JPH10190590A - Active noise controller for on-line feedback route modeling - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an active noise controller for preventing noise elimination signals generated so as to attenuate noise signals entering from a medium from adversely affecting from a feedback route. SOLUTION: A feed-forward active noise controller 50 performs on-line feedback route modeling and feedback route suppression and is provided with a reference sensor 16, a secondary source 18, an error sensor 20 and an active noise controller 10. The reference sensor 16 receives the noise signals and feedback signals 22 and generates primary signals x(n) the secondary source 18 receives secondary signals y (n) and generates corresponding noise elimination signals and the error sensor 20 receives residual signals and generates error signals e (n). The active noise controller 10 receives the primary signals x(n) and the error signals e (n) generates the secondary signals x (n) and performs the on-line feedback route modeling.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates generally to the field of controllers, and more particularly, to an active noise controller and method for on-line feedback path modeling.

[0002]

2. Description of the Related Art Active noise control devices are intended to reduce any kind of unwanted disturbance or noise signal coming from environmental noise sources, such as electricity, sound, vibration and other noise media. Since the noise source and environment often change over time, the frequency components, amplitude, and speed of the noise signal are generally not stationary. An active noise control device controls noise by providing a "noise-reducing" signal for cancellation from a suitable secondary source to the device environment or medium. An ideal noise-removed signal has the same amplitude as the noise signal and a phase of 180.
Different degrees. Therefore, when the noise removal signal and the noise signal are added at the acoustic addition junction, the two signals cancel or attenuate to reduce noise.

In order to greatly attenuate a noise signal, as described above, the amplitude and phase of the noise signal and the noise removal signal must match well. Generally, this is achieved with an active noise controller using an active noise controller controller. The active noise controller controller performs digital signal processing using one or more adaptive algorithms for adaptive filtering. Adaptive filtering, and more specifically, adaptive algorithms, minimize error signals and constantly track changes in the environment over time, tracking noise signals and any changes in the environment in real time. Adaptive filtering uses various known available adaptive algorithms, such as a least-squares ("LMS") algorithm, to determine the taps or coefficients of the associated adaptive filter, and to model the noise source and the environment to produce an error signal or environment. Reduce or minimize residual signals.

[0004] Compared to passive noise control devices, active noise control devices have the potential advantage of not only greatly attenuating noise, but also reducing size, weight, capacity and cost. Active noise control is an effective method of attenuating noise, which is often difficult and expensive with control using passive means, and is applicable to a wide range of issues in manufacturing, equipment operation, and consumer products.

[0005] The active noise control device is generally divided into a feedforward active noise control device and a feedback active noise control device. The present invention will be described along this line as it shows an example of application to a feedforward active noise control device.

A feedforward active noise controller generally comprises a reference sensor that detects a noise signal from a noise source and generates a corresponding primary signal; an active noise controller controller that generates a secondary signal; A secondary source downstream of the sensor for receiving the secondary signal and generating a noise removal signal for canceling or attenuating the noise signal; and an error sensor for detecting the residual signal and generating a corresponding error signal. The residual signal is the difference between the noise signal and the noise-removed signal applied to the error signal passing through the primary environment. The active noise controller controller receives the primary signal and the error signal and generates a secondary signal accordingly.

[0007] The active noise controller controller is implemented using a digital signal processor and performs digital signal processing using a specific adaptive algorithm for adaptive filtering, which depends on the type of cancellation method applied. Also, the reference sensor, the secondary source, and the error sensor may include an interface circuit that interfaces with the active noise controller controller. The interface circuit may include an analog-to-digital converter, a digital-to-analog converter, an analog filter such as a low-pass filter, and an automatic gain control amplifier.
That is, signals can be exchanged in the digital domain or the analog domain. The interface circuit may be provided separately.

[0008] The feedforward active noise control device includes a primary path with a transfer function P (z). The primary path is defined as the environment from the reference sensor to the error sensor.
Also, the feedforward active noise control device has a secondary path and a feedback path. The secondary path is the transfer function S
(Z). The secondary path is defined as the environment from the output of the active noise controller controller to the output of the error sensor. This includes interface circuits such as digital-to-analog converters, analog filters, power amplifiers, speakers, error microphones, and other equipment. The feedback path has a transfer function F (z). The feedback path is defined as the environment from the output of the active noise controller controller to the output of the reference sensor. An active noise controller controller using a digital signal processor comprises an adaptive filter W (z) that adaptively models the primary path. Adaptive filter W (z)
Is to minimize residual or error signals. The adaptive filtering performed by the adaptive filter W (z) is performed online or offline.

[0009]

Feedforward active noise control systems often have significant drawbacks that impair the performance of the overall system. When the secondary source produces a noise cancellation signal that cancels the noise signal, a portion of the noise cancellation signal propagates back and the reference sensor receives it along with the noise signal. The path through which the noise removal signal travels from the secondary source to the reference sensor is the feedback path. Again, the definition of the feedback path is the media environment from the output of the active noise controller controller to the output of the reference sensor. The portion of the noise removal signal flowing through the feedback path toward the reference sensor is a portion of the feedback signal traveling along the feedback path. If the reference sensor receives the feedback signal, the reference sensor will provide the wrong primary signal to the active noise controller controller, thereby degrading the performance of the overall device. If the feedback signal and the noise signal have the same phase, the reference sensor generates an excessive primary signal. If the feedback signal and the noise signal are out of phase, the reference sensor will still produce the wrong signal. Either way, the feedback signal is undesirable and detracts from the overall performance. Also, the feedback signal may cause a pole in the response of the transfer function of the device, and may become unstable when the gain of the feedback loop increases.

In some applications, unless the effects of the feedback path are modeled and suppressed, the performance of the entire device is significantly degraded. When the secondary source is in close proximity or in close communication with the reference sensor, modeling the feedback path and suppressing the feedback signal is particularly important to the performance of the overall active noise controller. Examples of such devices include refrigerators and window air conditioners with relatively short air ducts. In such an example, the secondary source is necessarily near the reference sensor, so the feedback signal and its adverse effects are large.

The problem of the feedback path has been recognized in the past, and several solutions have been proposed but with limited effectiveness. The first group of proposed solutions focuses on the use, type and location of reference sensors and secondary sources, while the second group of solutions focuses on signal processing. A first group of solutions is to use a directional reference sensor and a secondary source to limit or minimize the feedback signal. These solutions increase the cost and complexity of the device and reduce the overall reliability. On the other hand, obtaining excellent directivity over a wide range of frequencies is difficult, if not impossible.

A second group of proposed solutions focuses on signal processing techniques, but with limited effectiveness. The signal processing methods of this solution are generally divided into offline modeling methods and online modeling methods. Both off-line modeling and on-line modeling are device identification methods, in which a signal is provided to the device and the resulting signal is analyzed to create a model of the unknown device. In this method, an unknown path or environment is stimulated with a known signal, and a signal generated in response thereto is measured and analyzed.

In the off-line feedback path modeling method, a known signal is given in a state where the noise signal cancellation which is usually performed by the active noise control device is stopped. An adaptive algorithm is used to calculate the coefficients or taps of the adaptive filter to minimize the effects of the feedback path. Once the coefficients or taps are determined off-line, the taps or coefficients in the digital filter are fixed during operation of the actual active noise control device and are not changed during actual operation. Offline feedback path modeling may be sufficient,
When used in an apparatus whose parameters change frequently, sufficient performance cannot be obtained by off-line modeling. For example, if the temperature or the speed of the signal flow changes frequently, the feedback path model will be inaccurate due to these changes.

Another problem with off-line feedback path modeling is that noise signals must be removed, or cut off, to correctly model an unknown environment with off-line feedback path modeling. This is often not possible with real devices. For example, it is not easy to shut down a power transformer connected to a power source and supplying power to a consumer for offline modeling. In order to always accurately model the feedback path with a device that changes frequently, it is necessary to constantly model the offline feedback path. Performing off-line modeling when the noise source cannot be blocked involves providing a known or modeled signal for a very long period of time with very large amplitudes. Even so, off-line modeling is inaccurate.

The online feedback path modeling is performed when the noise signal enters an unknown environment and the active noise control device operates to cancel the noise signal.
Modeling the feedback path. Ideally,
When doing online feedback path modeling,
The problem that arises when performing an off-line feedback path modeling in a device environment that changes due to changes in temperature, flow, etc., while arbitrarily changing the device environment to be modeled and using an active noise control device. I want to avoid. However, online feedback path modeling attempted in the past has been unsatisfactory and modeling the feedback path online has failed.

One such method is to provide an adaptive suppression filter in parallel in the feedback path. The adaptive suppression filter method is described in, for example, US Pat.
No. 3,906, "Active Acoustic Attack"
enuator), but only works effectively in the offline feedback path modeling mode. This is because the adaptive suppression filter attempts to adapt even when the noise signal and the noise removal signal are completely cancelled. Feedback suppression models the feedback path to remove all parts of the primary signal that are correlated with the output of the adaptive filter, making the device ideally seemingly free of feedback. Since the primary noise signal has a high correlation with the noise removal signal,
The adaptive feedback suppression filter continues to adapt even if the feedback signal is completely canceled. Therefore, the feedback suppression filter adaptation must be deactivated when the device is online. Further, when the noise signal includes a narrow-band frequency component, the adaptive feedback suppression filter may not converge well when trying to adapt online.

Another on-line feedback path modeling method uses an infinite impulse response ("IIR") filter to compensate the feedback signal. The effectiveness of this method is limited. For example, U.S. Pat.
No. 7,677, "Active Speech Attenuator with Online Adaptive Feedback Cancellation" proposes an adaptive IIR filter structure for use in an active noise controller. This method considers the feedback path as part of the overall device model, but is not a true modeling of the feedback path. This method has some disadvantages specific to adaptive IIR filters. For example, some poles of an IIR filter are not unconditionally stable because they may move out of the unit circle and become unstable during the adaptation process. Also, since there is a local minimum, there is a possibility that by adaptation it will converge to one of this local minimum. Further, the adaptive algorithm used in the IIR filter often has a relatively slow convergence speed as compared with the FIR filter.

Another proposed online feedback path modeling method requires that the amplitude of the modeled signal be very large to distinguish it from the noise signal. This method introduces additional noise into the device, which adversely affects the operation and performance of the overall active noise control device.

[0019]

From the foregoing, it can be seen that there is a need for an active noise control apparatus and method for online feedback path modeling that eliminates or reduces the aforementioned problems. The present invention solves the signal processing problem of the feedback signal by performing an online modeling of the feedback path and suppressing its influence to operate the active noise control device more effectively and accurately. An active noise control device and method for feedback path modeling is provided. This is achieved even when the feedback path changes. The present invention attenuates both broadband and narrowband noise signals.

In one embodiment of the present invention, an active noise controller is provided to generate a noise rejection signal to attenuate the noise signal entering through the primary path medium. The active noise controller performs online feedback path modeling and feedback path suppression. The active noise control device
It comprises a reference sensor, a secondary source, an error sensor, and an active noise controller controller. The reference sensor receives the noise signal and the feedback signal and generates a primary signal. The secondary source receives the secondary signal and generates a corresponding denoised signal;
This is applied to the medium to attenuate the noise signal. The error sensor receives a residual signal that is a combination of the noise signal and the noise removal signal. The error sensor generates an error signal in response to receiving the residual signal. The active noise controller controller receives the primary signal and the error signal to generate a secondary signal,
On the other hand, online feedback path modeling is performed.

The present invention has various technical advantages. One technical advantage of the present invention is that it accurately performs online feedback path modeling to improve the overall performance of the active noise control system. Another technical advantage of the present invention is that the present invention can be implemented using existing digital signal processing methods and algorithms. Yet another technical advantage of the present invention is that it improves the stability of an active noise controller by eliminating the effects of the feedback path. Yet another advantage of the present invention is that both wideband and narrowband noise signals can be canceled or attenuated. Other advantages will be apparent to one skilled in the art from the following figures, descriptions, and claims.

[0022]

BEST MODE FOR CARRYING OUT THE INVENTION

FIG. 1 is a block diagram of a feedforward active noise control device 50. The feedforward active noise control device 50 includes a noise source 14 and a reference sensor 16.
The active noise controller controller 10 and the secondary source 1
8 and an error sensor 20. The noise source 14 generates a noise signal or provides it through the device environment,
6 receives this signal. In FIG. 1, the noise signal flows out of the noise source.

The reference sensor 16 receives a corresponding electrical signal x
(N) is generated. This is called a primary signal x (n). Reference sensor 16 may be implemented using any type of sensor, such as a microphone, tachometer, accelerometer, and the like. The reference sensor 16 includes an interface circuit 24,
The noise signal may be received as an analog signal and the corresponding primary signal x (n) may be output as a digital signal. Interface circuit 24 includes any of a variety of devices, such as analog-to-digital converters, analog filters, amplifiers controlled by automatic gain control circuits, and any other circuits such as antialiasing circuits. Good.

The active noise controller controller 10 comprises:
Upon receiving the primary signal x (n), it generates a corresponding electric signal y (n). This is called a secondary signal y (n). Secondary signal y
When (n) is given to the secondary source 18, the secondary source 18 receives it and returns it to the device environment as an analog signal. The output signal of the secondary source 18 is called a noise removal signal, and is designed to reduce, cancel, or suppress the noise signal from the noise source 14. Secondary source 18 may be implemented using any signal source, such as a speaker, shaker, or other available signal source. The secondary source 18 includes an interface circuit 26. The circuit 26 converts the secondary signal y (n) from the digital domain to the analog domain and gives it with a desired amplitude.
For example, the interface circuit 26 includes a digital-to-analog converter, an analog filter such as a low-pass filter, and an amplifier controlled by an automatic gain control circuit.
Including any type of circuit.

When the noise cancellation signal is introduced into the equipment environment, a portion of the noise cancellation signal is also transmitted to the reference sensor 16 through a feedback path defining the path from the output of the active noise controller controller 10 to the output of the reference sensor 16. Return. As shown, the feedback signal 22 flows through the feedback path, one of its components being part of the noise cancellation signal flowing through the feedback path.
This is called a noise removal feedback component. Feedback signal 22 also includes a modified modeling feedback component that is part of the present invention. Feedback signal 22
Are generated as a result of the modeled signal. This is part of the secondary signal y (n), which will flow through the feedback path, as will be explained in more detail below. Thus, the feedback signal 2
2 includes a denoising feedback component and a modified modeling feedback component. The reference sensor 16 receives the feedback signal 22 along with the noise signal, and generates a primary signal x (n) as a result. Therefore, the primary signal x (n)
Includes a noise signal component and a feedback signal component, and the feedback signal component includes a noise removal feedback component and a modified modeling feedback component. Reference sensor 16
, An incorrect primary signal x (n) is generated. This will be described later in detail.

The error sensor 20 receives a residual signal that is the result of adding the noise signal and the noise removal signal at the acoustic addition junction. Ideally, this residual signal is zero. When a noise removal signal having the same amplitude as the noise signal and having a phase difference of 180 degrees is provided to the acoustic addition junction to completely cancel the noise signal, the residual signal becomes zero.

The error sensor 20 receives the residual signal and generates a corresponding error signal e (n). The error sensor 20 may be realized using an arbitrary sensor. For example, like the reference sensor 16, the error sensor 20 may be implemented using a microphone, tachometer, accelerometer, or any other available sensor. The error signal e (n) may be provided in the digital domain using the interface circuit 28. The interface circuit 28 may be the same as the interface circuit 24, and may include circuits such as an analog-to-digital converter, a smoothing filter, and an amplifier controlled by an automatic gain control circuit. When the error signal e (n) is given to the active noise controller controller 10, the adaptive active noise controller filter 66 performs active noise control using this signal and generates the secondary signal y (n). Adjust to improve the overall performance of feedforward active noise controller 50. The adaptive active noise controller filter 66 is the main filter of the active noise controller controller 10 and is shown in FIG. Details will be described later. The active noise controller controller 10 also performs online feedback path modeling and feedback path suppression to reduce the effects of the feedback signal 22.

Although the interface circuit 24, interface circuit 26, and interface circuit 28 are shown in FIG. 1 as part of each sensor or source, the interface circuits may be provided independently, ie, separately, as discrete circuit components. . The present invention is not limited to a particular type of interface circuit.

The active noise controller controller 10, shown in detail in FIGS. 2 and 3, comprises a primary signal x (n) and an error signal e.
In response to (n), a secondary signal y (n) is generated. The active noise controller controller 10 includes an online feedback path modeling circuit and a feedback signal suppression circuit. The feedback path may be modeled by a transfer function F (z). Active noise control device controller 10
Comprises an adaptive active noise control device filter 66. It serves as an averaging filter and adaptively models the primary device or environment with the transfer function P (z).

The active noise controller controller 10
Comprises a modeled signal generator 64. When this is used to introduce the modeled signal into the feedforward active noise controller, the modeled signal passes through the feedback path and produces a feedback excitation signal, or a modified modeled signal. The modified modeling signal follows the feedback path and is therefore correlated with the feedback path. The modified modeling signal is provided as a modified modeling feedback component of the feedback signal 22 along with the noise removal feedback component of the feedback signal 22. Typically, the modeled signal is provided with a much smaller amplitude than the primary signal x (n) and the secondary signal y (n). The modified modeling signal is used together with the online feedback path modeling circuit and the feedback signal suppression circuit to perform online modeling and feedback signal suppression. The modeling signal and the modified modeling signal acting as a modified modeling feedback component of the feedback signal 22 will be described in detail later with reference to FIGS. The active noise controller controller 10 controls the feedforward active noise controller 5 by reducing or minimizing the error signal e (n).
0 and, on the other hand, perform on-line feedback path modeling of the feedback path to improve overall device performance and noise cancellation.

The active noise controller controller 10 comprises:
It may be realized using a digital circuit such as a digital signal processor. For example, Texas Instruments offers a family of digital signal processors, including the TMS320C25 and TMS320C30 digital signal processors. The advent of high-speed digital signal processors and associated hardware has made the invention easier to implement. Many digital signal processors use a fixed point data format. in this case,
Interface circuit 24 and interface circuit 28
In order to extend the dynamic range of the analog-to-digital converter, an automatic gain control circuit must be used for each data input.

FIG. 2 is a block diagram of the active noise controller controller 10. The active noise controller controller 10 receives the primary signal x (n) from the reference sensor 16 and receives the error signal e (n) from the error sensor 20 to perform various filtering, processing and modeling functions, and y (n) is generated and provided to the secondary source 18. The primary signal x (n) enters summing junction 52 with the output signal of feedback suppression filter 70. At the addition junction 52, the output signal of the feedback suppression filter 70 is subtracted from the primary signal x (n) to generate an output signal x '(n). Feedback signal 2, which is one component of the primary signal x (n)
The signal x ′ (n) may be referred to as a feedback-suppressed primary signal because the two noise-removal feedback components are removed by the feedback suppression filter 70.

The signal identification circuit 54 receives the feedback suppression primary signal x '(n) and generates an output signal v' (n). The noise signal component is a feedback suppressed primary signal x '
Since it is removed from (n), the signal v ′ (n) may be called a modified modeling signal. Modified modeling signal v '(n)
Is the modeled signal v after going through the feedback path
(N). Again, the definition of the feedback path is the device environment from the output of the active noise controller controller 10 to the output of the reference sensor 16. The signal identification circuit 54 is, in effect, a feedback suppression primary signal x ′.
Feedback signal 22 contained as one component of (n)
Extract the modified modeling feedback component of. This is achieved even though the amplitude of the modeled signal v (n) is generally much smaller than the amplitude of the noise signal. The signal identification circuit 54 uses a decorrelation delay device and a digital adaptive filter to provide the feedback signal 22
A prediction noise signal u (n) that does not include any of the components is generated. Next, the predicted noise signal u (n) is subtracted from the feedback suppressed primary signal x ′ (n) to obtain the modified modeled signal v ′.
(N) is generated. FIG. 3 shows the signal identification circuit 54.
And will be described later in detail.

Next, the feedback suppression primary signal x '
(N) to the summing junction 56 together with the modified modeling signal v ′ (n). At the summing junction 56, the modified modeling signal v 'is converted from the feedback suppression primary signal x' (n).
(N) minus the output signal r (n) called the processed primary signal
Generate The processed primary signal r (n) includes a noise signal component of the primary signal x (n) after removing the modified modeling feedback component of the feedback signal 22 from the feedback-suppressed primary signal x '(n). Next, the processed primary signal r (n) is supplied to the feedforward active noise control device 5.
0 to the main adaptive filter. This primary adaptive filter is
It includes an adaptive active noise controller filter 66 and an associated adaptive algorithm 72.

The adaptive active noise controller filter 66 and the adaptive algorithm 72 work together to generate an output signal s (n), called the generated secondary signal. The adaptive active noise controller filter 66 receives the processed primary signal r (n), and the adaptive algorithm 72 receives the processed primary signal r (n) and the error signal e (n). The adaptive algorithm 72 generates the coefficients or taps used by the adaptive active noise controller filter 66 to generate the appropriate value output signal s (n) to cancel the noise signal. The adaptive algorithm 72 calculates the error signal e
Generate taps or coefficients that minimize the value of (n).

Adaptive Active Noise Controller Filter 66
Can be any type of digital adaptive filter, such as FI
An R filter or Transversal filter, an IIR filter, a lattice filter, a subband filter, or any other filter capable of performing adaptive filtering may be implemented. Preferably,
The adaptive active noise controller filter 66 is implemented with an FIR filter to enhance stability and performance. The adaptation algorithm used for the adaptation algorithm 72 may be any known or available adaptation algorithm, such as a least mean square (LMS) algorithm, a normalized LMS algorithm, a correlated LMS algorithm, or a leaky LM.
S algorithm, partially updated LMS algorithm, variable step size LMS algorithm, signed LM
It may include an S algorithm, a complex LMS algorithm, and the like. The adaptive algorithm 72 uses a recursive or non-recursive algorithm, depending on how the adaptive active noise controller filter 66 is implemented. For example, if the adaptive active noise control device filter 66 is implemented by an IIR filter, a recursive LMS algorithm may be used for the adaptive algorithm 72. For an excellent overview of the main adaptive algorithms, see Sen M. Kue and Denni
sR. Morgan's "Adaptive Noise Control System: Algorithms and DSP Devices (Active Noise Control Systems: A
lgorithms and DSP Implementations), 1996.

Further, a modeled signal generator 64 is provided to provide a modeled signal v which is a white noise signal, ie, a random signal.
(N) is generated. Modeling signal generator 64 may generate the white noise signal, the random signal, or the chirp signal, using any method, but generally two basic signals used to generate the random or chirp signal. Use one of the methods. The first method is a look-up table method using a set of stored samples, and the second method uses a signal generation algorithm. Both generate sequences that repeat after a finite time, so they are not always completely random. The modeled signal v (n) is
It can be any signal that can model the environment or path.

The generated secondary signal s (n) and the modeled signal v
(N) is applied to the summing junction 68. Modeled signal v
The amplitude of (n) is generally much smaller than the amplitude of the noise signal or the noise rejection signal to reduce the effect on the feedforward active noise control device 50. These two signals are summed at summing junction 68 to produce a secondary signal y (n). Therefore, the secondary signal y (n) includes two components: (1) a modeled signal v (n) and (2) a generated secondary signal s (n).

An online feedback path modeling adaptive filter 60 and a corresponding adaptive algorithm 62 are also provided as part of the active noise controller controller 10. A feedback path is modeled using an online feedback path modeling adaptive filter 60 and an adaptive algorithm 62, and a filter coefficient, that is, a tap setting is periodically given to a feedback suppression filter 70. Again, the definition of the feedback path is based on the output of the active noise controller controller 10 from the reference sensor 16.
This is the device environment up to the output. The online feedback path modeling adaptive filter 60 gives the filter tap setting to the feedback suppression filter 70 for each fixed multiple of the sampling period. The fixed multiple of the sampling period is programmable, which is per sampling period, or preferably, a fixed multiple of the sampling period, to provide acceptable performance of the entire device. For example,
The fixed multiple of the sampling period is 20 sampling periods. This generally depends on how often the feedback path changes. The sampling period is the reciprocal of the sampling rate, and the sampling rate must be high enough to meet the Nyquist criterion that it must be at least twice the highest frequency of interest. Also, the real-time digital signal processing performed by the active noise controller controller 10 including the online feedback path modeling adaptive filter 60 must be performed at a sampling period smaller than the sampling period of the feedforward active noise controller 50.

The online feedback path modeling adaptive filter 60 and the adaptive algorithm 62 receive as input a modeled signal v (n). Adaptive algorithm 62 also receives the output signal of summing junction 58 as an input signal. This signal is combined with the modified modeling signal v '(n) and the online feedback path modeling adaptive filter 60.
Is the difference from the output signal. The function of the adaptive algorithm 62 is to adjust the taps or coefficients of the online feedback path modeling adaptive filter 60 to minimize the mean square value of the output signal from the summing junction 58. The output signal of summing junction 58 may be considered an error signal to be minimized, for example, a modeled error signal. Thus, the filter coefficients or taps are updated so that the error signal progressively becomes smaller at each sampling. The online feedback path modeling adaptive filter 60 and the adaptive algorithm 62 may be implemented as any digital adaptive filter as described above with respect to the adaptive active noise controller filter 66 and the adaptive algorithm 72. Online feedback path modeling adaptive filter 60 and adaptive algorithm 62 provide an online feedback path model.

The feedback suppression filter 70 is a non-adaptive digital filter, and receives taps or coefficients from the online feedback path modeling adaptive filter 60. As mentioned above, these coefficients are determined at each sampling period or, preferably, at selected intervals.
Online feedback path modeling adaptive filter 6
Copy from 0 to the feedback suppression filter 70. Feedback suppression filter 70 receives the tap or coefficient information and, in response, processes its input signal, the generated secondary signal s (n). Feedback suppression filter 7
0 filters this signal to produce an output signal approximately equal to the noise-removed feedback component of feedback signal 22 entering through the feedback path. Next, the output signal of the feedback suppression filter 70 is provided to the addition junction 52, and the noise removal feedback component of the feedback signal 22 is removed from the primary signal x (n).

In operation, the active noise controller controller 10 controls the error signal e from the error sensor 20
(N) and the primary signal x (n) from the reference sensor 16 are received as inputs. The primary signal x (n) may be considered to include a noise signal component and a feedback signal 22 component. Again, the feedback signal 22 includes at least two components, a denoising feedback component and a modified modeling feedback component. The primary signal x (n) passes through a summing junction 52 where a feedback suppression filter 70
To remove the noise-removed feedback component of the feedback signal 22 to obtain the feedback-suppressed primary signal x ′.
(N) is generated. Feedback suppression primary signal x '
(N) enters the signal discrimination circuit 54 and the summing junction 56.

The signal identification circuit 54 outputs the modified modeled signal v '
(N), and this modified modeling signal v ′ (n) also enters the summing junction 56. At the summing junction 56, the modified modeling signal v 'is converted from the feedback suppression primary signal x' (n).
Subtract (n) to remove the modified modeling feedback component of x '(n) and generate a processed primary signal r (n).
The processed primary signal r (n) enters an adaptive active noise controller filter 66 and an adaptive algorithm 72. Further, the adaptive algorithm 72 calculates the error signal e from the error sensor 20.
(N). Adaptive active noise control device filter 6
6 generates a secondary signal s (n) using an adaptive algorithm 72 that adjusts the coefficients or taps of the adaptive active noise controller filter 66 to minimize the error signal e (n). Ideally, after the generated secondary signal s (n) is substantially equal to a signal 180 degrees different in phase from the noise signal, and the generated secondary signal s (n) is converted to the analog domain by the secondary source 18, When added to the noise signal, the noise signal is canceled.

Another alternative is to provide no feedback junction primary signal x '(n) directly to the adaptive active noise controller filter 66 and the adaptive algorithm 72 without the addition junction 56. In this case, the feedback suppression primary signal x '
(N) functions as the processing primary signal r (n), but the processing primary signal r (n) includes the modified modeled signal v ′ (n) as a component. The reason for this is that the average amplitude of the modified modeled signal v '(n) is generally much smaller than the noise signal component of the primary signal x (n). The noise signal component is also included as one component of the feedback suppression primary signal x '(n).

On the other hand, the modeling signal generator 64 supplies the modeling signal v (n) to the summing junction 68, the online feedback path modeling adaptive filter 60, and the adaptive algorithm 62. The modeled signal v (n) and the generated secondary signal s (n) are added at an addition junction 68 to obtain a secondary signal y
(N) is generated. Next, the secondary signal y (n) is
8 to generate a corresponding noise removal signal that cancels the noise signal. Preferably, the amplitude of the modeled signal v (n) is slightly smaller than the amplitude of the noise signal. This is so that the modeling signal stimulates the feedback path without significantly affecting the overall device environment.

The online feedback path modeling adaptive filter 60 and the adaptive algorithm 62 receive the modeling signal v (n) and model the feedback path. To do this, the adaptive algorithm 62 calculates the appropriate taps or coefficients of the feedback suppression filter 70 and, at selected intervals, the feedback suppression filter 70.
Give to. As mentioned earlier, these are provided at each sampling period or at selected intervals. Next, the output of feedback suppression filter 70 enters summing junction 52, where the noise removal feedback component of feedback signal 22 is removed from primary signal x (n).

In this way, the active noise controller controller 10 controls the feedforward active noise controller 50, and the secondary source 18 generates a noise cancellation signal to cancel or attenuate the noise signal. The active noise controller controller 10 includes online feedback path modeling and suppression circuitry to improve the overall performance of the feedforward active noise controller 50, except for any adverse effects caused by the presence of the feedback path. With the active noise control device controller 10,
Both narrowband and wideband noise signals can be canceled.

FIG. 3 is a block diagram of the signal identification circuit 54.
De-correlation delay device 102 and adaptive identification filter 104
, An adaptive algorithm 106, and an addition junction 100. The decorrelation delay unit 102 and the summing junction 100 are connected to the feedback suppressing primary signal x ′ from the summing junction 52.
(N). The decorrelation delay unit 102 is a digital delay, and is a feedback suppression primary signal x '(n).
Is delayed by a selected multiple of the sampling period. Preferably, decorrelation delay device 102 provides a delay that is greater than or equal to the delay provided through the feedback path. The delay provided through the feedback path is, for example, the time required for the feedback signal 22 to travel from the output of the active noise controller controller 10 to the output of the reference sensor 16. Preferably, the decorrelation delay device 102
Is set to be longer than the delay of the feedback path, but the performance is high when the delay time is one sampling period. Thus, the present invention includes a delay of one sampling period or more.

The adaptive identification filter 104 and the adaptive algorithm 106 both receive the output signal of the decorrelation delay unit 102. The adaptive algorithm 106 also receives the modified modeled signal v ′ (n) as an input signal and uses it as an error signal. The adaptive algorithm 106 calculates taps or coefficients of the adaptive discrimination filter 104 so that the modified modeled signal v ′ (n) is minimized. In response,
The adaptive discrimination filter 104 receives the output of the decorrelation delay unit 102 and generates a predicted noise signal u (n) that is ideally equal to the actual noise signal. In this way, the modified modeling feedback component is removed, and the predicted noise signal u
(N) is applied to the summing junction 100 and subtracted from the feedback-suppressed primary signal x '(n) to remove the noise signal component of the feedback-suppressed primary signal x' (n), thereby obtaining a modified modeled signal v '(n). Generate

The adaptation algorithm 106 may be implemented using any type of known available adaptation algorithm, as described above with respect to the adaptation algorithms 72 and 62. The adaptive identification filter 104
It may be any type of digital filter, such as an FIR or IIR filter. The decorrelation delay device 102 may be implemented using computer memory or registers,
A desired delay is applied to the feedback suppression primary signal x '(n) to de-correlate the modified modeling feedback component of the feedback suppression primary signal x' (n), while continuing to correlate the narrowband components. As a result of the delay, the adaptive discriminating filter 104 can only predict or generate subsequently correlated signal components.

Thus, the present invention provides an active noise control apparatus and method for on-line feedback path modeling that eliminates or reduces the adverse effects of the feedback path on the operation of the overall device and has the advantages described above. give. Although the preferred embodiment has been described in detail, various modifications, substitutions, and alternatives can be made without departing from the scope of the invention. The invention can also be used to reduce any noise source, such as, but not limited to, vibration, acoustic signals, and electrical signals. In the preferred embodiment, the circuits and functional blocks have been shown as discrete or separate circuits or functional blocks, but they may be combined together or divided into separate circuits or functional blocks without departing from the scope of the invention. be able to.

Further, those skilled in the art will be able to modify the direct connection shown here so that the two circuits or functional blocks are not directly connected but are coupled to each other through intermediate circuits or functional blocks, and that the desired circuit shown in the present invention is not connected. Can be obtained. Also, those skilled in the art can modify certain signals shown here,
The signal can simply be processed or added with other signals in an intermediate stage while still achieving the desired results shown by the present invention. For example, the feedback suppressed primary signal may be provided to the adaptive active noise controller filter 66 with or without the modified modeled signal v '(n). Those skilled in the art can devise alterations, substitutions and other alternatives without departing from the spirit and scope of the invention as defined in the appended claims.

With respect to the above description, the following items are further disclosed. (1) An active noise control device for generating a noise removal signal, attenuating a noise signal entering through a medium, performing on-line feedback path modeling and feedback path suppression, and receiving the noise signal and the feedback signal, A reference sensor for generating a primary signal in response thereto;
Receiving a secondary signal, generating a corresponding noise-removed signal, applying the noise-reduced signal to the medium, and attenuating the noise signal. Receiving the residual signal is the error sensor, and generates an error signal in response to the error signal, an error sensor and the primary signal and the error signal to generate the secondary signal,
On the other hand, do online feedback path modeling,
An active noise control device comprising: an active noise control device controller.

(2) The active noise controller controller subtracts a noise removal feedback component from the primary signal to generate a feedback suppression primary signal, a first summing junction, the feedback suppression primary signal and the error. Receiving a signal, filtering the feedback suppressed primary signal to generate a generated secondary signal, a device adaptive filter, generating a modeled signal, a modeled signal generator, the generated secondary signal and the modeling 2. The active noise control device of claim 1, further comprising a second summing junction for adding signals to generate said secondary signal.

(3) The active noise controller controller receives the modeled signal and the modeled error signal,
Filtering the modeled signal to generate an output signal, an online feedback path modeling adaptive filter, receiving the feedback suppression primary signal to generate a modified modeled signal, a signal identification circuit, and from the modified modeled signal. A third summing junction for subtracting the output signal and providing the modeled error signal to be applied to an adaptive algorithm used by the online feedback path modeling adaptive filter; A feedback suppression filter for generating a signal including a component and applying the signal to the first summing junction, wherein the adaptive algorithm used by the online feedback path modeling adaptive filter is a filter tap of the online feedback path modeling adaptive filter. Total Minimizing the mean square value of the modeled error signal and providing the filter tap to the feedback suppression filter, which uses the feedback suppression filter to generate a signal including the noise removal feedback component, An active noise control device according to claim 1.

(4) a fourth summing junction for subtracting the modified modeling signal from the feedback suppression primary signal to generate a processing primary signal; and wherein the apparatus adaptive filter includes the processing primary signal and the processing primary signal. Receiving an error signal,
4. The active noise control device according to claim 3, wherein the processing primary signal is filtered to generate the generated secondary signal. (5) The active noise control device according to (1), further comprising: a feedback suppression filter that receives the generated secondary signal, generates a signal including the noise removal feedback component, and applies the generated signal to the first summing junction. controller.

(6) The adaptive algorithm used by the online feedback path modeling adaptive filter calculates filter taps of the online feedback path modeling adaptive filter to minimize the mean square value of the modeling error signal. 6. The active noise control device controller of claim 5, wherein: (7) The active noise control device controller according to (6), wherein the filter tap is provided to the feedback suppression filter, and the feedback suppression filter uses the feedback tap to generate a signal including the noise removal feedback component. (8) The active noise controller controller according to (7), wherein the filter tap is provided to the feedback suppression filter at a desired interval.

(9) The signal identifying circuit receives the delayed feedback suppression primary signal and the modified modeling signal, the decorrelation delay device delaying the feedback suppression primary signal to provide a delayed feedback suppression primary signal, An adaptive identification filter for filtering the delayed feedback suppressed primary signal to generate a prediction noise signal; and a fifth summing junction for subtracting the prediction noise signal from the feedback suppression primary signal to generate the modified modeled signal. 5. The active noise controller controller of claim 4, including:

10. The active noise controller controller of claim 9, wherein the delay of the decorrelation delay is a programmable delay. (11) The active noise controller controller according to (9), wherein the delay is equal to or longer than a delay of the feedback path to be modeled.

(12) The active noise control device controller according to item 11, wherein the feedback path is defined as an environment from an output of the active noise control device controller to an output of a reference sensor. (13) The active noise control device according to (1), wherein the active noise control device is a feedforward active noise control device.

(14) An on-line feedback path modeling adaptive filter for receiving the modeled signal and the modeling error signal and filtering the modeled signal to generate an output signal; and receiving and correcting the feedback suppression primary signal. Generating a modeled signal, a signal identification circuit, and subtracting the output signal from the modified modeled signal;
A third summing junction for generating the modeling error signal, which is given to an adaptive algorithm used by the online feedback path modeling adaptive filter, and the generated secondary signal, and generating a signal including the noise removal feedback component. A feedback suppression filter to be applied to the first summing junction, wherein the adaptive algorithm used by the online feedback path modeling adaptive filter calculates a filter tap of the online feedback path modeling adaptive filter to perform the modeling. The mean square value of the error signal is minimized, and the filter tap is provided to the feedback suppression filter, the feedback suppression filter uses the feedback suppression filter to generate a signal including the noise removal feedback component, and the signal identification circuit includes: The said Delaying the feedback suppression primary signal to provide a delayed feedback suppression primary signal, a decorrelation delay device, receiving the delayed feedback suppression primary signal and the modified modeling signal, filtering the delay feedback suppression primary signal to generate a prediction noise signal. 2. The active noise controller controller of claim 1 including: an adaptive discriminating filter; and a fifth summing junction for subtracting the predicted noise signal from the feedback suppressed primary signal to generate the modified modeled signal. .

(15) Provide a feedforward active noise control device 50 that generates a noise removal signal and attenuates a noise signal entering through a medium. The feedforward active noise controller 50 performs on-line feedback path modeling and feedback path suppression, and also includes a reference sensor 16, a secondary source 18, an error sensor 20,
An active noise control device controller 10 is provided. The reference sensor 16 receives the noise signal and the feedback signal 22 and generates a primary signal x (n). The secondary source 18 produces a secondary signal y
In response to (n), a corresponding noise removal signal is generated. The error sensor 20 receives the residual signal and generates an error signal e (n). The active noise controller controller 10 receives the primary signal x (n) and the error signal e (n) and receives the secondary signal y (n).
And perform online feedback path modeling.

[Brief description of the drawings]

For a better understanding of the present invention, refer to the accompanying drawings with a detailed description. In the drawings, like reference numbers indicate like parts.

FIG. 1 is a block diagram showing a feedforward active noise control device according to the present invention.

FIG. 2 is a block diagram showing an active noise control device controller of the feedforward active noise control device.

FIG. 3 is a block diagram showing a signal identification circuit of the active noise control device controller of the present invention.

[Explanation of symbols]

 Reference Signs List 10 Active noise controller controller 14 Noise source 16 Reference sensor 18 Secondary source 20 Error sensor 50 Feedforward active noise controller

Claims (1)

[Claims] 1. An active noise control device for generating a noise removal signal, attenuating a noise signal entering through a medium, performing on-line feedback path modeling and feedback path suppression, wherein the noise signal and the feedback signal are A reference sensor that receives and generates a primary signal in response thereto; and a secondary source that receives the secondary signal to generate a corresponding noise-removed signal and provides it to the medium to attenuate the noise signal. An error sensor, which receives a residual signal that is a sum of the noise signal and the noise removal signal to the error sensor and generates an error signal in response thereto, an error sensor, and the primary signal and the error signal. Receiving the secondary signal and performing on-line feedback path modeling on the other hand.