Robust front feedback hybrid active noise control system and method

Abstract

The invention discloses a robust front feedback hybrid active noise control system and method, and belongs to the technical field of active noise control. According to the system provided by the invention, by utilizing the linear prediction filtering subsystem, the broadband auxiliary filtering subsystem and the narrowband auxiliary filtering subsystem, the independence among the feedforward broadband active noise control subsystem, the feedforward narrowband active noise control subsystem and the feedback active noise control subsystem can be improved, and the convergence performance of the whole system is improved; and the separation performance of the broadband auxiliary filtering subsystem and the narrowband auxiliary filtering subsystem on the residual noise is respectively improved by using an efficient self-adaptive variable step size minimum mean square algorithm. The system provided by the invention not only can suppress the wide-narrow band mixed noise component related to the reference signal in the target noise, but also can suppress the narrow-band noise component from other noise sources in the target noise, thereby effectively overcoming the coupling influence among multiple noise sources, improving the noise reduction performance of the system, and widening the practical application range of the system.

Description

Technical Field

[0001] The invention relates to a robust forward feedback hybrid active noise control system and method, belonging to the technical field of active noise control. Background Art

[0002] Passive noise control technology achieves the purpose of noise reduction by means of material or structural design, such as sound elimination, sound absorption or sound barrier. It has good noise reduction effect on medium and high frequency audible sounds, but its volume is large and the cost is often high. In contrast, active noise control (ANC) technology is based on the principle of destructive interference of sound waves. It not only has good low-frequency noise suppression performance, but also has the advantages of small volume and low cost. It is a beneficial supplement to traditional passive noise control technology (L. Lu, K. Yin, RC de Lamare, et al, A survey on active noise control in the past decade, Part I: Linear systems, Signal Process., 183 (2021), 108039.). Active noise control technology has gradually been applied to noise reduction in automotive cockpits, rotating machinery and other occasions. According to the structure of the controller, ANC systems can be divided into three types: feedforward, feedback and feedforward hybrid.

[0003] In actual factories, there are a large number of periodic harmful noises generated by rotating equipment (such as cutting machine noise, engine noise, etc.), which usually have the characteristics of wide-band and narrow-band mixing. When the feedforward ANC system uses an acoustic sensor to obtain a reference signal, a corresponding feedforward active noise control system is designed to deal with the above wide-band and narrow-band mixed noise (Y. Xiao and J. Wang, A new feedforward hybrid active noise control system, IEEE Signal Process. Letts., 18 (10) (2011) 591-594.). In actual working conditions, due to the simultaneous operation of multiple rotating machines in the workshop, the target noise generated by a certain device may have both wide-band and narrow-band target noise components related to the reference signal and narrow-band target noise components from other devices that are unrelated to the reference signal, resulting in mutual coupling between multiple noise sources. Under this working condition, if only a feedforward ANC system is used, its noise reduction performance will be suppressed. At this time, it is necessary to introduce a feedback active noise control structure to deal with the narrow-band target noise components that are unrelated to the reference signal. Therefore, the development of a high-performance feedforward and feedback-type combined feedforward hybrid ANC system has more important practical application value.

[0004] A feedforward hybrid ANC system introduces an auxiliary filtering subsystem to enhance the independence between the feedforward active noise control subsystem and the feedback active noise control subsystem, and the noise reduction performance is significantly improved (MTAkhtar and W.Mitsuhashi, Improving performance of hybrid active noise control systems foruncorrelated narrowband disturbances, IEEE / ACM Trans.Audio, Speech, Lang.Process., vol.19, no.7, pp.2058-2066, Sept.2011.). Then, the system is based on the independence between the reference signal-related components and the reference signal-independent components in the target noise. However, the actual noise environment is relatively complex, mainly reflected in: 1) The target noise component related to the reference signal has the characteristics of wide-narrow band mixing; 2) The narrowband target noise from other sound sources may leak into the reference microphone through the acoustic channel, causing coupling effects between multiple sound sources. Therefore, the noise reduction performance of the above-mentioned feedforward hybrid ANC system for this complex noise is limited.

[0005] In order to simultaneously solve the problems that restrict system performance, such as the poor independence of the above-mentioned control subsystems, the wide-band and narrow-band mixing of target noise, and the mutual coupling between multiple sound sources, it is urgent to provide a more effective and practical forward feedback hybrid active noise control system and method to promote the application of active noise control technology. Summary of the invention

[0006] In order to solve the problems that still exist in the conventional forward feedback hybrid active noise control system, such as poor independence of each control subsystem, wide-band and narrow-band mixing of target noise, and mutual coupling between multiple sound sources, which restrict the system performance, the present invention provides a robust forward feedback hybrid active noise control system and method, and the technical solution is as follows:

[0007] The first object of the present invention is to provide a robust feedforward hybrid active noise control system, wherein the active noise control system uses a reference microphone to obtain a reference signal, a secondary loudspeaker to provide a secondary sound source, and an error microphone to obtain residual noise; the actual primary channel in the acoustic space is a channel model of the reference signal propagating from the sound source position to the error microphone; the actual secondary channel in the acoustic space is a channel model of the secondary sound source propagating to the error microphone via the secondary loudspeaker; the active noise control system includes a linear prediction filter subsystem, a feedforward broadband active noise control subsystem, a broadband auxiliary filter subsystem, a feedforward narrowband active noise control subsystem, a narrowband auxiliary filter subsystem, and a feedback active noise control subsystem;

[0008] The linear prediction filter subsystem is connected to the feedforward broadband active noise control subsystem, the broadband auxiliary filter subsystem, the feedforward narrowband active noise control subsystem, and the narrowband auxiliary filter subsystem respectively; the feedforward broadband active noise control subsystem is connected to the broadband auxiliary filter subsystem; the broadband auxiliary filter subsystem is connected to the narrowband auxiliary filter subsystem; the feedforward narrowband active noise control subsystem is connected to the narrowband auxiliary filter subsystem; the narrowband auxiliary filter subsystem is connected to the feedback active noise control subsystem;

[0009] The linear prediction filter subsystem is used to separate a broadband reference signal and a narrowband reference signal from the reference signal; the separated broadband reference signal is used as a reference input of the feedforward broadband active noise control subsystem and the broadband auxiliary filter subsystem; and the separated narrowband reference signal is used as a reference input of the feedforward narrowband active noise control subsystem and the narrowband auxiliary filter subsystem;

[0010] The feedforward broadband active noise control subsystem is used to synthesize broadband secondary sound sources related to broadband reference signals; the broadband auxiliary filtering subsystem is used to separate broadband residual noise and narrowband residual noise from residual noise; the feedforward narrowband active noise control subsystem is used to synthesize narrowband secondary sound sources related to narrowband reference signals; the narrowband auxiliary filtering subsystem is used to separate narrowband residual noise related to narrowband reference signals and narrowband residual noise unrelated to narrowband reference signals from narrowband residual noise; the feedback active noise control subsystem is used to synthesize narrowband secondary sound sources unrelated to narrowband reference signals;

[0011] The broadband residual noise and narrowband residual noise separated by the broadband auxiliary filtering subsystem are used as the error output of the feedforward broadband active noise control subsystem and the desired input of the narrowband auxiliary filtering subsystem, respectively; at the same time, the narrowband residual noise related to the narrowband reference signal separated by the narrowband auxiliary filtering subsystem is used as the error output of the feedforward narrowband active noise control subsystem; at the same time, the narrowband residual noise independent of the narrowband reference signal separated by the narrowband auxiliary filtering subsystem is used as the error output of the feedback active noise control subsystem and the reference input for synthesizing the feedback active noise control subsystem, respectively.

[0012] In one embodiment of the present invention, the linear prediction filter subsystem uses the reference signal x1(n) as input and separates the broadband reference signal x w (n) and the narrowband reference signal x f(n), and used as reference inputs of the feedforward broadband active noise control subsystem and the feedforward narrowband active noise control subsystem, respectively, to enhance the independence between the feedforward broadband active noise control subsystem and the feedforward narrowband active noise control subsystem;

[0013] The linear prediction filter subsystem includes a delay link and a linear prediction filter, the delay link and the linear prediction filter are cascaded, the order of the delay link is D; the linear prediction filter is a finite length impulse response filter model, and its coefficients and lengths are respectively and L Q , whose coefficients are updated using the least mean square algorithm, namely:

[0014] w Q (n+1)=w Q (n)+μ Q x w (n)x1(nDj)

[0015] Among them, μ Q is the update step size of the linear prediction filter, which is a positive value; n is the time, n≥0;

[0016] The narrowband reference signal and the wideband reference signal synthesized by the linear prediction filter subsystem are respectively:

[0017]

[0018] x w (n) = x1(n) - x f (n).

[0019] In one embodiment of the present invention, the feedforward broadband active noise control subsystem includes a feedforward broadband controller and a first FXLMS algorithm module; the feedforward broadband controller uses a linear finite impulse response filter, whose coefficients and lengths are respectively and L1;

[0020] The first FXLMS algorithm module uses the broadband auxiliary filtering subsystem to separate the broadband reference signal x w (n) The broadband residual noise y B (n) is used as an error output and is used to update the coefficients of the feedforward broadband controller; the coefficient update formula of the feedforward broadband controller is:

[0021]

[0022] Wherein, μ1 is the update step size of the feedforward broadband controller, which is a positive value; is the broadband reference signal x w(n) the output after the secondary channel estimation model in the first FXLMS algorithm module; the secondary channel estimation model The linear FIR filter is obtained by offline identification, and its coefficients and length are and

[0023] The feedforward broadband active noise control subsystem synthesizes a broadband secondary sound source related to the broadband reference signal as:

[0024]

[0025] In one embodiment of the present invention, the broadband auxiliary filtering subsystem includes a broadband auxiliary filter and a first VSS-LMS algorithm module;

[0026] The broadband auxiliary filter adopts a linear FIR filter, whose coefficients and lengths are respectively and L B , using the first VSS-LMS algorithm module to update the broadband auxiliary filter, that is:

[0027] w B,j (n+1)=w B,j (n)+μ B (n)e B (n)x w (nj)

[0028] Among them, μ B (n) is the update step size of the broadband auxiliary filter; e B (n) is the narrowband residual noise separated by the broadband auxiliary filtering subsystem;

[0029] The broadband residual noise related to the broadband reference signal separated by the broadband auxiliary filtering subsystem is:

[0030]

[0031] The narrowband residual noise separated by the broadband auxiliary filtering subsystem is B (n) = e(n) - y B (n), used as the desired input of the narrowband auxiliary filtering subsystem;

[0032] The first VSS-LMS algorithm module adaptively adjusts the step size of the broadband auxiliary filter to improve the separation performance of the broadband residual noise and the narrowband residual noise of the broadband auxiliary filtering subsystem. The variable step size strategy is:

[0033]

[0034] Among them, ξ B, η B , γ B1 and γ B2 is the user parameter, ξ B ∈(0.9,1),η B Take positive value, γ B1 and γ B2 The values ​​are all integers greater than or equal to zero.

[0035] In one embodiment of the present invention, the feedforward narrowband active noise control subsystem includes the feedforward narrowband controller and the second FXLMS algorithm module; the feedforward narrowband controller uses a linear FIR filter, whose coefficients and lengths are respectively and L2;

[0036] The second FXLMS algorithm module uses the narrowband auxiliary filtering subsystem to separate the narrowband reference signal x f (n) The relevant narrowband residual noise y N (n) is used as an error output and is used to update the coefficients of the feedforward narrowband controller; the coefficient update formula of the feedforward narrowband controller is:

[0037]

[0038] Wherein, μ2 is the update step size of the feedforward narrowband controller, which is a positive value; is the narrowband reference signal x f (n) output after passing through the secondary channel estimation model in the second FXLMS algorithm module;

[0039] The feedforward narrowband active noise control subsystem synthesizes a narrowband secondary sound source related to the narrowband reference signal as follows:

[0040]

[0041] In one embodiment of the present invention, the narrowband auxiliary filtering subsystem includes a narrowband auxiliary filter and a second VSS-LMS algorithm module;

[0042] The narrowband auxiliary filter adopts a linear FIR filter, whose coefficients and lengths are respectively and L N , using the second VSS-LMS algorithm module to update the narrowband auxiliary filter, that is:

[0043] w N,j (n+1)=w N,j (n)+μ N (n)e N (n)x f (nj)

[0044] Among them, μ N (n) is the update step size of the narrowband auxiliary filter; e N (n) is the narrowband residual noise separated by the narrowband auxiliary filtering subsystem and is independent of the narrowband reference signal;

[0045] The narrowband residual noise related to the narrowband reference signal separated by the narrowband auxiliary filtering subsystem is:

[0046]

[0047] The narrowband residual noise separated by the narrowband auxiliary filtering subsystem is independent of the narrowband reference signal. N (n) = e B (n)-y N (n), which is used as the error output of the feedback active noise control subsystem and is also used to synthesize the reference input x2(n) of the feedback active noise control subsystem.

[0048] In one embodiment of the present invention, the second VSS-LMS algorithm module adaptively adjusts the step size of the narrowband auxiliary filter to improve the narrowband residual noise separation performance of the narrowband auxiliary filtering subsystem related to and unrelated to the narrowband reference signal, and the variable step size strategy is:

[0049]

[0050] Among them, ξ N , η N , γ N1 and γ N2 is the user parameter, ξ N ∈(0.9,1),η N Take positive value, γ N1 and γ N2 The values ​​are all integers greater than or equal to zero.

[0051] In one embodiment of the present invention, the feedback active noise control subsystem includes an internal reference signal synthesis module, a feedback controller and a third FXLMS algorithm module. The internal reference signal synthesis module includes a secondary channel estimation model and a first-order delay link, and synthesizes an internal reference signal x2(n), that is:

[0052]

[0053] in, is the output of the feedback controller after passing through the secondary channel estimation model;

[0054] The feedback controller uses a linear FIR filter, whose coefficients and lengths are and L3;

[0055] The third FXLMS algorithm module uses the narrowband residual noise e that is independent of the narrowband reference signal separated by the narrowband auxiliary filtering subsystem. N (n) is used as the error output and is used to update the coefficients of the feedback controller. The updating formula is:

[0056]

[0057] Wherein, μ3 is the update step size of the feedback controller, which is a positive value; The output of x2(n) provided by the internal reference signal synthesis module after passing through the secondary channel estimation model in the third FXLMS algorithm module;

[0058] The narrowband secondary sound source obtained by the feedback active noise control subsystem and independent of the narrowband reference signal is:

[0059]

[0060] It is then used to suppress narrowband noise components in the target noise that are unrelated to the reference signal, so as to improve the noise reduction performance of the system.

[0061] In one embodiment of the present invention, the synthesized secondary sound source is:

[0062] y(n)=y1(n)+y2(n)+y3(n)

[0063] Furthermore, y(n) is provided to the secondary speaker and destructively interferes with the target noise in the acoustic space.

[0064] A second object of the present invention is to provide a robust feedforward hybrid active noise control method, the method using the robust feedforward hybrid active noise control system, the method comprising:

[0065] Step 1: Set user parameters and system initialization;

[0066] Respectively set the update step of the linear prediction filter subsystem, the feedforward broadband active noise control subsystem, the broadband auxiliary filter subsystem, the feedforward narrowband active noise control subsystem, the narrowband auxiliary filter subsystem, and the feedback active noise control subsystem, and set the length of the linear FIR filter and the initial value to zero; set the order D of the delay link; set the user parameters of the first VSS-LMS algorithm module and the second VSS-LMS algorithm module;

[0067] Step 2: Obtain a reference signal;

[0068] At time n, the reference signal x1(n) obtained by the reference microphone is passed through the linear prediction filter subsystem to obtain a broadband reference signal and a narrowband reference signal; the separated broadband reference signal is provided to the feedforward broadband secondary sound source synthesis subsystem and the broadband auxiliary filter subsystem respectively; the separated narrowband reference signal is provided to the feedforward narrowband secondary sound source synthesis subsystem and the narrowband auxiliary filter subsystem respectively; the internal reference signal synthesis module and the e at time n-1 are used to synthesize the reference signal. N (n-1) and Obtain the internal reference signal x2(n);

[0069] Step 3: At time n, first, the secondary sound source is synthesized by using the feedforward broadband controller, the feedforward narrowband controller and the feedback controller; then, the broadband auxiliary filtering subsystem is used to separate the broadband residual noise and the narrowband residual noise from the residual noise; finally, the narrowband auxiliary filtering subsystem is used to separate the narrowband residual noise related to the narrowband reference signal and the narrowband residual noise unrelated to the narrowband reference signal from the narrowband residual noise;

[0070] Step 4: Control system update;

[0071] The linear FIR filter coefficients in the linear prediction filter subsystem, the feedforward broadband active noise control subsystem, the broadband auxiliary filter subsystem, the feedforward narrowband active noise control subsystem, the narrowband auxiliary filter subsystem, and the feedback active noise control subsystem are updated respectively, and the coefficients of the corresponding linear FIR filters at time n+1 are obtained respectively; the step sizes in the first VSS-LMS algorithm module and the second VSS-LMS algorithm module are adaptively adjusted respectively, and the corresponding step sizes at time n+1 are obtained respectively;

[0072] Step 5: Return to step 2, and repeat steps 2 to 4 until the system gradually converges and reaches a steady state, thereby achieving noise control.

[0073] The beneficial effects of the present invention are:

[0074] 1. The present invention does not need to use a non-acoustic reference microphone to obtain frequency prior information, thus effectively avoiding the frequency offset problem.

[0075] 2. The present invention utilizes a linear prediction subsystem to achieve effective separation of narrowband reference signals and broadband reference signals, and provides reliable reference inputs for the feedforward narrowband controller and the feedforward broadband controller, respectively.

[0076] 3. The present invention introduces the VSS-LMS algorithm to improve the separation performance of the broadband auxiliary filtering subsystem and the narrowband auxiliary filtering subsystem for residual noise.

[0077] 4. The present invention utilizes the broadband residual noise related to the broadband reference signal separated by the broadband auxiliary filtering subsystem to update the feedforward broadband controller, utilizes the narrowband residual noise related to the narrowband reference signal separated by the narrowband auxiliary filtering subsystem to update the feedforward narrowband controller, and utilizes the narrowband residual noise independent of the narrowband reference signal separated by the narrowband auxiliary filtering subsystem to update the feedback controller, which can enhance the independence among the feedforward broadband controller, the feedforward narrowband controller and the feedback controller, thereby improving the dynamic performance of the system.

[0078] 5. The present invention utilizes a feedforward broadband controller and a feedforward narrowband controller to jointly deal with the wide-band and narrow-band mixed target noise components related to the reference signal, and utilizes a feedback controller to deal with the narrowband target noise components unrelated to the narrowband reference signal, thereby overcoming the coupling effect between multiple noise sources, improving the overall noise reduction performance of the system, and meeting practical application requirements. BRIEF DESCRIPTION OF THE DRAWINGS

[0079] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0080] Figure 1 It is a schematic diagram of a robust forward feedback hybrid active noise control system according to the first embodiment of the present invention.

[0081] FIG. 2( a ) is a dynamic variation curve of the residual noise mean square error of the third embodiment of the present invention.

[0082] FIG. 2( b ) is a curve showing the variation of the noise reduction amount with the number of independent operations according to the third embodiment of the present invention.

[0083] FIG2( c ) is a step-size convergence curve of the first VSS-LMS algorithm according to the third embodiment of the present invention.

[0084] FIG2( d ) is a step-size convergence curve of the second VSS-LMS algorithm according to the third embodiment of the present invention.

[0085] FIG3( a ) is a dynamic variation curve of the residual noise mean square error of the fourth embodiment of the present invention.

[0086] FIG3( b ) is a curve showing the variation of the noise reduction amount with the number of independent operations according to the fourth embodiment of the present invention.

[0087] FIG3( c ) is a step-size convergence curve of the first VSS-LMS algorithm according to the fourth embodiment of the present invention.

[0088] FIG3( d ) is a step-size convergence curve of the second VSS-LMS algorithm according to the fourth embodiment of the present invention.

[0089] In the figure: 1. Linear prediction filter subsystem; 2. Feedforward broadband active noise control subsystem; 3. Broadband auxiliary filter subsystem; 4. Feedforward narrowband active noise control subsystem; 5. Narrowband auxiliary filter subsystem; 6. Feedback active noise control subsystem; 11. Delay link; 12. Linear prediction filter; 21. Feedforward broadband controller; 22. First FXLMS algorithm module; 31. Broadband auxiliary filter; 32. First VSS-LMS algorithm module; 41. Feedforward narrowband controller; 42. Second FXLMS algorithm module; 51. Narrowband auxiliary filter; 52. Second VSS-LMS algorithm module; 61. Internal reference signal synthesis module; 62. Feedback controller; 63. Third FXLMS algorithm module. DETAILED DESCRIPTION

[0090] The present invention is described in detail below. In the following paragraphs, different aspects of the embodiments are defined in more detail. Each aspect so defined can be combined with any other aspect or aspects, unless explicitly stated not to be combined. In particular, any feature that is considered to be preferred or advantageous can be combined with one or more other features that are considered to be preferred or advantageous.

[0091] The terms "first", "second" and the like appearing in the present invention are only for the convenience of description to distinguish different components with the same name, and do not indicate a sequence or a primary and secondary relationship.

[0092] In addition, when an element is referred to as being "on" another element, the element may be directly on the other element, or may be indirectly on the other element with one or more intermediate elements interposed therebetween. In addition, when an element is referred to as being "connected to" another element, the element may be directly connected to the other element, or may be indirectly connected to the other element with one or more intermediate elements interposed therebetween. Hereinafter, the same reference numerals represent the same elements.

[0093] Embodiment 1:

[0094] This embodiment provides a robust forward feedback hybrid active noise control system. Figure 1 As shown in the system schematic diagram, the active noise control system uses a reference microphone to obtain a reference signal, a secondary speaker to provide a secondary sound source, and an error microphone to obtain residual noise; the active noise control system includes a linear prediction filter subsystem 1, a feedforward broadband active noise control subsystem 2, a broadband auxiliary filter subsystem 3, a feedforward narrowband active noise control subsystem 4, a narrowband auxiliary filter subsystem 5, and a feedback active noise control subsystem 6;

[0095] The linear prediction filter subsystem 1 is connected to the feedforward broadband active noise control subsystem 2, the broadband auxiliary filter subsystem 3, the feedforward narrowband active noise control subsystem 4, and the narrowband auxiliary filter subsystem 5 respectively; the feedforward broadband active noise control subsystem 2 is connected to the linear prediction filter subsystem 1 and the broadband auxiliary filter subsystem 3; the broadband auxiliary filter subsystem 3 is connected to the linear prediction filter subsystem 1, the feedforward broadband active noise control subsystem 2, and the narrowband auxiliary filter subsystem 5; the feedforward narrowband active noise control subsystem 4 is connected to the linear prediction filter subsystem 1 and the narrowband auxiliary filter subsystem 5; the narrowband auxiliary filter subsystem 5 is connected to the broadband auxiliary filter subsystem 3, the feedforward narrowband active noise control subsystem 4, and the feedback active noise control subsystem 6; the feedback active noise control subsystem 6 is connected to the narrowband auxiliary filter subsystem 5;

[0096] The linear prediction filter subsystem 1 is used to separate a broadband reference signal and a narrowband reference signal from a reference signal; the separated broadband reference signal is used as a reference input of a feedforward broadband active noise control subsystem 2 and a broadband auxiliary filter subsystem 3; and the separated narrowband reference signal is used as a reference input of a narrowband auxiliary filter subsystem 5 of a feedforward narrowband active noise control subsystem 4;

[0097] The feedforward broadband active noise control subsystem 2 is used to synthesize broadband secondary sound sources related to broadband reference signals; the broadband auxiliary filtering subsystem 3 is used to separate broadband residual noise and narrowband residual noise from residual noise; the feedforward narrowband active noise control subsystem 4 is used to synthesize narrowband secondary sound sources related to narrowband reference signals; the narrowband auxiliary filtering subsystem 5 is used to further separate narrowband residual noise related to narrowband reference signals and narrowband residual noise unrelated to narrowband reference signals from narrowband residual noise; the feedback active noise control subsystem 6 is used to synthesize narrowband secondary sound sources unrelated to narrowband reference signals;

[0098] The broadband residual noise and narrowband residual noise separated by the broadband auxiliary filtering subsystem 3 are used as the error output of the feedforward broadband active noise control subsystem 2 and the expected input of the narrowband auxiliary filtering subsystem 5 respectively; at the same time, the narrowband residual noise related to the narrowband reference signal separated by the narrowband auxiliary filtering subsystem 5 is used as the error output of the feedforward narrowband active noise control subsystem 4; at the same time, the narrowband residual noise independent of the narrowband reference signal separated by the narrowband auxiliary filtering subsystem 5 is used as the error output of the feedback active noise control subsystem 6 and the reference input of the synthetic feedback active noise control subsystem 6 respectively; the independence among the feedforward broadband active noise control subsystem 2, the feedforward narrowband active noise control subsystem 4 and the feedback active noise control subsystem 6 can be improved, and the dynamic performance of the system can be improved; the noise reduction performance of the system in dealing with narrowband target noise components that are both related to the reference signal and independent of the reference signal can be improved.

[0099] like Figure 1 As shown, the actual primary channel P(z) in the acoustic space is the channel model of the reference signal propagating from the sound source position to the error microphone, which is represented by a linear FIR filter with coefficients and lengths respectively. and M p ; The actual secondary channel S(z) in the acoustic space is the channel model of the secondary sound source propagating from the secondary speaker to the error microphone, which is represented by a linear FIR filter with coefficients and lengths of and M s .

[0100] The target noise is:

[0101] p(n)=p1(n)+p2(n)+v p (n)

[0102] Where p1(n) is the signal of the corresponding reference signal x1(n) in the acoustic space propagated to the error microphone through the actual secondary channel S(z); p2(n) is the narrowband target noise component containing q narrowband frequencies in the acoustic space; v p (n) has a mean of zero and a variance of Additive Gaussian white noise; n is the moment, n≥0.

[0103] The reference signal x1(n) contains a broadband component x B (n) and the narrowband component x N (n), where the narrowband component x N (n) consists of two parts: one is the narrowband frequency component in a certain noise source; the other is the signal of the narrowband frequency component in other noise sources propagating from the error microphone through the acoustic channel to the reference microphone. The above two narrowband frequency components are independent or related.

[0104] Linear prediction filter subsystem 1 takes the reference signal x1(n) as input and separates the broadband reference signal x w (n) and the narrowband reference signal x f (n), and are used as reference inputs of the feedforward broadband active noise control subsystem 2 and the feedforward narrowband active noise control subsystem 4, respectively, to enhance the independence between the feedforward broadband active noise control subsystem 2 and the feedforward narrowband active noise control subsystem 4;

[0105] The linear prediction filter subsystem 1 includes a delay link 11 and a linear prediction filter 12. The delay link 11 and the linear prediction filter 12 are cascaded. The order of the delay link 11 is D. The linear prediction filter 12 is a finite length impulse response filter model, and its coefficients and lengths are respectively and L Q , whose coefficients are updated using the least mean square algorithm (LMS), namely:

[0106] w Q (n+1)=w Q (n)+μ Q x w (n)x1(nDj)

[0107] Among them, μ Q is the update step size of the linear prediction filter 12, which is a positive value;

[0108] The narrowband reference signal and the wideband reference signal synthesized by the linear prediction filter subsystem 1 are:

[0109]

[0110] x w (n) = x1(n) - x f (n).

[0111] The feedforward broadband active noise control subsystem 2 includes a feedforward broadband controller 21 and a first filtered-X LMS (Filtered-X LMS, FXLMS) algorithm module 22;

[0112] The feedforward broadband controller 21 uses a linear finite impulse response (FIR) filter, whose coefficients and lengths are and L1;

[0113] The first FXLMS algorithm module 22 uses the broadband auxiliary filtering subsystem 3 to separate the broadband reference signal x w (n) The broadband residual noise y B(n) is used as the error output and is used to update the coefficients of the feedforward broadband controller 21; the coefficient update formula of the feedforward broadband controller 21 is:

[0114]

[0115] Wherein, μ1 is the update step size of the feedforward broadband controller 21, which is a positive value; is the broadband reference signal x w (n) Output after the secondary channel estimation model in the first FXLMS algorithm module 22; Secondary channel estimation model The linear FIR filter is obtained by offline identification, and its coefficients and length are and

[0116] The feedforward broadband active noise control subsystem 2 synthesizes a broadband secondary sound source related to the broadband reference signal as:

[0117]

[0118] The broadband auxiliary filtering subsystem 3 includes a broadband auxiliary filter 31 and a first variable step-size LMS (VSS-LMS) algorithm module 32;

[0119] The broadband auxiliary filter 31 adopts a linear FIR filter, whose coefficients and lengths are respectively and L B , the broadband auxiliary filter 31 is updated using the first VSS-LMS algorithm module 32, that is:

[0120] w B,j (n+1)=w B,j (n)+μ B (n)e B (n)x w (nj)

[0121] Among them, μ B (n) is the update step size of the broadband auxiliary filter 31; B (n) is the narrowband residual noise separated by the broadband auxiliary filtering subsystem 3;

[0122] The broadband residual noise related to the broadband reference signal separated by the broadband auxiliary filtering subsystem 3 is:

[0123]

[0124] The narrowband residual noise separated by the broadband auxiliary filtering subsystem 3 is B (n) = e(n) - y B(n), used as the desired input of the narrowband auxiliary filtering subsystem 5.

[0125] The first VSS-LMS algorithm module 32 improves the separation performance of the broadband residual noise and the narrowband residual noise of the broadband auxiliary filtering subsystem 3 by adaptively adjusting the step size of the broadband auxiliary filter 31. The variable step size strategy is:

[0126]

[0127] Among them, ξ B , η B , γ B1 and γ B2 is the user parameter, ξ B ∈(0.9,1),η B Take a smaller positive value, γ B1 and γ B2 The values ​​are all integers greater than or equal to zero.

[0128] The feedforward narrowband active noise control subsystem 4 includes a feedforward narrowband controller 41 and a second FXLMS algorithm module 42; the feedforward narrowband controller 41 uses a linear FIR filter, whose coefficients and lengths are respectively and L2;

[0129] The second FXLMS algorithm module 42 uses the narrowband auxiliary filtering subsystem 5 to separate the narrowband reference signal x f (n) The relevant narrowband residual noise y N (n) is used as an error output and is used to update the coefficients of the feedforward narrowband controller 41; the coefficient update formula of the feedforward narrowband controller 41 is:

[0130]

[0131] Wherein, μ2 is the update step size of the feedforward narrowband controller 41, which is a positive value; is the narrowband reference signal x f (n) output after passing through the secondary channel estimation model in the second FXLMS algorithm module 42;

[0132] The feedforward narrowband active noise control subsystem 4 synthesizes the narrowband secondary sound source related to the narrowband reference signal as:

[0133]

[0134] The narrowband auxiliary filtering subsystem 5 includes a narrowband auxiliary filter 51 and a second VSS-LMS algorithm module 52;

[0135] The narrowband auxiliary filter 51 adopts a linear FIR filter, whose coefficients and lengths are respectively and LN , the coefficients of the narrowband auxiliary filter 51 are updated using the second VSS-LMS algorithm module 52, namely:

[0136] w N,j (n+1)=w N,j (n)+μ N (n)e N (n)x f (nj)

[0137] Among them, μ N (n) is the update step size of the narrowband auxiliary filter 51; N (n) is the narrowband residual noise separated by the narrowband auxiliary filtering subsystem 5 and is independent of the narrowband reference signal;

[0138] The narrowband residual noise related to the narrowband reference signal separated by the narrowband auxiliary filtering subsystem 5 is:

[0139]

[0140] The narrowband residual noise e that is independent of the narrowband reference signal separated by the narrowband auxiliary filtering subsystem 5 N (n) = e B (n)-y N (n), used as the error output of the feedback active noise control subsystem 6, and also used to synthesize the reference input x2(n) of the feedback active noise control subsystem 6.

[0141] The second VSS-LMS algorithm module 52 further improves the separation performance of the narrowband residual noise related to and unrelated to the narrowband reference signal of the narrowband auxiliary filtering subsystem 5 by adaptively adjusting the step size of the narrowband auxiliary filter 51. The variable step size strategy is:

[0142]

[0143] Among them, ξ N , η N , γ N1 and γ N2 is the user parameter, ξ N ∈(0.9,1),η N Take a smaller positive value, γ N1 and γ N2 The values ​​are all integers greater than or equal to zero.

[0144] The feedback active noise control subsystem 6 includes an internal reference signal synthesis module 61, a feedback controller 62 and a third FXLMS algorithm module 63. The internal reference signal synthesis module 61 includes a secondary channel estimation model and a first-order delay link, and synthesizes the internal reference signal x2(n), that is:

[0145]

[0146] in, is the output of the feedback controller 62 after passing through the secondary channel estimation model;

[0147] The feedback controller 62 uses a linear FIR filter, whose coefficients and lengths are and L3;

[0148] The third FXLMS algorithm module 63 uses the narrowband residual noise e that is independent of the narrowband reference signal separated by the narrowband auxiliary filtering subsystem 5 N (n) is used as the error output and is used to update the coefficients of the feedback controller 62. The update formula is:

[0149]

[0150] Wherein, μ3 is the update step size of the feedback controller 62, which is a positive value; The output of x2(n) provided by the internal reference signal synthesis module 61 after passing through the secondary channel estimation model in the third FXLMS algorithm module 63;

[0151] The narrow-band secondary sound source obtained by the feedback active noise control subsystem 6 and having nothing to do with the narrow-band reference signal is:

[0152]

[0153] It is then used to suppress narrowband noise components in the target noise that are unrelated to the reference signal, further improving the noise reduction performance of the system.

[0154] The synthesized secondary sound source is:

[0155] y(n)=y1(n)+y2(n)+y3(n)

[0156] Furthermore, the target noise p(n) and the signal y after the secondary sound source y(n) provided by the secondary speaker passes through the actual secondary channel S(z) p The difference between (n) and (n) is the residual noise, that is, e(n) = p(n) - y p (n) Interference and destructive interference in the acoustic space to achieve active noise control.

[0157] In summary, the system provided by the present invention utilizes a linear prediction filter subsystem, a broadband auxiliary filter subsystem, and a narrowband auxiliary filter subsystem to enhance the independence among the feedforward broadband active noise control subsystem, the feedforward narrowband active noise control subsystem, and the feedback active noise control subsystem, and improve the convergence performance of the overall system; utilizing an efficient adaptive variable step-size minimum mean square algorithm, the separation performance of the broadband auxiliary filter subsystem and the narrowband auxiliary filter subsystem for residual noise is enhanced respectively. The system provided by the present invention can not only suppress the broadband and narrowband mixed noise components related to the reference signal in the target noise, but also suppress the narrowband noise components from other noise sources in the target noise, effectively overcome the coupling effect between multiple noise sources, enhance the noise reduction performance of the system, and broaden its practical application scope.

[0158] Embodiment 2:

[0159] This embodiment provides a forward feedback hybrid active noise control method, which uses the foregoing forward feedback hybrid active noise control system. The method includes:

[0160] Step 1: Set user parameters and system initialization;

[0161] Respectively set the update step of the linear prediction filter subsystem 1, the feedforward broadband active noise control subsystem 2, the broadband auxiliary filter subsystem 3, the feedforward narrowband active noise control subsystem 4, the narrowband auxiliary filter subsystem 5, and the feedback active noise control subsystem 6, and set the length of the linear FIR filter and the initial value to zero; set the order D of the delay link 11; set the user parameters of the first VSS-LMS algorithm module 32 and the second VSS-LMS algorithm module 52;

[0162] Step 2: Obtain a reference signal;

[0163] At time n, the reference signal x1(n) obtained by the reference microphone is passed through the linear prediction filter subsystem 1 to obtain a broadband reference signal and a narrowband reference signal; the separated broadband reference signal is provided to the feedforward broadband secondary sound source synthesis subsystem 2 and the broadband auxiliary filter subsystem 3 respectively; the separated narrowband reference signal is provided to the feedforward narrowband secondary sound source synthesis subsystem 4 and the narrowband auxiliary filter subsystem 5 respectively; the internal reference signal synthesis module 61 and the e at time n-1 are used to synthesize the reference signal. N (n-1) and Obtain the internal reference signal x2(n);

[0164] Step 3: At time n, first, the feedforward broadband controller 21, the feedforward narrowband controller 41 and the feedback controller 62 are used to synthesize the secondary sound source; then, the broadband auxiliary filtering subsystem 3 is used to separate the broadband residual noise and the narrowband residual noise from the residual noise; finally, the narrowband auxiliary filtering subsystem 5 is used to separate the narrowband residual noise related to the narrowband reference signal and the narrowband residual noise unrelated to the narrowband reference signal from the narrowband residual noise;

[0165] Step 4: Control system update;

[0166] The linear FIR filter coefficients in the linear prediction filter subsystem 1, the feedforward broadband active noise control subsystem 2, the broadband auxiliary filter subsystem 3, the feedforward narrowband active noise control subsystem 4, the narrowband auxiliary filter subsystem 5, and the feedback active noise control subsystem 6 are updated respectively, and the coefficients of the corresponding linear FIR filters at the time n+1 are obtained respectively; the step sizes in the first VSS-LMS algorithm module 32 and the second VSS-LMS algorithm module 52 are adaptively adjusted respectively, and the corresponding step sizes at the time n+1 are obtained respectively;

[0167] Step 5: Return to step 2 above, and repeat steps 2 to 4 above until the system gradually converges and reaches a steady state, thus achieving noise control.

[0168] Example 3: Simulation scenario 1 verification

[0169] Assume that the narrowband component x in the reference signal x1(n) N (n) consists of two parts: one is a narrowband component x containing three frequencies from a noise source N,1 (n), the normalized angular frequencies are ω1 = 0.1π, ω2 = 0.15π, and ω3 = 0.2π, the corresponding cosine amplitudes are 1, 0.5, and 0.25, and the initial phases are 0, Another narrowband target noise component p2(n) from another noise source propagates to the reference microphone signal x N,2 (n). The frequencies of p2(n) are 0.15π and 0.3π, the corresponding cosine amplitudes are 1 and 0.5, and the initial phases are 0. B (n) is Gaussian white noise, with a mean of 0 and a variance equal to the narrowband component x N The additive noise in the target noise p(n) is a Gaussian white noise with a mean of zero and a variance of 0.01. It is assumed that the secondary channel estimation model is the same as the actual secondary channel S(z), and both use a linear FIR model with a length and cutoff frequency of 21 and 0.45π respectively; the length and cutoff frequency of the actual primary channel are 48 and 0.45π respectively.

[0170] The length of the feedforward broadband controller, feedforward narrowband controller and feedback controller are all 37, and the corresponding step sizes are 0.0005, 0.0005 and 0.001 respectively. The length of the linear prediction filter is 120, the corresponding step size is 0.0001, and the delay link order is 10. The length of the narrowband auxiliary filter and the broadband auxiliary filter are all 37, and the corresponding step size updates the required user parameter ξ N =ξ B =0.9996, η N =η B =0.00001,γ N1 =γ B1 =2 and γ N2 =γ B2 = 2. The length of simulation data is 30000; the number of independent runs is 100.

[0171] FIG2(a) is a dynamic change curve of the residual noise mean square error of Example 3; FIG2(b) is a curve of the noise reduction amount change with the number of independent runs of Example 3; FIG2(c) is a step-length convergence curve of the first VSS-LMS algorithm of Example 3; FIG2(d) is a step-length convergence curve of the second VSS-LMS algorithm of Example 3. As shown in FIG2(a), the system of the present invention has good dynamic performance, and the steady-state value of the residual noise mean square error of the system is 0.12, which tends to the variance of the additive Gaussian white noise in the target noise, indicating that the system of the present invention can not only suppress the wide- and narrow-band mixed target noise components, but also suppress the narrow-band target noise components from other noise sources. FIG2(b) further shows that the average noise reduction amount of the system of the present invention is 13.91dB, and good noise suppression performance is obtained each time it is independently run, which indirectly shows that the system has good stability. FIG2(c) and FIG2(d) show that the VSS-FXLMS algorithm of the present invention improves the dynamic performance of the broadband auxiliary filter and the narrowband auxiliary filter by adaptively adjusting the step size.

[0172] Example 4: Verification of Simulation Scenario 2

[0173] Suppose the broadband component x in the reference signal x1(n) is B (n) is colored noise, using x B (n) = 0.75x B (n-1)+randn(1,1) is generated, and its variance is equal to the narrowband component x N The step size of the feedforward broadband controller, feedforward narrowband controller and feedback controller is 0.0002. The variable step size updates the required user parameter ξ N =ξ B =0.99975, η N =η B =2×10 -6; The simulation data length is 40000; other parameter settings are the same as those in Example 3.

[0174] FIG3(a) is a dynamic change curve of the residual noise mean square error of Example 4; FIG3(b) is a curve of the noise reduction amount change with the number of independent runs of Example 4; FIG3(c) is a step-size convergence curve of the first VSS-LMS algorithm of Example 4; FIG3(d) is a step-size convergence curve of the second VSS-LMS algorithm of Example 4. As shown in FIG3(a), when the broadband noise component in the reference signal is colored noise, the system of the present invention still has good dynamic performance, and the steady-state value of the residual noise mean square error of the system is 0.14, which tends to the variance of the additive Gaussian white noise in the target noise, further indicating that the system of the present invention can not only suppress the wide- and narrow-band mixed target noise components, but also suppress the narrow-band target noise components from other noise sources. FIG3(b) further shows that the average noise reduction amount of the system of the present invention is 14.97 dB, and good noise suppression performance is obtained in each independent run, which also indirectly indicates that the system has good stability. FIG3( c ) and FIG3( d ) further show that the VSS-FXLMS algorithm of the present invention can improve the dynamic performance of the broadband auxiliary filter and the narrowband auxiliary filter by adaptively adjusting the step size.

[0175] The above embodiments 3 and 4 respectively verify the effectiveness and practicality of a robust forward feedback hybrid active noise control system and method provided by the present invention from the two cases where the reference signal contains a colored broadband noise component and a white broadband noise component, which will further promote the practical application of active noise control technology.

[0176] Some steps in the embodiments of the present invention may be implemented using software, and the corresponding software program may be stored in a readable storage medium, such as a CD or a hard disk.

[0177] Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments, or to make equivalent replacements for some of the technical features therein. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A robust feedforward hybrid active noise control system, characterized in that: The active noise control system uses a reference microphone to obtain a reference signal, a secondary loudspeaker to provide a secondary sound source, and an error microphone to obtain residual noise; the actual primary channel in the acoustic space is a channel model of the reference signal propagating from the sound source position to the error microphone; the actual secondary channel in the acoustic space is a channel model of the secondary sound source propagating to the error microphone via the secondary loudspeaker; the active noise control system comprises a linear prediction filter subsystem (1), a feedforward broadband active noise control subsystem (2), a broadband auxiliary filter subsystem (3), a feedforward narrowband active noise control subsystem (4), a narrowband auxiliary filter subsystem (5), and a feedback active noise control subsystem (6); The linear prediction filter subsystem (1) is respectively connected to the feedforward broadband active noise control subsystem (2), the broadband auxiliary filter subsystem (3), the feedforward narrowband active noise control subsystem (4), and the narrowband auxiliary filter subsystem (5); the feedforward broadband active noise control subsystem (2) is connected to the broadband auxiliary filter subsystem (3); the broadband auxiliary filter subsystem (3) is connected to the narrowband auxiliary filter subsystem (5); the feedforward narrowband active noise control subsystem (4) is connected to the narrowband auxiliary filter subsystem (5); the narrowband auxiliary filter subsystem (5) is connected to the feedback active noise control subsystem (6); The linear prediction filter subsystem (1) is used to separate a broadband reference signal and a narrowband reference signal from a reference signal; the separated broadband reference signal is used as a reference input of the feedforward broadband active noise control subsystem (2) and the broadband auxiliary filter subsystem (3); and the separated narrowband reference signal is used as a reference input of the feedforward narrowband active noise control subsystem (4) and the narrowband auxiliary filter subsystem (5); The feedforward broadband active noise control subsystem (2) is used to synthesize a broadband secondary sound source related to a broadband reference signal; the broadband auxiliary filtering subsystem (3) is used to separate broadband residual noise and narrowband residual noise from residual noise; the feedforward narrowband active noise control subsystem (4) is used to synthesize a narrowband secondary sound source related to a narrowband reference signal; the narrowband auxiliary filtering subsystem (5) is used to separate narrowband residual noise related to the narrowband reference signal and narrowband residual noise unrelated to the narrowband reference signal from the narrowband residual noise; the feedback active noise control subsystem (6) is used to synthesize a narrowband secondary sound source unrelated to the narrowband reference signal; The broadband residual noise and narrowband residual noise separated by the broadband auxiliary filtering subsystem (3) are used as the error output of the feedforward broadband active noise control subsystem (2) and the desired input of the narrowband auxiliary filtering subsystem (5), respectively; at the same time, the narrowband residual noise related to the narrowband reference signal separated by the narrowband auxiliary filtering subsystem (5) is used as the error output of the feedforward narrowband active noise control subsystem (4); at the same time, the narrowband residual noise independent of the narrowband reference signal separated by the narrowband auxiliary filtering subsystem (5) is used as the error output of the feedback active noise control subsystem (6) and the reference input for synthesizing the feedback active noise control subsystem (6), respectively.

2. A robust feedforward hybrid active noise control system as claimed in claim 1, characterized in that: The linear prediction filter subsystem (1) takes the reference signal x1(n) as input and separates the broadband reference signal x w (n) and the narrowband reference signal x f (n), and are used as reference inputs of the feedforward broadband active noise control subsystem (2) and the feedforward narrowband active noise control subsystem (4), respectively, so as to enhance the independence between the feedforward broadband active noise control subsystem (2) and the feedforward narrowband active noise control subsystem (4); The linear prediction filter subsystem (1) comprises a delay link (11) and a linear prediction filter (12), wherein the delay link (11) and the linear prediction filter (12) are cascaded, and the order of the delay link (11) is D; the linear prediction filter (12) is a finite length impulse response filter model, and its coefficients and length are respectively and L Q , whose coefficients are updated using the least mean square algorithm, namely: w Q (n+1)=w Q (n)+μ Q x w (n)x1(nDj) Among them, μ Q is the update step size of the linear prediction filter (12), which is a positive value; n is the time, n≥0; The narrowband reference signal and the wideband reference signal synthesized by the linear prediction filter subsystem (1) are respectively: x w (n)=x1(n)-x f (n) 3. A robust feedforward hybrid active noise control system as claimed in claim 2, characterized in that: The feedforward broadband active noise control subsystem (2) comprises a feedforward broadband controller (21) and a first FXLMS algorithm module (22); the feedforward broadband controller (21) adopts a linear finite impulse response filter, the coefficients and length of which are respectively and L1; The first FXLMS algorithm module (22) uses the broadband auxiliary filtering subsystem (3) to separate the broadband reference signal x w (n) The broadband residual noise y B (n) is used as an error output and is used to update the coefficients of the feedforward broadband controller (21); the coefficient update formula of the feedforward broadband controller (21) is: Wherein, μ1 is the update step size of the feedforward broadband controller (21), and is a positive value; is the broadband reference signal x w (n) an output after passing through the secondary channel estimation model in the first FXLMS algorithm module (22); a secondary channel estimation model The linear FIR filter is obtained by offline identification, and its coefficients and length are and The feedforward broadband active noise control subsystem (2) synthesizes a broadband secondary sound source related to the broadband reference signal as follows:

4. A robust feedforward hybrid active noise control system as claimed in claim 3, characterized in that: The broadband auxiliary filtering subsystem (3) comprises a broadband auxiliary filter (31) and a first VSS-LMS algorithm module (32); The broadband auxiliary filter (31) adopts a linear FIR filter, whose coefficients and length are respectively and L B , using the first VSS-LMS algorithm module (32) to update the broadband auxiliary filter (31), that is: w B,j (n+1)=w B,j (n)+µ B (n)e B (n)x w (n) Among them, μ B (n) is the update step size of the broadband auxiliary filter (31); e B (n) is the narrowband residual noise separated by the broadband auxiliary filtering subsystem (3); The broadband residual noise related to the broadband reference signal separated by the broadband auxiliary filtering subsystem (3) is: The narrowband residual noise e separated by the broadband auxiliary filtering subsystem (3) B (n) = e(n) - y B (n), used as the desired input of the narrowband auxiliary filtering subsystem (5); The first VSS-LMS algorithm module (32) adaptively adjusts the step size of the broadband auxiliary filter (31) to improve the separation performance of the broadband residual noise and the narrowband residual noise of the broadband auxiliary filtering subsystem (3), and its variable step size strategy is: Among them, ξ B , η B , γ B1 and γ B2 is the user parameter, ξ B ∈(0.9,1),η B Take positive value, γ B1 and γ B2 The values ​​are all integers greater than or equal to zero.

5. A robust feedforward hybrid active noise control system as claimed in claim 4, characterized in that: The feedforward narrowband active noise control subsystem (4) comprises the feedforward narrowband controller (41) and a second FXLMS algorithm module (42); the feedforward narrowband controller (41) adopts a linear FIR filter, the coefficients and length of which are respectively and L2; The second FXLMS algorithm module (42) uses the narrowband auxiliary filtering subsystem (5) to separate the narrowband reference signal x f (n) The relevant narrowband residual noise y N (n) is used as an error output and is used to update the coefficients of a feedforward narrowband controller (41); the coefficient update formula of the feedforward narrowband controller (41) is: Wherein, μ2 is the update step size of the feedforward narrowband controller (41), and is a positive value; is the narrowband reference signal x f (n) output after passing through the secondary channel estimation model in the second FXLMS algorithm module (42); The feedforward narrowband active noise control subsystem (4) synthesizes a narrowband secondary sound source related to the narrowband reference signal as follows:

6. A robust feedforward hybrid active noise control system as claimed in claim 5, characterized in that: The narrowband auxiliary filtering subsystem (5) comprises a narrowband auxiliary filter (51) and a second VSS-LMS algorithm module (52); The narrowband auxiliary filter (51) adopts a linear FIR filter, and its coefficients and length are respectively and L N , using the second VSS-LMS algorithm module (52) to update the narrowband auxiliary filter (51), that is: w N,j (n+1)=w N,j (n)+µ N (n)e N (n)x f (n) Among them, μ N (n) is the update step size of the narrowband auxiliary filter (51); N (n) is the narrowband residual noise separated by the narrowband auxiliary filtering subsystem (5) and is independent of the narrowband reference signal; The narrowband residual noise related to the narrowband reference signal separated by the narrowband auxiliary filtering subsystem (5) is: The narrowband residual noise e which is independent of the narrowband reference signal separated by the narrowband auxiliary filtering subsystem (5) N (n) = e B (n)-y N (n), used as the error output of the feedback active noise control subsystem (6), and also used to synthesize the reference input x2(n) of the feedback active noise control subsystem (6).

7. A robust feedforward hybrid active noise control system as claimed in claim 6, characterized in that: The second VSS-LMS algorithm module (52) adaptively adjusts the step size of the narrowband auxiliary filter (51) to improve the separation performance of the narrowband residual noise related to the narrowband reference signal and the narrowband residual noise unrelated to the narrowband reference signal of the narrowband auxiliary filtering subsystem (5), and the variable step size strategy is: Among them, ξ N , η N , γ N1 and γ N2 is the user parameter, ξ N ∈(0.9,1),η N Take positive value, γ N1 and γ N2 The values ​​are all integers greater than or equal to zero.

8. A robust feedforward hybrid active noise control system as claimed in claim 7, characterized in that: The feedback active noise control subsystem (6) comprises an internal reference signal synthesis module (61), a feedback controller (62) and a third FXLMS algorithm module (63). The internal reference signal synthesis module (61) comprises a secondary channel estimation model and a first-order delay link, and synthesizes an internal reference signal x2(n), namely: in, The output of the feedback controller (62) after being passed through the secondary channel estimation model; The feedback controller (62) uses a linear FIR filter, whose coefficients and length are respectively and L3; The third FXLMS algorithm module (63) uses the narrowband residual noise e that is independent of the narrowband reference signal separated by the narrowband auxiliary filtering subsystem (5) N (n) is used as the error output and is used to update the coefficients of the feedback controller (62), and the updating formula is: Wherein, μ3 is the update step size of the feedback controller (62), and is a positive value; The output of x2(n) provided by the internal reference signal synthesis module (61) after passing through the secondary channel estimation model in the third FXLMS algorithm module (63); The narrow-band secondary sound source obtained by the feedback active noise control subsystem (6) and having nothing to do with the narrow-band reference signal is: It is then used to suppress narrowband noise components in the target noise that are unrelated to the reference signal, so as to improve the noise reduction performance of the system.

9. A robust feedforward hybrid active noise control system as claimed in claim 8, characterized in that: The synthesized secondary sound source is: y(n)=y1(n)+y2(n)+y3(n) Furthermore, y(n) is provided to the secondary speaker and destructively interferes with the target noise in the acoustic space.

10. A robust forward feedback hybrid active noise control method, characterized in that: The method uses a robust forward feedback hybrid active noise control system according to any one of claims 1 to 9, and the method comprises: Step 1: Set user parameters and system initialization; The update step sizes of the linear prediction filter subsystem (1), the feedforward broadband active noise control subsystem (2), the broadband auxiliary filter subsystem (3), the feedforward narrowband active noise control subsystem (4), the narrowband auxiliary filter subsystem (5), and the feedback active noise control subsystem (6) are respectively set, and the length of the linear FIR filter and the initial value are set to zero; the order D of the delay link (11) is set; and the user parameters of the first VSS-LMS algorithm module (32) and the second VSS-LMS algorithm module (52) are set; Step 2: Obtain a reference signal; At time n, a reference signal x1(n) is obtained by using a reference microphone, and after passing through a linear prediction filter subsystem (1), a broadband reference signal and a narrowband reference signal are obtained; the separated broadband reference signal is respectively provided to a feedforward broadband secondary sound source synthesis subsystem (2) and a broadband auxiliary filter subsystem (3); the separated narrowband reference signal is respectively provided to a feedforward narrowband secondary sound source synthesis subsystem (4) and a narrowband auxiliary filter subsystem (5); an internal reference signal synthesis module (61) and the e at time n-1 are used to synthesize the reference signal. N (n-1) and Obtain the internal reference signal x2(n); Step 3: At time n, first, the feedforward broadband controller (21), the feedforward narrowband controller (41) and the feedback controller (62) are used to synthesize the secondary sound source; then, the broadband auxiliary filtering subsystem (3) is used to separate the broadband residual noise and the narrowband residual noise from the residual noise; finally, the narrowband auxiliary filtering subsystem (5) is used to separate the narrowband residual noise related to the narrowband reference signal and the narrowband residual noise unrelated to the narrowband reference signal from the narrowband residual noise; Step 4: Control system update; The linear FIR filter coefficients in the linear prediction filter subsystem (1), the feedforward broadband active noise control subsystem (2), the broadband auxiliary filter subsystem (3), the feedforward narrowband active noise control subsystem (4), the narrowband auxiliary filter subsystem (5), and the feedback active noise control subsystem (6) are updated respectively, and the coefficients of the corresponding linear FIR filters at the time n+1 are obtained respectively; the step sizes in the first VSS-LMS algorithm module (32) and the second VSS-LMS algorithm module (52) are adaptively adjusted respectively, and the corresponding step sizes at the time n+1 are obtained respectively; Step 5: Return to step 2, and repeat steps 2 to 4 until the system gradually converges and reaches a steady state, thereby achieving noise control.

Images