CN113655436A - Method and device for optimizing broadband beam forming by particle swarm with channel calibration - Google Patents
Method and device for optimizing broadband beam forming by particle swarm with channel calibration Download PDFInfo
- Publication number
- CN113655436A CN113655436A CN202110839718.XA CN202110839718A CN113655436A CN 113655436 A CN113655436 A CN 113655436A CN 202110839718 A CN202110839718 A CN 202110839718A CN 113655436 A CN113655436 A CN 113655436A
- Authority
- CN
- China
- Prior art keywords
- calibration
- frequency
- particle swarm
- compensation
- channel
- 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.)
- Granted
Links
- 239000002245 particle Substances 0.000 title claims abstract description 95
- 238000000034 method Methods 0.000 title claims abstract description 69
- 238000005457 optimization Methods 0.000 claims abstract description 73
- 239000011159 matrix material Substances 0.000 claims abstract description 45
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 42
- 230000001629 suppression Effects 0.000 claims abstract description 32
- 230000003595 spectral effect Effects 0.000 claims abstract description 17
- 238000001914 filtration Methods 0.000 claims abstract description 13
- 239000013598 vector Substances 0.000 claims description 57
- 230000003044 adaptive effect Effects 0.000 claims description 29
- 238000005070 sampling Methods 0.000 claims description 17
- 238000010606 normalization Methods 0.000 claims description 10
- 230000009467 reduction Effects 0.000 claims description 10
- 238000001228 spectrum Methods 0.000 claims description 10
- 230000006870 function Effects 0.000 claims description 7
- 230000003321 amplification Effects 0.000 claims description 4
- 238000004590 computer program Methods 0.000 claims description 4
- 238000003199 nucleic acid amplification method Methods 0.000 claims description 4
- 238000012937 correction Methods 0.000 claims description 3
- 230000001934 delay Effects 0.000 claims description 2
- 230000017105 transposition Effects 0.000 claims description 2
- 238000010586 diagram Methods 0.000 description 7
- 238000012545 processing Methods 0.000 description 6
- 238000004891 communication Methods 0.000 description 4
- 238000013461 design Methods 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 2
- 239000011541 reaction mixture Substances 0.000 description 2
- 230000005236 sound signal Effects 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 238000005311 autocorrelation function Methods 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 125000004432 carbon atom Chemical group C* 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000005316 response function Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S3/00—Direction-finders for determining the direction from which infrasonic, sonic, ultrasonic, or electromagnetic waves, or particle emission, not having a directional significance, are being received
- G01S3/80—Direction-finders for determining the direction from which infrasonic, sonic, ultrasonic, or electromagnetic waves, or particle emission, not having a directional significance, are being received using ultrasonic, sonic or infrasonic waves
- G01S3/8003—Diversity systems specially adapted for direction finding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Remote Sensing (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Radar, Positioning & Navigation (AREA)
- Circuit For Audible Band Transducer (AREA)
Abstract
The invention discloses a method and a device for optimizing broadband beam forming by a particle swarm with a channel calibration function. The method comprises the following steps: calculating compensation coefficients of the self-adaptive calibration compensation filters corresponding to the microphone element channels; determining a frequency band range in which a user is interested and a main lobe and a side lobe of an array beam on the frequency band range; calculating an input spectral power autocorrelation matrix in a frequency band range in which a user is interested, and constructing a problem of solving an optimal FIR filter weight coefficient under a linear constraint minimum variance criterion; converting the problem into an unconstrained optimization problem, and solving by using a particle swarm optimization algorithm to obtain a weight coefficient of an optimal FIR filter; summing the digital signals after filtering compensation by using the weight coefficient of the optimal FIR filter to output time domain signals after beam forming; if the incoming wave direction of the target sound source pointed by the wave beam or the requirements of side lobe constraint, interference and noise suppression are changed, the weight coefficient of the FIR filter is recalculated.
Description
Technical Field
The invention relates to a method for optimizing broadband beam forming by a particle swarm with channel calibration, and also relates to a corresponding broadband beam forming device, belonging to the technical field of acoustics.
Background
With the development of array signal processing technology, beam forming (beam forming) method is widely applied in the fields of radar, communication, intelligent voice and the like. Array speech enhancement applications require the formation of relatively uniform wideband beams over the audio frequency range, increasing the array directivity, effectively enhancing the target speech signal, while suppressing interference and various random noises.
The broadband beam forming method can be divided into a fixed beam forming method and a self-adaptive beam forming method, and although the self-adaptive beam forming method fully utilizes the statistical characteristics of array receiving signals, in practical application, particularly when the number of array elements is large, the statistical characteristics are difficult to obtain in real time, the algorithm computation amount is too large, and the application of the self-adaptive beam forming method is limited. The fixed beam forming method does not need real-time updating operation, and the beam is designed on a wide frequency band according to the array characteristics and the performance index requirements, so that the practical value is higher.
In the prior art, a time domain broadband fixed beam former is mainly adopted, and an expected broadband beam meeting constraint conditions is formed by optimizing weight coefficients of each channel filter formed by a beam. The beam former generally adopts a time domain broadband beam forming method with side lobe control, converts the problem of solving beam forming weight coefficients into a convex optimization problem, sets main lobe broadband constraints and side lobe suppression constraints on a broadband, searches for the optimal solution of the convex optimization problem, and can obtain array broadband beams with good space directivity and interference noise suppression. However, this method has the following disadvantages: firstly, the interference and noise types in the actual acoustic environment are complex, and the array processing performance is reduced due to the influence of factors such as scattering noise and strong directional interference sources; secondly, the synthesized wave beam is not accurate enough; thirdly, the robustness of the method is poor, and the mismatching of the amplitude or the phase among array elements can cause the serious reduction of the array beam performance, thereby seriously influencing the practical degree of the array beam.
Disclosure of Invention
The invention aims to solve the primary technical problem of providing a particle swarm optimization broadband beam forming method with channel calibration.
Another technical problem to be solved by the present invention is to provide an optimized wideband beam forming apparatus with channel calibration for particle swarm.
In order to achieve the purpose, the invention adopts the following technical scheme:
according to a first aspect of the embodiments of the present invention, there is provided a wideband beamforming method with channel calibration particle swarm optimization, including the following steps:
step S1, self-adaptive calibration is carried out in the calibration stage, and the compensation coefficients of the self-adaptive calibration compensation filters corresponding to the microphone element channels are calculated;
step S2, determining the frequency band range of interest of the user and the main lobe and the side lobe of the array beam on the frequency band range;
step S3, calculating an input spectrum power autocorrelation matrix in a frequency band range in which a user is interested, and constructing a problem of solving the weight coefficient of the optimal FIR filter under the linear constraint minimum variance criterion;
step S4, converting the problem into an unconstrained optimization problem, and solving by using a particle swarm optimization algorithm to obtain a weight coefficient of the optimal FIR filter;
step S5, summing the digital signals after filtering compensation by the adaptive calibration compensation filter by using the weight coefficient of the optimal FIR filter, and outputting the time domain signals after beam forming; and if the incoming wave direction of the target sound source pointed by the wave beam or the requirements of sidelobe constraint, interference and noise suppression are not changed, the step S5 is repeatedly executed, otherwise, the step S2 is returned, and the weight coefficient of the FIR filter is recalculated.
Preferably, the step S2 includes the following sub-steps:
step S11, collecting the played broadband calibration signal by using a microphone array;
and step S12, taking the first microphone element channel as a reference channel, calculating the error signal target value of the adaptive calibration between each of the rest microphone element channels and the reference channel, and calculating the compensation coefficient of the adaptive calibration compensation filter corresponding to each microphone element channel by adopting a proportional normalization least mean square algorithm.
Preferably, in step S12, the remaining error signal target values of the adaptive calibration between each microphone element channel and the reference channel are represented as:
|en(m)|2＝|x0(m)-sn(m)gn(m)|2
in the above formula, en(m) a digital signal x representing the output of the adaptive calibration compensation filter corresponding to the nth microphone element channeln(m) and a reference target signal x0(m) error between; broadband calibration signal s collected by nth microphone elementn(m)＝[sn(m),...,sn(m-M+1)]TM denotes the sampling point of the first microphone element channel, M denotes the number of delays of the sampling point, gnAnd (m) represents the compensation coefficient corresponding to the current sampling point.
Preferably, when the proportional normalization least mean square algorithm calculates the compensation coefficients of the adaptive calibration compensation filters corresponding to the microphone element channels, the iterative formula adopted is as follows:
in the above formula, gn(m +1) represents the compensation coefficient corresponding to the next sampling point, gn(m) represents a compensation coefficient corresponding to the current sampling point, eta represents a corrected step constant, and delta is a smaller integer, so that the stability is prevented from being reduced due to the fact that the step constant eta is too large because the inner product of the input vector is too small; t represents transposition; g (m +1) ═ diag { beta [ ]1(m+1),β2(m+1),...,βM(m +1) } is the step size control matrix, diag { } denotes the diagonal matrix, and β (m +1) denotes the element in the step size control matrix.
Preferably, the element β (m +1) in the step control matrix is obtained according to the following formula;
compensation coefficient feedback value xil(m+1)＝max{ρmax{υ,|g1(m)|,...,|gM(m)|},|gl(m) is used for preventing iteration from being invalid due to too small compensation coefficient, upsilon is a correction value for preventing the compensation coefficients from being all zero, and rho is a compensation coefficient feedback scale factor.
Preferably, the step S2 includes the following sub-steps:
step S21, determining the frequency band range of interest of the user, dividing the frequency band range into K +1 frequency points, selecting R +1 reference frequency points from the frequency points, and calculating the frequency values of the related frequency points;
step S22, determining a reference beam width corresponding to the frequency reference value of each reference frequency point, and performing interpolation operation on the reference beam width to obtain beam main lobe widths corresponding to the frequency values of all frequency points within a frequency band range in which a user is interested;
and step S23, determining a side lobe region and a main lobe beam corresponding to the frequency values of all frequency points in the frequency band range interested by the user according to the main lobe width of the beam, and finely dividing the side lobe region to apply constraint to obtain side lobe constraint direction values on all frequency points.
Preferably, the step S3 includes the following sub-steps:
step S31, dividing N again in the frequency band range of interest to the userf+1 frequency points, and calculating the frequency value of each frequency point;
step S32, calculating the spectral power autocorrelation matrixes of the target sound source, the uncorrelated background noise, the scattering noise component and the directional interference respectively;
and step S33, summing and regularizing the spectral power autocorrelation matrixes of the target sound source, the uncorrelated background noise, the scattering noise component and the directional interference to obtain an input spectral power autocorrelation matrix, and constructing a problem of solving the weight coefficient of the optimal FIR filter under the linear constraint minimum variance criterion.
Preferably, in step S33, the following constraint is satisfied,constructing the problem of solving the weight coefficient w of the optimal FIR filter under the linear constraint minimum variance criterion:
constraint conditions
In the above formula, Vs＝[vs(0),vs(1),...,vs(K)]A matrix composed of K +1 frequency point target signal guide vectors divided in a frequency band range in which a user is interested,
representing a K +1 dimensional distortion-free constraint vector, fkRepresenting the frequency value f corresponding to each frequency point in K +1 frequency pointssRepresenting a digital audio sampling frequency;
the matrix is composed of E-th sidelobe constraint guide vectors of K +1 frequency points, wherein | · | | represents the Euclidean norm, epsilonslAnd εnParameters of the side lobe suppression degree and white noise gain amplification are respectively expressed.
Preferably, the step S4 includes the following sub-steps:
step S41, converting the problem of the constructed optimal FIR filter weight coefficient meeting the constraint condition into an unconstrained optimization problem:
where Ω (w) represents the objective function of the optimization problem, w represents the weight coefficients of the FIR filter, RxRepresenting the input spectral power autocorrelation matrix, NslRepresenting the number of side lobe constraints, and lambda represents a positive weight coefficient;
step S42, setting relevant parameters of the particle swarm optimization algorithm and initializing in a solution spaceEach population particle and setting the position vector and velocity vector of each population particle, respectively
And
np is the number of the population particles, Q is the serial number of the particles, and t represents the iteration times of the particle swarm optimization algorithm;
step S43, order
Substituting the position vectors of all the population particles into the unconstrained optimization problem to calculate an objective function
And searching and finding out the self historical optimal solution vector of the population particles
And historical optimal solution vectors for all population particles
t' represents the iteration times of the particle swarm optimization algorithm;
step S44, when the iteration time T of the particle swarm optimization algorithm reaches the preset maximum iteration time TmaxIf so, terminating the iteration of the particle swarm optimization algorithm to obtain the weight coefficient of the optimal FIR filter
And performs step S5; otherwise, updating the velocity vector and the position vector of the particle swarm, and returning to the step S43 to continue the iteration.
Preferably, in step S44, when the iteration number T of the particle swarm optimization algorithm does not reach the preset maximum iteration number TmaxAccording to the optimal solution vector of the population particles of the iteration
And
updating the velocity vector and the position vector of the particle swarm:
in the above formula, the inertia factor ω and the learning factor
Selecting [0,1 ] according to specific application]The number of intervals.
According to a second aspect of embodiments of the present invention, there is provided a particle swarm optimization wideband beamforming device with channel calibration, comprising a processor and a memory, wherein the processor reads a computer program or instructions in the memory to perform the following operations:
step S1, self-adaptive calibration is carried out in the calibration stage, and the compensation coefficients of the self-adaptive calibration compensation filters corresponding to the microphone element channels are calculated;
step S2, determining the frequency band range of interest of the user and the main lobe and the side lobe of the array beam on the frequency band range;
step S3, calculating an input spectrum power autocorrelation matrix in a frequency band range in which a user is interested, and constructing a problem of solving the weight coefficient of the optimal FIR filter under the linear constraint minimum variance criterion;
step S4, converting the problem into an unconstrained optimization problem, and solving by using a particle swarm optimization algorithm to obtain a weight coefficient of the optimal FIR filter;
step S5, summing the digital signals after filtering compensation by the adaptive calibration compensation filter by using the weight coefficient of the optimal FIR filter, and outputting the time domain signals after beam forming; and if the incoming wave direction of the target sound source pointed by the wave beam or the requirements of sidelobe constraint, interference and noise suppression are not changed, the step S5 is repeatedly executed, otherwise, the step S2 is returned, and the weight coefficient of the FIR filter is recalculated.
Compared with the prior art, the method and the device for optimizing the broadband beam forming by the particle swarm with the channel calibration function have the following characteristics:
(1) the beam design with beam control and complex interference noise suppression is considered, the strong directional interference, the scattering noise and the white noise gain are restrained in a targeted manner, the strong directional interference can be suppressed obviously under the condition of meeting the requirements of the main lobe width and the side lobe level, and the output signal-to-noise ratio is improved.
(2) Aiming at the optimization problem, a particle swarm optimization method is adopted, the optimal solution of the weight coefficient of the FIR filter can be rapidly and accurately obtained, the expected beam is accurately synthesized, and the beam directivity and the interference noise suppression capability are effectively improved.
(3) In the calibration stage, the compensation coefficient of the calibration compensation filter is obtained by using the adaptive filtering framework and the proportional normalization least mean square algorithm, the performance reduction of an array beam caused by the consistency difference between microphone element channels is eliminated, the effectiveness and the robustness of the technical method are obviously improved, and the method has higher practical value.
Drawings
Fig. 1 is a flowchart of optimizing broadband beam forming by using a particle swarm with channel calibration according to an embodiment of the present invention;
fig. 2 is a schematic block diagram of a wideband beamforming method with channel calibration particle swarm optimization according to an embodiment of the present invention;
fig. 3 is a schematic block diagram of adaptive calibration compensation in a wideband beamforming method with channel calibration for particle swarm optimization according to an embodiment of the present invention;
fig. 4 is a schematic diagram of comparing beam patterns of a wideband beam forming method with channel calibration particle swarm optimization according to an embodiment of the present invention with beam patterns of other beam forming methods;
fig. 5 is a schematic diagram of a comparison between a beam pattern formed by a broadband beam forming method with channel calibration particle swarm optimization and a beam pattern formed by a calibration beam without a channel provided in an embodiment of the present invention;
fig. 6a to 6c are schematic diagrams illustrating comparison between output signal waveforms and speech spectrograms of the method for calibrating particle swarm optimization broadband beam forming with channels according to the embodiment of the present invention and the existing beam forming method;
fig. 7 is a schematic structural diagram of a wideband beam forming apparatus with channel calibration and particle swarm optimization according to an embodiment of the present invention.
Detailed Description
The technical contents of the present invention will be further described in detail with reference to the accompanying drawings and specific embodiments.
In order to obtain a multi-channel filtering weight coefficient for beamforming, implement the optimal desired waveform satisfying constraint conditions, and significantly improve beam space directivity, specific interference noise suppression capability and robustness, as shown in fig. 1, an embodiment of the present invention provides a method for optimizing wideband beamforming by using particle swarm with channel calibration, including the following steps:
step S1, performing adaptive calibration in the calibration stage, and calculating the compensation coefficients of the adaptive calibration compensation filters corresponding to the microphone element channels.
The method comprises the following substeps:
and step S11, collecting the played broadband calibration signal by adopting a microphone array.
In the embodiment of the invention, the played broadband calibration signal is acquired by adopting the linear microphone array consisting of N omnidirectional microphone arrays. The linear microphone array is distributed on a z coordinate axis of a rectangular coordinate system, each microphone element serves as a microphone element channel, the microphone element channel is represented by a serial number N, wherein N is 0,1, the
d represents the spacing of the microphone elements.
In the calibration stage, the played broadband calibration signal may be a speech or broadband noise signal, which serves as a calibration input sound source having an incident direction at a 90 ° vertical incident angle with respect to the linear microphone array, and the sound signal may be considered to propagate to each microphone element without delay according to a far-field acoustic model.
And step S12, taking the first microphone element channel as a reference channel, calculating the error signal target value of the self-adaptive calibration between each of the rest microphone element channels and the reference channel, and calculating the compensation coefficient of the self-adaptive calibration compensation filter corresponding to each microphone element channel by adopting a proportional normalization least mean square algorithm.
As shown in fig. 2 and 3, the broadband calibration signal s collected by each microphone elementn(m) after passing through the cascade adaptive calibration compensation filter, sending the signal to the FIR filter, wherein the input signal of the FIR filter is xn(m) of the reaction mixture. Selecting the first microphone element channel 0 as a reference channel, wherein the digital signal x output by the reference channel0(m) as a reference target signal, the impulse response function of the microphone element channel 0 is g0And (M) is delta (M-M), M represents a sampling point of the first microphone element channel 0, M represents the delay number of the sampling point, the delay number is the same as the order of the self-adaptive calibration compensation filter, and delta is a small integer, so that the stability reduction caused by the overlarge step factor eta due to the overlarge inner product of the input vector is prevented.
The target values of the error signals of the adaptive calibration between the remaining microphone element channels and the reference channel are expressed as:
|en(m)|2＝|x0(m)-sn(m)gn(m)|2 (1)
in the above formula, en(m) a digital signal x representing the output of the adaptive calibration compensation filter corresponding to the nth microphone element channeln(m) and a reference target signal x0Error between (m), i.e. en(m)＝x0(m)-xn(m) of the reaction mixture. Broadband calibration signal s collected by nth microphone elementn(m)＝[sn(m),...,sn(m-M+1)]T. Adopting proportional normalization least mean square algorithm to calculate | en(m)|2G corresponding to the minimumnAs the compensation coefficients of the adaptively calibrated compensation filter corresponding to the nth microphone element channel.
When the proportional normalization least mean square algorithm calculates the compensation coefficient of the adaptive calibration compensation filter corresponding to each microphone element channel, the adopted iterative formula is as follows:
in the above formula, gn(m +1) represents the compensation coefficient corresponding to the next sampling point, gn(m) represents a compensation coefficient corresponding to the current sampling point, η represents a modified step constant, en(m) a digital signal x representing the output of the adaptive calibration compensation filter corresponding to the nth microphone element channeln(m) and a reference target signal x0(m) the error delta is a small integer, so that the stability is prevented from being reduced due to the fact that the step length constant eta is too large when the inner product of the input vectors is too small;Trepresenting a transpose; g (m +1) ═ diag { beta [ ]1(m+1),β2(m+1),...,βMAnd (m +1) is a step size control matrix, diag represents a diagonal matrix, and beta (m +1) represents an element in the step size control matrix, and is calculated according to the following formula.
Compensation coefficient feedback value xil(m+1)＝max{ρmax{υ,|g1(m)|,...,|gM(m)|},|gl(M) | } prevents iteration failure caused by too small compensation coefficient, upsilon is a correction value for preventing the compensation coefficients from being all zero, ρ is a compensation coefficient feedback scale factor, and ρ is generally selected to be more than or equal to 1/M and less than or equal to 5/M. The proportional normalization least mean square algorithm has high convergence speed, and the value of | e isn(m)|2If epsilon is less than epsilon, the iterative convergence condition is satisfied, and a fixed calibration filter coefficient g is obtained after convergencenN-1, compensating for errors caused by channel-to-channel transfer function mismatch. Where ε represents the mean square error threshold.
Step S2, determining the frequency band range of interest to the user and the main and side lobes of the array beam over the frequency band range.
The main lobe of the array beam in the frequency band range in which the user is interested comprises the beam main lobe width and the beam main lobe direction corresponding to the frequency values of all the frequency points, and a side lobe area and a side lobe constraint direction.
The method comprises the following substeps:
and step S21, determining the frequency band range in which the user is interested, dividing the frequency band range into K +1 frequency points, selecting R +1 reference frequency points from the frequency points, and calculating the frequency values of the related frequency points.
Determining the lower limit f of the frequency according to the requirements in practical applicationLAnd an upper frequency limit fHConstituting a frequency band range of interest to the user fL,fH]. Wherein, f is more than 0L＜fH≤fs/2，fsIs the digital audio sampling frequency.
In the frequency band of interest to the user fL,fH]Frequency value f corresponding to each frequency point in internally divided K +1 frequency pointskComprises the following steps:
e.g. in the frequency band of interest to the user fL,fH]And 120 frequency points are internally divided.
Uniformly selecting the frequency lower limit f from the divided K +1 frequency pointsLAnd an upper frequency limit fHThe R +1 reference frequency points generally require that K can be divided by R, and the frequency reference value frComprises the following steps:
for example, 5 frequency points are selected from the 120 divided frequency points as reference frequency points, and are respectively [ B (f)0)，B(f1)，B(f2)，B(f3)，B(f4)]＝[100°,80°,65°，60°,55°]。
Step S22, determining frequency reference value f of each reference frequency pointrCorresponding reference beam width B (f)r) And carrying out interpolation operation by referring to the beam width to obtain the beam main lobe width corresponding to the frequency values of all frequency points in the frequency band range interested by the user.
Taking the beam width of the conventional method as the basis according to the actual situationThe application requirements are that the main lobe beam is widened and narrowed, and different frequency reference values f are determinedrCorresponding reference beam width B (f)r). According to different frequency reference values frCorresponding reference beam width B (f)r) Calculating the frequency values f of all frequency points in the frequency band range of interest of the user by adopting an interpolation algorithmkCorresponding main lobe width B (f)k). Usually, a piecewise linear interpolation method is adopted to obtain frequency values f of all frequency pointskCorresponding main lobe width B (f)k). The calculation formula under the piecewise linear interpolation method is as follows:
and step S23, determining a side lobe region and a main lobe beam corresponding to the frequency values of all the frequency points in the frequency band range interested by the user according to the main lobe width of the beam corresponding to the frequency values of all the frequency points, and finely dividing the side lobe region to apply constraint to obtain side lobe constraint direction values on all the frequency points.
Knowing the frequency values f of all frequency points in the frequency band range of interest to the userkCorresponding main lobe width B (f) of the beamk) And then, regarding the part outside the main lobe width of the beam as a side lobe area, and in order to obtain better side lobe suppression, finely dividing the side lobe area to exert constraint, and enabling N to be NslFor the number of side lobe constraints, θsFor the target audio source direction angle, the sidelobe constraint direction values at all frequency points are:
in the above formula, v is 0sl-1，
In addition, because the main lobe direction of the wave beam follows the angle of the target sound source, the frequency value f of all frequency points is obtainedkThe corresponding main lobe width of the wave beam and the angle of the target sound source can also determine all the frequenciesFrequency value f of pointkCorresponding main lobe beam, particularly denoted as θs+B(fk)/2 [θs-B(fk)/2，θs+B(fk)/2]. For example, the frequency value f of a frequency pointkThe width of the corresponding main lobe of the wave beam is 20 degrees, the angle of the target sound source is 70 degrees to 90 degrees, and the frequency value f of the frequency point can be obtainedkThe corresponding main lobe direction of the beam is 80-100 degrees.
And step S3, calculating an input spectrum power autocorrelation matrix in a frequency band range interested by a user according to the design requirement of the expected wave beam, the interference suppression requirement and the noise reduction requirement, and constructing a problem of solving the optimal FIR filter weight coefficient under the linear constraint minimum variance criterion.
The method comprises the following substeps:
step S31, dividing N again in the frequency band range of interest to the userfAnd +1 frequency points, and calculating the frequency value of each frequency point.
In the frequency band of interest to the user fL,fH]N of inner divisionfFrequency value f corresponding to each frequency point in +1 frequency pointsqComprises the following steps:
wherein N isfThe larger the value is, the finer the frequency division is, the higher the calculation precision of the broadband input spectrum power autocorrelation matrix is, and the more favorable the accurate obtaining of the weight coefficient of the FIR filter meeting the requirements is.
And step S32, calculating the spectral power autocorrelation matrixes of the target sound source, the uncorrelated background noise, the scattered noise component and the directional interference respectively.
If the direction angle theta of the target sound source is knownsIts spectral power autocorrelation matrix RsObtained according to the following formula.
In the above formula, σsWeighting coefficient representing target sound source signalThe value range is [0,1 ]]Re denotes the real part, H denotes the conjugate transpose, vs(q) is a frequency value f corresponding to a frequency pointqThe target direction steering vector of (2) may be expressed as:
in the above formula, I represents an NxN dimensional identity matrix,
which represents the kronecker product of,
representing a vector of dimension L x 1, L representing the FIR filter order,
representing an N x 1 dimensional vector, where j represents an imaginary part and d ═ z0,z1,...,zN-1]TA coordinate vector representing the position of the microphone array, and c represents the speed of sound propagation in air.
Assuming that the background noise on each microphone array channel is uncorrelated and satisfies a Gaussian distribution, let the variance of the uncorrelated noise be
The spectral power autocorrelation matrix of the uncorrelated background noise is obtained according to the following formula.
In the above formula, INLRepresents a NL × NL dimensional identity matrix.
Assuming that the case is considered where the actual environment is a diffuse noise field,
is the variance of the scattering noise, the spectral power self-phasing of the scattering noise componentThe relation matrix RdfObtained according to the following formula.
In the above formula, the first and second carbon atoms are,
is a Toplitz matrix, eqFor the first column of elements of the matrix,
the first row of elements of the matrix, sinc (-) is a sine function, D is an N-dimensional matrix representing position difference information of the microphone elements, and the element { D } in the Uth row and the Y-th column of the matrixUY＝|zU-zY|， U,Y＝0,1,...,N-1。
Setting interference zero limit for the condition of strong directional interference signal, then the spectral power autocorrelation matrix R of directional interferenceBObtained according to the following formula.
In the above formula, vF(q) is a guide vector of the interference direction, the number of interference zero limits is J, and the interference zero limit direction is thetaFF1, 2,.., J, corresponding to an interference variance parameter of
d′＝[z0,z1,...,zN-1]TA coordinate vector representing the location of the microphone array.
And step S33, summing and regularizing the spectral power autocorrelation matrixes of the target sound source, the uncorrelated background noise, the scattering noise component and the directional interference to obtain an input spectral power autocorrelation matrix, and constructing a problem of solving the weight coefficient of the optimal FIR filter under the linear constraint minimum variance criterion.
The input spectrum power autocorrelation matrix is the input spectrum power autocorrelation matrix corresponding to the digital signal after the calibration of the adaptive calibration compensation filter. The input spectral power autocorrelation matrix is obtained by the following regularization processing formula.
Rx＝Rs+Rn+Rdf+RB+μINL (12)
In the above equation, μ represents a regularization coefficient.
On the premise of ensuring the improvement of the operation robustness, the following constraint conditions are met, and the problem of solving the optimal FIR filter weight coefficient w under the linear constraint minimum variance criterion is constructed:
constraint conditions
In the above formula, Vs＝[vs(0),vs(1),...,vs(K)]Indicating the frequency band range of interest to the user fL,fH]A matrix formed by the target signal guide vectors of the middle divided K +1 frequency points,
representing a K +1 dimensional distortion-free constraint vector,
indicating the frequency band range of interest to the user fL,fH]A matrix formed by E-th sidelobe constraint guide vectors of the middle-divided K +1 frequency points, | | · | | represents the Euclidean norm, epsilonslAnd εnParameters of side lobe suppression degree and white noise gain amplification are respectively expressed.
And step S4, converting the constructed problem of the optimal FIR filter weight coefficient meeting the constraint condition into an unconstrained optimization problem, and solving by using a particle swarm optimization algorithm to obtain the weight coefficient of the optimal FIR filter.
The method comprises the following substeps:
step S41, converting the problem of the constructed optimal FIR filter weight coefficient meeting the constraint condition into an unconstrained optimization problem:
wherein Ω (w) is an objective function of the optimization problem, and a weight coefficient w of the FIR filter which can minimize Ω (w) is calculated, and λ is a positive weight coefficient.
Step S42, setting relevant parameters of the particle swarm optimization algorithm, initializing each population particle in the solution space and setting the position vector and the velocity vector of each population particle, wherein the position vector and the velocity vector are respectively
And
np is the number of the population particles, Q is the number of the particles, and t represents the iteration number of the particle swarm optimization algorithm.
In the embodiment of the invention, the particle swarm optimization algorithm is an optimization algorithm for finding the optimal solution through cooperation and information sharing among individuals in a swarm.
Setting relevant parameters of the particle swarm optimization algorithm including iteration times T and maximum iteration times TmaxNumber of population particles NpThe inertia factor omega and the learning factor are respectively
Initializing N in the solution space by making the iteration number t equal to 0pThe NL-dimensional position vector and velocity vector of each population particle are respectively set as:
and
any one of the set position vectors is expressed as:
in the above formula, IK+1Is a K +1 dimensional identity matrix and α is a regularization parameter. The position vectors and velocity vectors of other populations of particles may be randomly generated in the solution space.
Step S43, order
Substituting the position vectors of all the population particles into the unconstrained optimization problem to calculate an objective function
And searching and finding out the self historical optimal solution vector of the population particles
And historical optimal solution vectors for all population particles
t' represents the iteration number of the particle swarm optimization algorithm.
Step S44, when the iteration time T of the particle swarm optimization algorithm reaches the preset maximum iteration time TmaxIf so, terminating the iteration of the particle swarm optimization algorithm to obtain the weight coefficient of the optimal FIR filter
And performs step S5; otherwise, updating the velocity vector and the position vector of the particle swarm, and returning to the step S43 to continue the iteration.
As shown in FIG. 2, the weight coefficients w of the optimal FIR filter obtained by the particle swarm optimization algorithm are used as the weightsOn an FIR filter, which can be expressed as hn(m)，n＝0,1,...,N-1。
When the iteration time T of the particle swarm optimization algorithm does not reach the preset maximum iteration time TmaxAccording to the optimal solution vector of the population particles of the iteration
And
updating the velocity vector and the position vector of the particle swarm:
in the above formula, the inertia factor ω and the learning factor
Selecting [0,1 ] according to specific application]The number of intervals.
Step S5, summing the digital signals after filtering compensation by the adaptive calibration compensation filter corresponding to each microphone element channel by using the weight coefficient of the optimal FIR filter, and outputting the time domain signals after beam forming; and if the incoming wave direction of the target sound source pointed by the wave beam or the requirements of side lobe restriction, interference and noise suppression are not changed, the step S5 is repeatedly executed, otherwise, the step S2 is returned, and the weight coefficients of the FIR filter are recalculated.
As shown in fig. 2, the ambient sound signal s is collected by each microphone element channel in real timen(m) digital signal x after filtering compensation of calibration compensation filtern(m) outputting to the FIR filter having the optimal weight coefficient to sum and output the beamformed time domain signal y (m).
Fig. 4 is a comparison of beam patterns of the present invention with other beamforming methods. It can be seen that the beam sidelobe formed by the conventional frequency domain delay-sum beam forming method is high, and in addition, the directivity and robustness in the low frequency part are poor. The linear constraint minimum variance beam forming method adopts an FIR filtering and adding framework, the optimal criterion and constraint conditions are directly met in a wide frequency band range through time domain weighted filtering, and the side lobe suppression degree is improved by about 1 to 2dB compared with a delay adding method. The improved linear constraint minimum variance beam forming method can flexibly control the side lobe suppression degree by increasing the side lobe suppression constraint and constructing the input spectrum power autocorrelation function with the directional interference and the noise suppression on the basis of the original method, the average suppression degree of the side lobe is 20dB in the example, strong suppression null can be formed in a specific direction, and the problem of noise amplification is avoided. The method adopts the particle swarm optimization algorithm to search the optimal weight coefficient of the FIR filter on the basis of the improved linear constraint minimum variance method, can more accurately approach the global optimal solution of the optimization problem while keeping the advantages, has the side lobe average suppression degree of about 26dB and has stronger noise reduction capability.
Figure 5 is a comparison of the beam patterns of the present invention with calibration-without-channel beamforming. Because random deviation of amplitude or phase exists among microphone element channels, the sidelobe suppression degree of the channel-free calibration beam forming method is obviously reduced compared with that under the ideal condition that the channels are consistent, and the sidelobe curve of the channel-carrying calibration beam forming method provided by the invention is very close to the sidelobe curve under each ideal condition, and the sidelobe suppression degree can be improved to about 10 dB. The method can solve the problems of performance reduction and robustness caused by channel consistency by simply determining the compensation coefficient in the calibration stage through the self-adaptive calibration process.
Fig. 6a to 6c are comparison results of output signal waveforms and speech spectrogram of the method of the present invention and the conventional beamforming method. FIG. 6a is a diagram of an input signal including a target sound source signal, white noise and two sets of directional interference signals, one set being a 115 degree voice signal and the other set being a 40 degree mono signal; fig. 6b is the output of the prior art method, and it can be seen that because the beam sidelobe is high, the suppression of two sets of directional interference signals is poor, and the output waveform interference signal remains much; fig. 6c shows that the method of the present invention can reduce the side lobe, and also form strong zero-limit suppression on the single-tone interference signal in the specific direction, so that it can be seen that the voice signal in the 115-degree direction is almost completely suppressed by the strong zero-limit suppression, and the single-tone interference signal in the 40-degree direction of the side lobe has only weak residue, and in addition, the method of the present invention also has good noise reduction capability.
In addition, as shown in fig. 7, an embodiment of the present invention further provides a wideband beamforming device with channel calibration particle swarm optimization, which includes a processor 32 and a memory 31, and may further include a communication component, a sensor component, a power component, a multimedia component, and an input/output interface according to actual needs. The memory, communication components, sensor components, power components, multimedia components, and input/output interfaces are all connected to the processor 32. As mentioned above, the memory 31 may be a Static Random Access Memory (SRAM), an electrically erasable programmable read-only memory (EEPROM), an erasable programmable read-only memory (EPROM), a programmable read-only memory (PROM), a read-only memory (ROM), a magnetic memory, a flash memory, etc.; the processor 32 may be a Central Processing Unit (CPU), Graphics Processing Unit (GPU), Field Programmable Gate Array (FPGA), Application Specific Integrated Circuit (ASIC), Digital Signal Processing (DSP) chip, or the like. Other communication components, sensor components, power components, multimedia components, etc. may be implemented using common components found in existing smartphones and are not specifically described herein.
In addition, the wideband beam forming apparatus with channel calibration particle swarm optimization provided by the embodiment of the present invention includes a processor 32 and a memory 31, where the processor 32 reads a computer program or an instruction in the memory 31 to perform the following operations:
step S1, performing adaptive calibration in the calibration stage, and calculating the compensation coefficients of the adaptive calibration compensation filters corresponding to the microphone element channels.
Step S2, determining the frequency band range of interest to the user and the main and side lobes of the array beam over the frequency band range.
And step S3, calculating an input spectrum power autocorrelation matrix in a frequency band range interested by a user according to the design requirement of the expected wave beam, the interference suppression requirement and the noise reduction requirement, and constructing a problem of solving the optimal FIR filter weight coefficient under the linear constraint minimum variance criterion.
And step S4, converting the constructed problem of the optimal FIR filter weight coefficient meeting the constraint condition into an unconstrained optimization problem, and solving by using a particle swarm optimization algorithm to obtain the weight coefficient of the optimal FIR filter.
Step S5, summing the digital signals after filtering compensation by the adaptive calibration compensation filter corresponding to each microphone element channel by using the weight coefficient of the optimal FIR filter, and outputting the time domain signals after beam forming; and if the incoming wave direction of the target sound source pointed by the wave beam or the requirements of side lobe restriction, interference and noise suppression are not changed, the step S5 is repeatedly executed, otherwise, the step S2 is returned, and the weight coefficients of the FIR filter are recalculated.
In addition, an embodiment of the present invention further provides a computer-readable storage medium, where instructions are stored on the computer-readable storage medium, and when the instructions are run on a computer, the computer is enabled to execute the method for wideband beamforming with calibration of particle swarm optimization with channel as described in fig. 1, and details of a specific implementation thereof are not repeated here.
In addition, an embodiment of the present invention further provides a computer program product including instructions, which when run on a computer, causes the computer to execute the method for wideband beamforming with calibration particle swarm optimization as described in fig. 1, and details of the implementation thereof are not repeated here.
Compared with the prior art, the method and the device for optimizing the broadband beam forming by the particle swarm with the channel calibration function have the following characteristics:
(1) the beam design with beam control and complex interference noise suppression is considered, the strong directional interference, the scattering noise and the white noise gain are restrained in a targeted manner, the strong directional interference can be suppressed obviously under the condition of meeting the requirements of the main lobe width and the side lobe level, and the output signal-to-noise ratio is improved.
(2) Aiming at the optimization problem, a particle swarm optimization method is adopted, the optimal solution of the weight coefficient of the FIR filter can be rapidly and accurately obtained, the expected beam is accurately synthesized, and the beam directivity and the interference noise suppression capability are effectively improved.
(3) In the calibration stage, the compensation coefficient of the calibration compensation filter is obtained by using the adaptive filtering framework and the proportional normalization least mean square algorithm, the performance reduction of an array beam caused by the consistency difference between microphone element channels is eliminated, the effectiveness and the robustness of the technical method are obviously improved, and the method has higher practical value.
The method and the device for forming the wideband beam by optimizing the particle swarm with the channel calibration provided by the invention are explained in detail above. It will be apparent to those skilled in the art that various modifications can be made without departing from the spirit of the invention.
Claims (11)
1. A method for optimizing broadband beam forming by using particle swarm with channel calibration is characterized by comprising the following steps:
step S1, self-adaptive calibration is carried out in the calibration stage, and the compensation coefficients of the self-adaptive calibration compensation filters corresponding to the microphone element channels are calculated;
step S2, determining the frequency band range of interest of the user and the main lobe and the side lobe of the array beam on the frequency band range;
step S3, calculating an input spectrum power autocorrelation matrix in a frequency band range in which a user is interested, and constructing a problem of solving the weight coefficient of the optimal FIR filter under the linear constraint minimum variance criterion;
step S4, converting the problem into an unconstrained optimization problem, and solving by using a particle swarm optimization algorithm to obtain a weight coefficient of the optimal FIR filter;
step S5, summing the digital signals after filtering compensation by the adaptive calibration compensation filter by using the weight coefficient of the optimal FIR filter, and outputting the time domain signals after beam forming; and if the incoming wave direction of the target sound source pointed by the wave beam or the requirements of sidelobe constraint, interference and noise suppression are not changed, repeatedly executing the step S5, otherwise, returning to the step S2 and recalculating the weight coefficient of the FIR filter.
2. The method for wideband beamforming with channel calibration particle swarm optimization according to claim 1, wherein step S2 comprises the following sub-steps:
step S11, collecting the played broadband calibration signal by using a microphone array;
and step S12, taking the first microphone element channel as a reference channel, calculating the error signal target value of the adaptive calibration between each of the rest microphone element channels and the reference channel, and calculating the compensation coefficient of the adaptive calibration compensation filter corresponding to each microphone element channel by adopting a proportional normalization least mean square algorithm.
3. The particle-swarm optimized wideband beamforming method for calibration with channel as claimed in claim 2, wherein in step S12,
the remaining error signal target values of the adaptive calibration between each microphone element channel and the reference channel are expressed as:
|en(m)|2＝|x0(m)-sn(m)gn(m)|2
in the above formula, en(m) a digital signal x representing the output of the adaptive calibration compensation filter corresponding to the nth microphone element channeln(m) and a reference target signal x0(m) error between; broadband calibration signal s collected by nth microphone elementn(m)＝[sn(m),...,sn(m-M+1)]TM denotes the sampling point of the first microphone element channel, M denotes the number of delays of the sampling point, gnAnd (m) represents a compensation coefficient corresponding to the current sampling point.
4. The particle-swarm-optimized wideband beamforming method with channel calibration according to claim 3, characterized in that:
when the proportional normalization least mean square algorithm calculates the compensation coefficient of the adaptive calibration compensation filter corresponding to each microphone element channel, the adopted iterative formula is as follows:
in the above formula, gn(m +1) represents the compensation coefficient corresponding to the next sampling point, gn(m) represents a compensation coefficient corresponding to the current sampling point, eta represents a modified step constant, and delta is a smaller integer, so that the stability reduction caused by the overlarge step constant eta due to the overlarge inner product of the input vector is prevented; t represents transposition; g (m +1) ═ diag { beta [ ]1(m+1),β2(m+1),...,βM(m +1) } is the step size control matrix, diag { } denotes the diagonal matrix, and β (m +1) denotes the element in the step size control matrix.
5. The particle-swarm-optimized wideband beamforming method with channel calibration according to claim 4, characterized in that:
the element beta (m +1) in the step control matrix is obtained according to the following formula;
compensation coefficient feedback value xil(m+1)＝max{ρmax{υ,|g1(m)|,...,|gM(m)|},|glAnd (m) is used for preventing iteration from being invalid due to too small compensation coefficient, upsilon is a correction value for preventing the compensation coefficients from being all zero, and ρ is a compensation coefficient feedback scale factor.
6. The method for wideband beamforming with channel calibration particle swarm optimization according to claim 1, wherein step S2 comprises the following sub-steps:
step S21, determining the frequency band range of interest of the user, dividing the frequency band range into K +1 frequency points, selecting R +1 reference frequency points from the frequency points, and calculating the frequency values of the relevant frequency points;
step S22, determining a reference beam width corresponding to the frequency reference value of each reference frequency point, and obtaining beam main lobe widths corresponding to the frequency values of all frequency points in the frequency band range in which a user is interested by performing interpolation operation on the reference beam width;
and step S23, determining a side lobe region and a main lobe beam corresponding to the frequency values of all frequency points in the frequency band range interested by the user according to the main lobe width of the beam, and finely dividing the side lobe region to apply constraint to obtain side lobe constraint direction values on all frequency points.
7. The method for wideband beamforming with channel calibration particle swarm optimization according to claim 1, wherein step S3 comprises the following sub-steps:
step S31, dividing N again in the frequency band range of interest to the userf+1 frequency points, and calculating the frequency value of each frequency point;
step S32, calculating the spectral power autocorrelation matrixes of the target sound source, the uncorrelated background noise, the scattering noise component and the directional interference respectively;
and step S33, summing and regularizing the spectral power autocorrelation matrixes of the target sound source, the uncorrelated background noise, the scattering noise component and the directional interference to obtain an input spectral power autocorrelation matrix, and constructing a linear constraint minimum variance criterion to solve the problem of the optimal FIR filter weight coefficient.
8. The method for forming wideband beam by particle swarm optimization with channel calibration as claimed in claim 7, wherein in step S33, the following constraint conditions are satisfied, and the problem of solving the optimal FIR filter weight coefficient w under the linear constraint minimum variance criterion is constructed:
constraint conditions
In the above formula, Vs＝[vs(0),vs(1),...,vs(K)]K +1 representing a division in a frequency band of interest to a userA matrix formed by the target signal guide vectors of the frequency points,
representing a K +1 dimensional distortion-free constraint vector, fkRepresenting the frequency value f corresponding to each frequency point in K +1 frequency pointssRepresenting a digital audio sampling frequency;
the matrix is composed of E-th sidelobe constraint guide vectors of K +1 frequency points, wherein | · | | represents the Euclidean norm, epsilonslAnd εnParameters of the side lobe suppression degree and white noise gain amplification are respectively expressed.
9. The method for wideband beamforming with channel calibration particle swarm optimization according to claim 8, wherein step S4 comprises the following sub-steps:
step S41, converting the problem of the constructed optimal FIR filter weight coefficient meeting the constraint condition into an unconstrained optimization problem:
where Ω (w) represents the objective function of the optimization problem, w represents the weight coefficients of the FIR filter, RxRepresenting the input spectral power autocorrelation matrix, NslRepresenting the number of side lobe constraints, and lambda represents a positive weight coefficient;
step S42, setting relevant parameters of the particle swarm optimization algorithm, initializing each population particle in the solution space and setting the position vector and the velocity vector of each population particle, wherein the position vector and the velocity vector are respectively
And
np is the number of population particles, Q is the number of particles, tRepresenting the iteration times of the particle swarm optimization algorithm;
step S43, order
Substituting the position vectors of all the population particles into the unconstrained optimization problem to calculate an objective function
And searching and finding out the self historical optimal solution vector of the population particles
And historical optimal solution vectors for all population particles
t' represents the iteration times of the particle swarm optimization algorithm;
step S44, when the iteration number T of the particle swarm optimization algorithm reaches the preset maximum iteration number TmaxIf so, terminating the iteration of the particle swarm optimization algorithm to obtain the weight coefficient of the optimal FIR filter
And performs step S5; otherwise, updating the velocity vector and the position vector of the particle swarm, and returning to the step S43 to continue the iteration.
10. The method for wideband beamforming with channel calibration particle swarm optimization as claimed in claim 9, wherein in step S44, when the iteration number T of the particle swarm optimization algorithm does not reach the preset maximum iteration number TmaxAccording to the optimal solution vector of the population particles of the iteration
And
updating the speed of a particle swarmVector and position vector:
in the above formula, the inertia factor ω and the learning factor
Selecting [0,1 ] according to specific application]The number of intervals.
11. A particle swarm optimization wideband beamforming device with channel calibration, comprising a processor and a memory, the processor reading a computer program or instructions in the memory for performing the following operations:
step S1, self-adaptive calibration is carried out in the calibration stage, and the compensation coefficients of the self-adaptive calibration compensation filters corresponding to the microphone element channels are calculated;
step S2, determining the frequency band range of interest of the user and the main lobe and the side lobe of the array beam on the frequency band range;
step S3, calculating an input spectrum power autocorrelation matrix in a frequency band range in which a user is interested, and constructing a problem of solving the weight coefficient of the optimal FIR filter under the linear constraint minimum variance criterion;
step S4, converting the problem into an unconstrained optimization problem, and solving by using a particle swarm optimization algorithm to obtain a weight coefficient of the optimal FIR filter;
step S5, summing the digital signals after filtering compensation by the adaptive calibration compensation filter by using the weight coefficient of the optimal FIR filter, and outputting the time domain signals after beam forming; and if the incoming wave direction of the target sound source pointed by the wave beam or the requirements of sidelobe constraint, interference and noise suppression are not changed, repeatedly executing the step S5, otherwise, returning to the step S2 and recalculating the weight coefficient of the FIR filter.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110839718.XA CN113655436B (en) | 2021-07-23 | 2021-07-23 | Method and device for optimizing broadband wave beam formation through channel calibration particle swarm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110839718.XA CN113655436B (en) | 2021-07-23 | 2021-07-23 | Method and device for optimizing broadband wave beam formation through channel calibration particle swarm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113655436A true CN113655436A (en) | 2021-11-16 |
| CN113655436B CN113655436B (en) | 2024-04-16 |
Family
ID=78478115
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110839718.XA Active CN113655436B (en) | 2021-07-23 | 2021-07-23 | Method and device for optimizing broadband wave beam formation through channel calibration particle swarm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113655436B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114280544A (en) * | 2021-12-02 | 2022-04-05 | 电子科技大学 | Minimum transition bandwidth directional diagram shaping method based on relaxation optimization |
| CN116611223A (en) * | 2023-05-05 | 2023-08-18 | 中国科学院声学研究所 | Accurate array response control method and device combined with white noise gain constraint |
| CN116683965A (en) * | 2023-07-20 | 2023-09-01 | 之江实验室 | Digital beam forming device, method and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108872870A (en) * | 2018-06-21 | 2018-11-23 | 浙江工业大学 | A kind of lithium battery SOC estimation method based on particle group optimizing expanded Kalman filtration algorithm |
| CN109959917A (en) * | 2019-03-08 | 2019-07-02 | 南京航空航天大学 | A kind of non-frequency in broadband becomes the array Sparse methods of multi-beam imaging sonar |
| CN111487594A (en) * | 2020-04-23 | 2020-08-04 | 中国民航大学 | Circular array beam forming method based on particle swarm optimization |
| CN112202483A (en) * | 2020-10-12 | 2021-01-08 | 西安工程大学 | Beam forming method and device, electronic equipment and storage medium |
-
2021
- 2021-07-23 CN CN202110839718.XA patent/CN113655436B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108872870A (en) * | 2018-06-21 | 2018-11-23 | 浙江工业大学 | A kind of lithium battery SOC estimation method based on particle group optimizing expanded Kalman filtration algorithm |
| CN109959917A (en) * | 2019-03-08 | 2019-07-02 | 南京航空航天大学 | A kind of non-frequency in broadband becomes the array Sparse methods of multi-beam imaging sonar |
| CN111487594A (en) * | 2020-04-23 | 2020-08-04 | 中国民航大学 | Circular array beam forming method based on particle swarm optimization |
| CN112202483A (en) * | 2020-10-12 | 2021-01-08 | 西安工程大学 | Beam forming method and device, electronic equipment and storage medium |
Non-Patent Citations (1)
| Title |
|---|
| 周飞红;刘辉;廖子贞;: "粒子群优化算法在FIR数字滤波器设计中的应用", 计算机工程与应用, no. 33 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114280544A (en) * | 2021-12-02 | 2022-04-05 | 电子科技大学 | Minimum transition bandwidth directional diagram shaping method based on relaxation optimization |
| CN116611223A (en) * | 2023-05-05 | 2023-08-18 | 中国科学院声学研究所 | Accurate array response control method and device combined with white noise gain constraint |
| CN116611223B (en) * | 2023-05-05 | 2023-12-19 | 中国科学院声学研究所 | Accurate array response control method and device combined with white noise gain constraint |
| CN116683965A (en) * | 2023-07-20 | 2023-09-01 | 之江实验室 | Digital beam forming device, method and storage medium |
| CN116683965B (en) * | 2023-07-20 | 2023-10-20 | 之江实验室 | Digital beam forming device, method and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113655436B (en) | 2024-04-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113655436B (en) | Method and device for optimizing broadband wave beam formation through channel calibration particle swarm | |
| US8223989B2 (en) | Signal processing system and method for calibrating channel signals supplied from an array of sensors having different operating characteristics | |
| CN111415676B (en) | Blind source separation method and system based on separation matrix initialization frequency point selection | |
| US9031257B2 (en) | Processing signals | |
| JP4973657B2 (en) | Adaptive array control device, method, program, and adaptive array processing device, method, program | |
| CN105137409B (en) | The sane space-time adaptive processing method of echo signal mutually constrained based on width | |
| CN113422630B (en) | Self-adaptive focusing broadband beam forming method and system | |
| CN103984676A (en) | Rectangular projection adaptive beamforming method based on covariance matrix reconstruction | |
| EP2949130A1 (en) | Method of configuring planar transducer arrays for broadband signal processing by three-dimensional beamforming and signal processing systems using such method, particularly an acoustic camera | |
| WO2007123051A1 (en) | Adaptive array controlling device, method, program, and adaptive array processing device, method, program | |
| JPH10207490A (en) | Signal processor | |
| CN113889136A (en) | Pickup method, pickup device and storage medium based on microphone array | |
| CN108877828B (en) | Speech enhancement method/system, computer-readable storage medium, and electronic device | |
| CN109040913B (en) | Beam forming method of window function weighting electroacoustic transducer transmitting array | |
| CN109669172B (en) | Weak target direction estimation method based on strong interference suppression in main lobe | |
| Zhang et al. | Selective frequency invariant uniform circular broadband beamformer | |
| CN112834981A (en) | Null array arrival direction estimation method under impulse noise background | |
| CN111162878A (en) | Multi-domain joint anti-interference method based on subarray dimension reduction band constraint | |
| CN113491137B (en) | Flexible differential microphone array with fractional order | |
| Zhang et al. | Frequency-invariant beamformer design via ADPM approach | |
| McCowan et al. | Multi-channel sub-band speech recognition | |
| CN112953609A (en) | Fast iteration least square broadband beam forming method | |
| CN113702900A (en) | Planar array sub-band weight optimization broadband beam forming method and device | |
| CN116611223B (en) | Accurate array response control method and device combined with white noise gain constraint | |
| CN116962948B (en) | Non-uniform linear sparse microphone array design method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |