Feedback active noise control method based on sound field compensation

Abstract

The invention discloses a feedback active noise control method based on sound field compensation. The method comprises the following steps: measuring a secondary path transfer function from a secondary source to a microphone and a human ear position and a transfer function between the microphone and the human ear in a primary sound field; a compensation filter C (z) is obtained through off-line optimization of a measurement result; compensating the microphone signal by using a compensation filter C (z) to obtain a target control signal at the microphone; and a servo control system is used for tracking a target control signal at the microphone, so that noise reduction at human ears is realized. According to the method, the sound field space characteristics between the microphone and the human ear are used for compensating the microphone signal to obtain the target control signal at the microphone, and meanwhile, the advantage of short time delay of the transfer function between the secondary source and the microphone is utilized, so that the target control signal can be tracked in a relatively wide frequency band, and a relatively good noise reduction effect at the position of the human ear is realized.

Description

Technical Field

[0001] The invention relates to the field of noise control, and in particular to a feedback active noise control method based on sound field compensation. Background Art

[0002] Active Noise Control (ANC) technology uses the principle of destructive interference of sound waves to reduce noise. It has been applied to scenarios such as active noise-cancelling headphones, active noise reduction of automobile engines, active noise reduction of helicopter propellers, and active control of automobile road noise. It is also used in other fields such as active control of machine noise, natural ventilation soundproof windows, and active noise-cancelling ducts.

[0003] Active noise reduction headrest is a typical active noise control system, which uses the speaker on the headrest as a secondary source to control the noise at the microphone arranged near the human ear to achieve noise reduction at the human ear. Active noise reduction headrest system has been applied to noise control in closed mobile spaces such as cars, airplanes and high-speed rail cabins.

[0004] Active noise control can be divided into feedforward active noise control and feedback active noise control, where the performance of feedforward active noise control depends on the coherence between the reference signal and the primary noise signal at the target control position. In order to obtain a better noise reduction effect, the active noise reduction head based on the feedforward control strategy usually needs to optimize the number and position of the reference sensors that pick up the reference signal to ensure that there is sufficient coherence and causality between the reference signal and the primary noise signal. The system cost is complex and the implementation cost is high.

[0005] In contrast, feedback active noise control does not require a reference signal. It can effectively reduce the noise at the error microphone by only forming a feedback closed loop through a feedback controller, a secondary source, and an error microphone. This method has the advantages of simple system structure and low implementation cost. However, the noise reduction bandwidth of the feedback active control system decreases with the increase of the secondary path delay. For active noise reduction headrests, in order to facilitate the movement of the user's head, the secondary source is at a certain distance from the human head. If the error microphone is placed closer to the secondary source, the secondary path delay is shorter, and a better noise reduction effect can be obtained at the error microphone, and a certain range of quiet zone is generated near the microphone. However, the size of the quiet zone is limited, and it is a spherical area with a diameter not exceeding 1 / 10 of the wavelength, which is difficult to effectively cover the position of the human ear. If the error microphone is placed very close to the human ear, the quiet zone generated at the error microphone can cover the human ear. However, this also causes two problems. On the one hand, the error microphone is too close to the human ear, which affects the movement of the head. On the other hand, the secondary path delay is long, and the noise reduction effect at the error microphone is limited, which in turn affects the noise reduction effect at the human ear.

[0006] Therefore, it is of great significance to provide a feedback control method that can effectively improve the noise reduction performance at the human ear. Summary of the invention

[0007] In order to overcome the shortcomings of the prior art, the present invention provides a method for improving the noise reduction performance of a feedback active noise reduction headrest by utilizing the spatial characteristics of the sound field. The method installs a microphone near the headrest speaker, utilizes the spatial characteristics of the sound field at the human ear to compensate the microphone signal to obtain a target control signal at the microphone, and tracks the target control signal through a servo control system to achieve noise reduction at the human ear position.

[0008] To achieve the above object, the technical solution adopted by the present invention is:

[0009] A feedback active noise control method based on sound field compensation comprises the following steps:

[0010] Step 1: Measure the transfer function from the secondary source to the microphone in the active noise control system and obtain the transfer function estimate n is the time index, M is the path unit impulse response length, and the superscript T is the transposition operator symbol. The z-domain form of

[0011] Step 2: Measure the transfer function from the secondary source to the target noise reduction position in the active noise control system and obtain the transfer function estimate Its z-domain form can be expressed as

[0012] Step 3: Measure the transfer function between the microphone and the target noise reduction position in the primary noise field, and obtain the transfer function as follows: L is its unit impulse response length, The z-domain form of

[0013] Step 4, use and Primary noise of microphone p m (n) Filter to get the expected signal p′ m (n), the z-domain of this process is expressed as Use p m (n) The filtered signal r m (n) is used as the reference signal, and then the least mean square algorithm (LMS) is used to identify the reference signal and the expected signal p′ m (n), the compensation filter C(z) can be obtained, and its time domain optimal Wiener solution c opt =[c(0),c(1),...,c(K-1)]T (K is the compensation filter length) is:

[0014]

[0015] in is the reference signal r m (n) = [r m (n), r m (n-1), ..., r m (n-K+1)] T The autocorrelation matrix of For r m (n) and the expected signal p′ m (n) cross-correlation vector;

[0016] Step 5: Use the controller output signal u(n) Filter to get the secondary sound estimate at the microphone After taking the negative value, it is equal to the signal e picked up by the microphone m (n) is added to obtain the primary noise estimate at the microphone The z-domain calculation process can be expressed as in E m (z) and U(z) are e m The z-domain form of (n) and u(n);

[0017] Step 6: Use the compensation filter C(z) to estimate the microphone primary noise Filter to get the target control signal at the microphone

[0018] Step 7: Use the servo control system to adjust the secondary sound field at the microphone. Regulate to minimize the cost function shown in formula (2),

[0019]

[0020] in for With the target control signal The servo tracking system is an internal model control (IMC) structure, and its control filter w(n) = [w(0), w(1), ..., w(I-1)] T (I is the control filter length) the iterative formula is:

[0021] w(n+1)=w(n)+μe t (n)s′ t (n) (3)

[0022] Where μ is the iteration step size, s′ t (n) for use Control signal to target The filtered signal vector;

[0023] Step 8, continuously iteratively control the filter coefficients to minimize the cost function shown in equation (2) and finally achieve noise reduction at the target position.

[0024] Ideally, it is assumed that there is no error between the transfer function estimates in steps 1-3 and the transfer function, that is, The optimal solution of the compensation filter is C opt (z)=-Q(z)G m (z) / G e (z).

[0025] In the actual implementation process of the compensation filter C(z) described in step 4, a finite length impulse response (FIR) filter can be used to fit the compensation value -Q(z)G m (z) / G e (z) frequency response.

[0026] The controller H(z) in the servo system described in step 7 is:

[0027]

[0028] Where W(z) is the z-domain form of w(n);

[0029] Compared with the prior art, the present invention has the following beneficial effects:

[0030] (1) Compared with directly implementing feedback control on the microphone, the method proposed in the present invention utilizes the spatial characteristics of the sound field between the microphone and the human ear to compensate the microphone signal to obtain the target control signal at the microphone, and tracks the target control signal through the feedback control system to achieve a noise reduction effect at the human ear.

[0031] (2) The method proposed in the present invention uses a feedback control system to track the target control signal at the microphone. It can take advantage of the short delay of the transfer function between the secondary source and the microphone to track the target control signal within a wider frequency band, thereby achieving a better noise reduction effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a flowchart of the method of the present invention.

[0033] Figure 2 It is a block diagram of the off-line optimization process of the compensation filter C(z) in the method of the present invention.

[0034] Figure 3 is the amplitude-frequency response curve of the compensation filter C(z) in the embodiment of the present invention.

[0035] Figure 4 is the estimated value of the signal output by the servo control system at the microphone in the embodiment of the present invention With the target control signal Sound pressure level curve.

[0036] Figure 5 1 is the error signal spectrum before and after noise reduction in the embodiment of the present invention, (a) is the error signal spectrum before and after noise reduction at the microphone, and (b) is the error signal spectrum before and after noise reduction at the human ear. DETAILED DESCRIPTION

[0037] The technical solution of the present invention is further explained below in conjunction with the accompanying drawings and specific embodiments. It should be understood that these examples are only used to illustrate the present invention and are not used to limit the scope of the present invention. After reading the present invention, various equivalent forms of modifications to the present invention by those skilled in the art all fall within the scope defined by the claims attached to this application.

[0038] A feedback active noise control method based on sound field compensation, such as Figure 1 As shown, the following steps are included:

[0039] Step 1: Measure the transfer function from the secondary source to the microphone in the active noise control system and obtain the transfer function estimate n is the time index, M is the path unit impulse response length, and the superscript T is the transposition operator symbol. The z-domain form of

[0040] Step 2: Measure the transfer function from the secondary source to the target noise reduction position (i.e., the human ear) in the active noise control system to obtain the transfer function estimate Its z-domain form can be expressed as

[0041] Step 3, measure the transfer function between the microphone and the human ear in the primary noise field, and obtain the transfer function as L is its unit impulse response length, The z-domain form of

[0042] Step 4, use and Primary noise of microphone p m (n) Filter to get the expected signal p′ m (n), the z-domain of this process is expressed as Use p m (n) The filtered signal rm (n) is used as the reference signal, and then the LMS algorithm is used to identify the reference signal and the expected signal p′ m (n), the compensation filter C(z) can be obtained, and its time domain optimal Wiener solution c opt =[c(0),c(1),...,c(K-1)] T (K is the compensation filter length) is:

[0043]

[0044] in is the reference signal r m (n) = [r m (n), r m (n-1), ..., r m (n-K+1)] T The autocorrelation matrix of For r m (n) and the expected signal p′ m (n) is the cross-correlation vector. Assume that the transfer function estimates in steps 1-3 have no error with the transfer function, that is, Ideally, the optimal solution of the compensation filter can be expressed as C opt (z)=-Q(z)G m (z) / G e (z). The compensation filter C(z) includes the primary sound field spatial characteristics Q(z) and the secondary sound field spatial characteristics G m (z) / G e (z), the optimization process can be regarded as using a finite length impulse response (FIR) filter to fit the compensation value Q(z)G m (z) / G e (z) frequency response.

[0045] Step 5: Use the controller output signal u(n) Filter to get the secondary sound estimate at the microphone After taking the negative value, it is equal to the signal e picked up by the microphone m (n) is added to obtain the primary noise estimate at the microphone The z-domain calculation process can be expressed as in E m (z) and U(z) are e m The z-domain form of (n) and u(n);

[0046] Step 6: Use the compensation filter C(z) to estimate the microphone signal Filter to get the target control signal at the microphone

[0047] Step 7: Use the servo control system to adjust the secondary sound field at the microphone. Regulate to minimize the cost function shown in formula (6),

[0048]

[0049] in for With the target control signal The servo tracking system can be transformed into a feedforward structure by introducing an internal model control structure, using the tracking error signal e t (n) plus the estimated value of the secondary sound field at the microphone The target control signal can be restored That is, the system is finally converted into a reference signal and an expected signal. The feedforward control system is: w(n) = [w(0), w(1), ..., w(I-1)] T represents the control filter in the feedforward control system, I is the length of the control filter, and the update iterative formula can be obtained by using the gradient descent method:

[0050] w(n+1)=w(n)+μe t (n)s′t(n) (7)

[0051] Where μ is the iteration step size, s′ t (n) for use Control signal to target The relationship between the controller H(z) and the feedforward control filter w(n) in the servo system can be expressed as:

[0052]

[0053] Where W(z) is the z-domain form of w(n);

[0054] Step 8: Continuously iterate and control the filter coefficients to minimize the cost function shown in equation (6) and finally achieve noise reduction at the target position.

[0055] Figure 1 The flowchart of the method of the present invention is shown in FIG. m (z) and G e (z) are the transfer functions from the secondary source to the microphone and the human ear, respectively, and The estimated results for both are: is the target control signal at the microphone The time domain form of e t(n) is the tracking error of the servo system, and u(n) is the control signal output by the controller H(z). e (n) and y m (n) are the control signals at the ear position and the microphone, respectively. for y m The estimated result of (n). e (n) and p m (n) are the time domain forms of the primary noise signals at the human ear and microphone, respectively, e(n) and e m (n) are the time domain forms of the error signals at the human ear and microphone, respectively. is the primary noise p at the microphone m (n). Using the compensation filter C(z) The target control signal can be obtained by filtering

[0056] The compensation filter can be obtained by offline optimization of steps 1-4. This process can be regarded as using a finite impulse response filter to fit the frequency response of the sound field compensation filter. The corresponding flow chart is as follows: Figure 2 As shown, Q(z) is the transfer function between the microphone and the human ear in the primary noise field, G m (z) and G e (z) are the transfer functions from the secondary source to the microphone and the human ear, respectively, and p′ m (n) is the primary noise signal p of the microphone m (n) passes through Q(z) and G m (z) The desired signal obtained by filtering, r m (n) is p m (n) After G e (z) The reference signal obtained by filtering, e f (n) is the fitting error. In the optimization process, the compensation filter C(z) is iteratively optimized using the LMS algorithm, with the goal of minimizing the fitting error e f (n). Through this iterative process, the required compensation filter is finally obtained.

[0057] The present invention uses the spatial characteristics of the sound field between the microphone and the human ear to compensate the microphone signal to obtain the target control signal at the microphone, and utilizes the advantage of the short delay of the transfer function between the secondary source and the microphone to track the target control signal within a wider frequency band, thereby obtaining a better noise reduction effect at the human ear position, which is hereinafter referred to as the compensation feedback control algorithm.

[0058] The effect of the present invention is illustrated below by taking the active noise reduction headrest system in the car cabin as an example. The active noise reduction headrest system in the cabin is usually arranged near the human ear, and the seat headrest speaker is used as a secondary sound source to generate a quiet zone at the human ear. Each ear uses one control channel, including one secondary source and one microphone installed near the secondary source, and the distance between the microphone and the human ear is about 10 cm. There are two channels for two ears. Generally speaking, the two channels of the active noise reduction headrest system are weakly coupled, and each channel can be regarded as a single-channel system. The common method for the active noise reduction headrest system using feedback control strategy is to directly use the microphone signal for feedback control (direct feedback control, Direct control). The method proposed in the present invention is implemented using the above steps 1-7, and noise reduction is achieved at the human ear position using compensation feedback control.

[0059] In order to illustrate the performance of this method, the following experiment is used to compare and verify the noise reduction performance of the direct feedback control algorithm and the compensation feedback control algorithm proposed by the present invention at the human ear. The system sampling rate is set to 16kHz, and the secondary path transfer function from the secondary source to the microphone and the human ear is an M=512-point FIR filter. The measurement results are and Driving an electric car at a constant speed, recording the primary noise signals at the microphone and the human ear, and modeling the transfer function between the microphone and the human ear Its length is L=512. Then, the compensation filter C(z) is obtained by offline optimization, and the length of the compensation filter is set to K=512. Figure 3 The figure shows the frequency response comparison between the optimized compensation filter and its unconstrained solution in the frequency domain. The gray solid line is the unconstrained solution in the frequency domain, and the black dotted line is the optimized compensation filter. It can be seen that the optimization result is in good agreement with the unconstrained solution in the frequency range of 100 to 400 Hz.

[0060] During the control process, the servo control system (i.e. Figure 1 The black dashed line in the middle (shown in the middle) adjusts the secondary sound field at the microphone to track the target control signal Assume the control filter length I is 512, Figure 4 Compared with the target control signal The secondary signal generated at the microphone by the servo control system The gray solid line in the figure is the spectrum of the microphone signal using the compensation filter C(z). The target control signal generated by compensation The black dotted line is the tracking result of the servo control system on the target control signal. Since the low-frequency response of the secondary source is weak, it is difficult to effectively track the target control signal with many low-frequency components. Therefore, the A-weighted filter network is selected to constrain the low-frequency components of the control target. The black dotted line in the figure is the target control signal tracking result after being constrained by the A-weighted filter network.

[0061] Figure 5 The figure shows the noise reduction results of the real car at the right microphone and right ear position at a speed of 60km / h. The gray solid line in the figure is the primary noise spectrum, and the black solid line and gray dotted line are the error signal spectra after noise reduction using compensation feedback control and direct feedback control, respectively. At the microphone position, Figure 5 (a) shows that the overall noise reduction effect of direct feedback control in the 100-500Hz frequency band is good, and the A-weighted noise reduction of direct feedback control and compensation feedback control is 3.8dBA and 1.6dBA respectively. At the target, that is, the right ear position, Figure 5 (b) shows that the main noise reduction frequency band of compensation feedback control and direct feedback control is 150-400Hz, among which compensation feedback control can achieve better noise reduction effect in the narrow band of 170-350Hz, and the average noise reduction in this band is 2.0dBA greater than that of direct feedback control, but the compensation feedback control in the frequency band of 400-550Hz has a slight noise amplification. The A-weighted noise reduction of compensation feedback control in the full frequency band is 1.8dBA, which is 0.3dBA higher than that of direct feedback control.

[0062] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A feedback active noise control method based on sound field compensation, characterized in that: The steps include: Step 1: Measure the transfer function from the secondary source to the microphone in the active noise control system and obtain the transfer function estimate n is the time index, M is the path unit impulse response length, and the superscript T is the transposition operator symbol. The z-domain form of Step 2: Measure the transfer function from the secondary source to the target noise reduction position in the active noise control system and obtain the transfer function estimate Its z-domain form can be expressed as Step 3: Measure the transfer function between the microphone and the target noise reduction position in the primary noise field, and obtain the transfer function as follows: L is its unit impulse response length, The z-domain form of Step 4, use and Primary noise of microphone p m (n) Filter to get the expected signal p′ m (n), the z-domain of this process is expressed as Use pm(n) The filtered signal rm(n) is used as the reference signal, and then the least mean square algorithm is used to identify the reference signal and the expected signal p′ m (n), the compensation filter C(z) can be obtained, and its time domain optimal Wiener solution c opt =[c(0),c(1),...,c(K-1)] T (K is the compensation filter length) is: in is the reference signal r m (n) = [r m (n), r m (n-1), ..., r m (n-K+1)] T The autocorrelation matrix of For r m (n) and the expected signal p′ m (n) is the cross-correlation vector. Step 5: Use the controller output signal u(n) Filter to get the secondary sound estimate at the microphone After taking the negative value, it is equal to the signal e picked up by the microphone m( n) to obtain the primary noise estimate at the microphone The z-domain calculation process can be expressed as in E m (z) and U(z) are e m The z-domain form of (n) and u(n); Step 6: Use the compensation filter C(z) to estimate the microphone primary noise Filter to get the target control signal at the microphone Step 7: Use the servo control system to adjust the secondary sound field at the microphone. Regulate to minimize the cost function shown in formula (2), in for With the target control signal The servo tracking system is an internal model control structure, and its control filter w(n) = [w(0), w(1), ..., w(I-1)] T (I is the control filter length) the iterative formula is: w(n+1)=w(n)+μe t (n)s′ t (n) (3) Where μ is the iteration step size, s′ t (n) for use Control signal to target The filtered signal vector. Step 8, continuously iteratively control the filter coefficients to minimize the cost function shown in equation (2) and finally achieve noise reduction at the target position.

2. A feedback active noise control method based on sound field compensation as claimed in claim 1, characterized in that: Ideally, it is assumed that there is no error between the transfer function estimates in steps 1-3 and the transfer function, that is, The optimal solution of the compensation filter is C opt (z)=-Q(z)G m (z) / G e (z).

3. A feedback active noise control method based on sound field compensation as claimed in claims 1 and 2, characterized in that: In the actual implementation process of the compensation filter C(z) described in step 4, a finite length impulse response (FIR) filter can be used to fit the compensation value -Q(z)G m (z) / G e (z) frequency response.

4. According to the feedback active noise control method based on sound field compensation as claimed in claims 1-3, the controller H(z) in the servo system in step 7 is: where W(z) is the z-domain form of w(n).

Images