 20dB,Imbalance curve comparison when unknown system under 30dB, 40dB signal to noise ratio condition shown in drawing for estimate 2; Figure7 to Fig. 9 is sef-adapting filter drawing for estimate 3 under 20dB, 30dB, 40dB signal to noise ratio condition respectivelyImbalance curve comparison when shown unknown system.
From experimental result:
1) convergence rate of variable element ratio sef-adapting filter disclosed by the invention is adaptive faster than NLMSAnswer wave filter, and the fluctuation of stable state imbalance is suitable with NLMS sef-adapting filter.
2) convergence rate and the IPNLMS of variable element ratio sef-adapting filter disclosed by the invention are adaptiveAnswer the convergence rate of wave filter suitable, and the fluctuation of stable state imbalance is filtered far below IPNLMS self adaptationThe fluctuation of ripple device stable state imbalance. Therefore the performance of variable element ratio sef-adapting filter is better than NLMSWith IPNLMS sef-adapting filter.
Above-described embodiment is only explanation technical conceive of the present invention and feature, and its object is to allow is familiar with thisThe people of technology can understand content of the present invention and implement according to this, can not limit guarantor of the present invention with thisProtect scope. All equivalent transformations that Spirit Essence does according to the present invention or modification, all should be encompassed in the present inventionProtection domain within.
Claims (3)
1. a variable element ratio sef-adapting filter, is characterized in that described wave filter comprises:
Noise power estimation module, for the power when sef-adapting filter estimating system noise in static time;
Error power estimation module, for carrying out time smoothing estimation error signal to the output error signal of sef-adapting filterPower;
Intermediate variable generation module, produces intermediate variable for the power of the power by error signal and system noise, described inBetween variable by the power of error signal and the power ratio of system noise, through ask logarithm obtain;
Time-varying parameter generation module, for intermediate variable is changed by Sigmoid function, obtains for ratio adaptiveAnswer the time-varying parameter of wave filter;
Ratio matrix builds module, for asked for the gain of each coefficient by the time-varying parameter obtaining, then is built by coefficient gainRatio matrix;
Filter coefficient update module, for carry out the coefficient update of sef-adapting filter according to the ratio matrix building, andAnd calculate new error signal value;
Described noise power estimation module is carried out the power of estimating system noiseSchilling input signal u (n)=0, outputError e (n) is system noise v (n); By time averaging method, try to achieve the power of system noise
Described error power estimation module is carried out error power estimation and is carried out in accordance with the following steps:
1) pass through input signal u (n) and desired signal d (n) according to e (n)=d (n)-wT(n) u (n) error signalValue, wherein w (n)=[w1(n),w2(n),…,wM(n)] be the coefficient vector of sef-adapting filter in the n moment;u(n)＝[u(n),u(n-1),…,u(n-M+1)]TFor sef-adapting filter is at the input signal vector in n moment, this vector is by inputtingThe sample value that signal is current and M-1 sampling value before it form;
2) according toEstimate the power of output error signalWherein λ is for flatThe sliding factor;
Described intermediate variable generation module according to system noise power and error signal power according to?To intermediate variable x (n);
Described time-varying parameter generation module according to intermediate variable x (n) according toα (n)=(2 α+2)/{ 1+exp[-β x (n)] }-(α+2) obtain the value of time-varying parameter α (n), and wherein α is compromise parameter; β isThe form parameter of Sigmoid function;
Described ratio matrix build module first according to time-varying parameter according togm(n)＝[1-α(n)]/2M+[1+α(n)]|wm(n)|/[2||w(n)||1+ ε] obtain ratio entry of a matrix element, whereinm＝1,2,…,M-1，wm(n) be m coefficient of sef-adapting filter in the value in n moment, || ||1Represent L1Norm,ε is the little positive number of introducing; Then the M obtaining a ratio entry of a matrix element formed to diagonal matrixG(n)＝diag[g1(n),g2(n),…,gM(n)], wherein in G (n) each diagonal element corresponding to the increasing of each filter coefficientBenefit gm(n)。
2. variable element ratio sef-adapting filter according to claim 1, is characterized in that described filter coefficient moreNew module according to the ratio matrix generating according to new formula more
w(n+1)＝w(n)+μG(n)u(n)e(n)/[uT(n) G (n) u (n)+δ] upgrade the coefficient vector of sef-adapting filter, wherein δ isWith the regularization parameter that solves numerical computations difficulty.
3. a variable element ratio adaptive filter coefficient vector update method, is characterized in that described method comprises following stepRapid:
(1) when the power of sef-adapting filter estimating system noise in static time;
(2) output error signal of sef-adapting filter is carried out to the power of time smoothing estimation error signal;
(3) produce intermediate variable by the power of error signal and the power of system noise, described intermediate variable is by error signalThe power ratio of power and system noise, through asking logarithm to obtain;
(4) intermediate variable is changed by Sigmoid function, obtained for ratio sef-adapting filter time-varying parameter;
(5) asked for the gain of each coefficient by the time-varying parameter obtaining, then by coefficient gain structure ratio matrix;
(6) carry out the coefficient update of sef-adapting filter according to the ratio matrix building, and calculate new error signal value;
Described method is specifically carried out in accordance with the following steps:
(1) make input signal u (n)=0, output error e (n) is system noise v (n); By time averaging method,Try to achieve the power of system noise
(2) carrying out error power estimation carries out in accordance with the following steps:
1) pass through input signal u (n) and desired signal d (n) according to e (n)=d (n)-wT(n) u (n) error signalValue, wherein w (n)=[w1(n),w2(n),…,wM(n)] be the coefficient vector of sef-adapting filter in the n moment;u(n)＝[u(n),u(n-1),…,u(n-M+1)]TFor sef-adapting filter is at the input signal vector in n moment, this vector is by inputtingThe sample value that signal is current and M-1 sampling value before it form;
2) according toEstimate the power of output error signalWherein λ is level and smoothThe factor;
(3) according to system noise power and error signal power according toObtain intermediate variable x (n);
(4) obtain time-varying parameter α (n) according to intermediate variable x (n) according to α (n)=(2 α+2)/{ 1+exp[-β x (n)] }-(α+2)Value, wherein α for compromise parameter, β is the form parameter of Sigmoid function;
(5) first according to time-varying parameter according to gm(n)＝[1-α(n)]/2M+[1+α(n)]|wm(n)|/[2||w(n)||1+ ε] obtainGet ratio entry of a matrix element, wherein m=1,2 ..., M-1, wm(n) be that m coefficient of sef-adapting filter is in the n momentValue, || ||1Represent L1Norm, ε is the little positive number of introducing; Then the M obtaining a ratio entry of a matrix element formed to diagonal matrixG(n)＝diag[g1(n),g2(n),…,gM(n)], wherein in G (n) each diagonal element corresponding to the increasing of each filter coefficientBenefit gm(n)；
(6) according to the ratio matrix generating according to new formula more
w(n+1)＝w(n)+μG(n)u(n)e(n)/[uT(n) G (n) u (n)+δ] upgrade the coefficient vector of sef-adapting filter, wherein δ is for usingSolve the regularization parameter of numerical computations difficulty.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410015022.5A CN103716013B (en) | 2014-01-14 | 2014-01-14 | Variable element ratio sef-adapting filter |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410015022.5A CN103716013B (en) | 2014-01-14 | 2014-01-14 | Variable element ratio sef-adapting filter |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103716013A CN103716013A (en) | 2014-04-09 |
| CN103716013B true CN103716013B (en) | 2016-05-04 |
Family
ID=50408674
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410015022.5A Expired - Fee Related CN103716013B (en) | 2014-01-14 | 2014-01-14 | Variable element ratio sef-adapting filter |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103716013B (en) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104202018B (en) * | 2014-08-22 | 2017-11-28 | 华为技术有限公司 | The method and apparatus for determining the stable factor of sef-adapting filter |
| CN104363003A (en) * | 2014-10-27 | 2015-02-18 | 常州听觉工坊智能科技有限公司 | Filtering method and device |
| DK3311591T3 (en) * | 2015-06-19 | 2021-11-08 | Widex As | PROCEDURE FOR OPERATING A HEARING AID SYSTEM AND A HEARING AID SYSTEM |
| CN105871762B (en) * | 2016-05-23 | 2018-10-12 | 苏州大学 | A kind of adaptive network for the estimation of Sparse parameter vector |
| CN108200522B (en) * | 2017-11-24 | 2020-02-18 | 华侨大学 | Regularization proportion normalization subband self-adaptive filtering method |
| CN111313865A (en) * | 2018-12-12 | 2020-06-19 | 哈尔滨工业大学 | Step length regularization method for random Fourier feature kernel LMS algorithm |
| CN110459197B (en) * | 2019-07-10 | 2021-09-24 | 哈尔滨工业大学（深圳） | Signal enhancer and method for denoising and extracting weak blind signals |
| CN110767245B (en) * | 2019-10-30 | 2022-03-25 | 西南交通大学 | Voice communication self-adaptive echo cancellation method based on S-shaped function |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101283514A (en) * | 2005-08-15 | 2008-10-08 | 捷讯研究有限公司 | Joint space-time optimum filters (ISTOF) for interference cancellation |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2524719B2 (en) * | 1986-11-12 | 1996-08-14 | 富士通株式会社 | Space-time adaptive filter |
-
2014
- 2014-01-14 CN CN201410015022.5A patent/CN103716013B/en not_active Expired - Fee Related
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101283514A (en) * | 2005-08-15 | 2008-10-08 | 捷讯研究有限公司 | Joint space-time optimum filters (ISTOF) for interference cancellation |
Non-Patent Citations (1)
| Title |
|---|
| 变设计参数子带自适应滤波器研究;倪锦根;《中国博士学位论文全文数据库》;20111215;第77页-第86页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103716013A (en) | 2014-04-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103716013B (en) | Variable element ratio sef-adapting filter | |
| CN107168071B (en) | A kind of nonlinear system Auto-disturbance-rejection Control based on interference observer | |
| CN104616667B (en) | A kind of active denoising method in automobile | |
| CN109257030B (en) | Variable step length lpSparse system identification method of norm LMS algorithm | |
| CN108200522B (en) | Regularization proportion normalization subband self-adaptive filtering method | |
| CN104283528A (en) | Variable-step LMS adaptive filtering method | |
| CN112017679B (en) | Method, device and equipment for updating adaptive filter coefficients | |
| CN109977914A (en) | Self-adaptation noise reduction method based on VMD | |
| CN110402540A (en) | Active denoising method, device, chip, active control system and storage medium | |
| DE102017101497B4 (en) | Micro-electro-mechanical system (MEMS) circuit and method for reconstructing a disturbance variable | |
| Galrinho et al. | A weighted least-squares method for parameter estimation in structured models | |
| Rombouts et al. | Robust and Efficient Implementation of the PEM—AFROW Algorithm for Acousic Feedback Cancellation | |
| CN110048694B (en) | Random Fourier feature kernel least mean square algorithm based on variable step length | |
| CN109379652A (en) | A kind of the secondary channel offline identification method and system of earphone Active noise control | |
| CN103929704B (en) | The method and system that a kind of adaptive acoustic feedback based on transform domain is eliminated | |
| CN105070295A (en) | Adaptive method, applied to echo cancellation, of active factor proportional sub band | |
| DE60311862T2 (en) | Equalizer for sample-wise decision feedback in the transformation domain | |
| Saengmuang et al. | Convergence and Stability Analysis of Spline Adaptive Filtering based on Adaptive Averaging Step-size Normalized Least Mean Square Algorithm | |
| CN112886947A (en) | Variable-step robust affine projection adaptive filter | |
| CN111161700A (en) | Main channel time delay estimation method in space noise reduction system | |
| Gannot et al. | On the application of the unscented Kalman filter to speech processing | |
| CN102724152A (en) | Multinomial adaptive active noise cancellation method based on Laguerre structure | |
| CN113299284B (en) | Speech recognition device, method, equipment and medium based on adaptive filtering | |
| CN112866149B (en) | Adaptive estimation method based on bias compensation block sparse proportional NLMS | |
| CN113676157B (en) | Self-adaptive filtering algorithm for counteracting direct current bias |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20160504 Termination date: 20190114 |