JPH07210178A - Active noise control device - Google Patents

Abstract

(57) [Abstract] [Purpose] The objective is to efficiently and actively mute random noise that has a spectral structure distributed in a wide band and has a dominant low frequency range. A reference signal after passing through a low-pass filter 41 is divided into two systems, one of which is a low-frequency signal control system and the other is a high-frequency signal control system for parallel control. A signal is input to the high frequency signal control system via high pass filters 410 and 411.
For the adaptation of each control system, the convergence coefficient is set so that the low-frequency signal control system is initially faster, and the high-frequency noise is adapted after removing the low-frequency noise. [Effect] Compared with the case where broadband noise is collectively processed by a single adaptive filter, low-frequency noise and high-frequency noise can be controlled more efficiently in a well-balanced manner.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention is applied to random noise having a spectral structure which is distributed over a wide band such as vehicle interior noise generated when a vehicle is running and whose absolute sound pressure level is superior at lower frequencies. The present invention relates to a device that actively silences sound.

[0002]

2. Description of the Related Art Noise generated when a vehicle is driven is
There are various engine sounds, wind noises, road noises, etc.,
Of these, the noise generated by the vibration of tires and suspensions being propagated in the vehicle interior due to the unevenness of the road surface during traveling is generally called road noise, and usually has a broadband spectrum of 30 to 300 Hz. Road noise
Since the noise occurs on a rough road surface at a speed of about 40 to 60 km / h and is an unpleasant sound for humans, various efforts have been made to reduce it.

Incidentally, as a countermeasure against these noises, a so-called "passive" method such as a change in the structural design of the vehicle body or a countermeasure using a sound insulating material has been generally used. On the other hand, active noise control technology has been attracting attention that artificially creates a secondary sound having an opposite phase to the generated noise to "actively" eliminate the sound. In particular, as an example of research on a system that uses a secondary sound source to actively mute road noise, see, for example, the document “Active Noise and V
ibration Control within the Automobile ", AM Mcdon
ald, et al, International Symposium on Active Cont
rol of Sound and Vibration, ASJ Proc. '91, Tokyo,
April 9-11, 1991, pp. 147-156.
Among them, AM Mcdonald and colleagues used the vibration of the rear suspension as a reference signal using the measurement output of two accelerometers mounted close to the wheel hub as an active component consisting of two microphones and two speakers. We have reported an example where the noise reduction system can significantly reduce the noise level near 100 Hz at the position of the rear seat.

The basic idea of this active noise control is old, and since the pioneering research conducted by Lueg in the 1930s, research has been conducted by Olson, Conver, etc. in the 1950s. It has been relatively recently applied to products. This is largely due to the advent of high-speed arithmetic elements for enabling control such as digital signal processors, but it can also be mentioned that the theoretical aspects of control algorithms have been improved.

Examples of recent noteworthy research on active noise control technology include those by GBB Chaplin (for example, Japanese Patent Publication No. 56-501062) and PANelson / SJEllio.
by t (for example, published Japanese Patent Publication No. 1-501344)
There are two examples. The difference between the two control methods is that the former control uses "repetitive control" that is based on the target noise periodicity, while the latter control uses the LMS algorithm, which is a type of steepest descent method. The target noise does not necessarily need to be periodic because the adaptive signal processing is performed.

This LMS adaptive control algorithm is 195
Although it is a method systematized by Widrow in the 0's, most of recent studies on active noise control rely on this control algorithm because of its versatility. Also in the present invention, since it is basically premised on the use of this control algorithm, the above-mentioned Japanese Patent Publication No. 1-501344
(PANelson / SJElliot) is taken as an example to explain the conventional technology.

FIG. 7 shows an active noise control device described in the above-mentioned published patent, which silences a specific closed space such as a passenger compartment of an automobile by a plurality of loudspeakers and microphones. This is to output a secondary sound for noise reduction by interfering with primary sound (noise) and secondary sound at each microphone position and three microphones for measuring sound pressure at a predetermined position in a closed space. Two loudspeakers,
A reference signal generator that generates a signal synchronized with the rotation of the engine, and a control circuit that has a pair of adaptive filters for driving the loudspeaker by outputting a signal that drives the speaker by phase-amplitude modulating the reference signal Has been done. Further, an engine rotation signal (for example, an ignition timing signal, a crank angle sensor signal, etc.) is input to the reference signal generator, and the reference signal generator outputs a sine wave proportional to an integral multiple of the momentary engine rotation cycle. Generating a signal.

In the LMS adaptive control algorithm, the coefficient of the adaptive filter is updated every moment so that the square value of the sound pressure at each microphone position is minimized. However, the primary sound and the secondary sound interfere well with each other. To reduce noise,
The standard signal or the reference signal which is the source of the standard signal must include a component having a sufficiently high correlation with the primary sound. Usually, a dimensionless quantity that takes a value between 0 and 1 called coherence is defined as an index indicating the degree of correlation between two signals. From the results of rigorous theoretical analysis,
It is known that the noise reduction amount by the active noise control system based on the LMS adaptive control algorithm is determined by the value of this coherence.

In the active noise control system in the passenger compartment of an automobile as shown in FIG. 7, the noise caused by the rotational vibration of the engine is the object of control, and the engine rotation signal is supplied as a reference signal to synchronize with it. By generating the sine wave signal, the reference signal having high coherence with respect to the engine noise component is obtained. For example, when the engine is a four-cycle four-cylinder engine, vibration with a frequency twice the engine rotation is large, and is generally called engine secondary vibration. This is due to gas torque fluctuations due to gas combustion every 1/2 rotation and inertia torque fluctuations due to imbalance of the crankshaft system moment.
Then, this becomes an exciting torque to vibrate the vehicle body, and this is propagated into the vehicle interior to generate standing wave noise. Therefore, at this time, the active noise control of the secondary vibration noise becomes possible by supplying the rotation signal for each crank angle of 180 degrees as the reference signal.

[0010]

By the way, as shown in FIG. 6, road noise has a characteristic of 30 to 300H.
Absolute sound pressure level (linear scale without auditory correction), although it is a broadband spectrum distributed in z
When compared with, it has the characteristic that the low frequency sound is excellent. And, it has a tendency to decrease to the right with a peak around 30 Hz, and approximately 50 to 60 between 30 and 300 Hz.
It has a dynamic range of dB. In an active noise control system based on the LMS adaptive control algorithm, normally F
An IR type adaptive digital filter is used. If one secondary sound output means and one secondary sound control signal are used to mute road noise having a wide dynamic range as described above, it is necessary to increase the number of taps of the FIR filter from the viewpoint of control resolution. However, if the number of taps is too large, it is not a good idea because it causes a decrease in adaptation speed and a large increase in the amount of calculation.

Regarding the difference between the low frequency component and the high frequency component of the road noise, the following can be mentioned in addition to the sound pressure level described above. First, low frequency (mainly below 100Hz)
Since the wavelength is long, the vibration damping time is relatively long even if the vibration damping ratio is the same. Further, since the basic characteristics of the vehicle (rigidity, mass, damping, spatial layout, etc.) define the approximate characteristics, the vibration and noise transmission system has a relatively small change with time and a time-invariant characteristic. On the other hand, at high frequencies, the wavelength is short, so the vibrations are damped relatively quickly. However, high-frequency waves are subject to various disturbances such as the influence of non-linearity included in the vibration and noise transmission system of the vehicle, and their transfer characteristics are likely to change.

Therefore, when the secondary sound is divided into low frequency control and high frequency control, the following method can be considered in consideration of the difference in these characteristics.

First, for the low frequency control system, the sampling frequency of the digital control system is lowered as compared with the high frequency system so that a long attenuation time can be expressed with a small number of filter taps. On the other hand, in the high frequency control system, the sampling frequency is kept unchanged, and the level of the low frequency component contained in the input (noise signal and reference signal) is attenuated by the high pass filter to improve the high frequency digital resolution. Then, for each secondary sound control signal, another secondary sound output means (loudspeaker) executes mute control. In this case, a relatively small loudspeaker can be used for high-frequency control, which saves space.

By the way, if the adaptive speeds of the low-frequency control system and the high-frequency control system (specified by the convergence coefficient of the updating formula of an adaptive filter described later) are made the same, the following problems occur.

In the high frequency control system, the level of the low frequency component is attenuated by the high pass filter. in this case,
As for the noise signal, a low frequency component can be considerably attenuated by using a filter having a certain degree of steepness (high order), but it is difficult to completely eliminate the low frequency component. On the other hand, for the reference signal, a filter that is too steep (high order) cannot be used because of the problem of the response delay of the filter. Therefore, a low-order low-attenuation gradual high-pass filter is used, but in this case, the low-frequency component after passing through the high-pass filter has a level of at most the same level as the high-frequency component.

The adaptive filter is adapted from the frequency where the coherence of the noise signal and the reference signal is relatively high, but the coherence is often higher in the low frequency region. In such a situation, the adaptation to the low frequency component is performed by both the low frequency control system and the high frequency control system.
A small speaker for high frequency is originally used for the secondary sound output of the high frequency control system, and a sufficient low tone cannot be output as compared with the output level of the secondary sound control signal, which is extremely inefficient.

The present invention has been made in view of the above problems, and a problem to be solved by the present invention is that, like road noise, it is distributed over a wide band and the absolute sound pressure level becomes more prominent in the lower frequency range. For random noise with a certain spectral structure, by separating the control system for the low frequency region and the high frequency region and giving priority to the low frequency control system, the low frequency noise and the high frequency noise can be controlled in good balance. To provide a noise control system.

[0018]

In order to solve the above-mentioned problems, in the present invention, a random noise having a spectral structure which is distributed over a wide band and whose absolute sound pressure level is superior in a lower frequency range is adopted. On the other hand, one or a plurality of noise detecting means and one or a plurality of noise detectors that output a secondary sound for canceling the noise (primary sound) by actively interfering with the noise (primary sound) at the noise detection position. Next sound output means, means for detecting, as a reference signal, one or a plurality of signals close to the noise source of the detected noise and having a high correlation with the noise, and the above-mentioned means for minimizing a certain evaluation function. An active noise control device comprising: an adaptive signal processing means having one or a plurality of adaptive filters for generating a secondary sound control signal from a reference signal obtained by a reference signal detecting means, the device comprising: Signal and the reference signal are separated into a low frequency component and a high frequency component, the secondary sound control signal generated by the adaptive signal processing means is separated into a low frequency sound control signal and a high frequency sound control signal, and a secondary sound output means. Low frequency sound control and high frequency sound control separately, and at the initial stage of adaptation, the adaptation speed of low frequency sound control is set to be higher than that of high frequency sound control, and after a certain control effect is achieved. The feature is that the adaptive speed of the high frequency sound control is increased and executed.

[0019]

In the active noise control system of the present invention, the noise detecting means detects noise. The secondary sound output means outputs a secondary sound for canceling the noise (primary sound) by actively interfering with the noise (primary sound) at the noise detection position. As the reference signal, a signal highly correlated with the detected noise is used. The adaptive signal processing means separates the detected noise signal and the reference signal into a low frequency component and a high frequency component, and separately generates a secondary sound control signal for low frequency sound control and high frequency sound control. At that time, in the initial stage of adaptation, the control by the low frequency control system is prioritized and executed, and the operation of the high frequency control system is executed after the low frequency noise is sufficiently suppressed.

[0020]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The contents of the present invention will be described in detail below with reference to the drawings. FIG. 1 shows an example of the overall configuration of an active noise control device for road noise. The system is mainly composed of an acceleration sensor attached to each suspension for reference signals, a noise detection microphone arranged in the vehicle compartment, a speaker 3 for outputting a secondary sound for silencing, a control controller 4, etc. A plurality of sensors, microphones, and speakers are used.

When the tires are vibrated by the unevenness of the road surface and each suspension vibrates during traveling, the acceleration sensor 1 detects the vibration acceleration and supplies it as a sensor detection signal 101 to the controller 4. The sensor detection signal 101 is converted into a digital signal 102 via an analog low pass filter 41 and an A / D converter 42. The microprocessor 43 in the controller 4 convolves the adaptive digital filter 103 with the digital signal 102 as a reference signal to generate a secondary sound output control signal 104. The output control signal 104 is amplified by the power amplifier 5 via the D / A converter 44 and the analog low-pass filter 45, and is output as the secondary sound 105 from the speaker 3. On the other hand, the vehicle interior sound pressure signal 10 obtained from the microphone 2
6 is an analog low-pass filter 40 and an A / D converter 50
Is converted into a digital signal via the input terminal and input to the microprocessor 43. The adaptive control algorithm sometimes sets each filter coefficient of the adaptive digital filter at a certain sampling interval so that the sound pressure in the vehicle interior is minimized due to the interference between the secondary sound output of the speaker 3 and the primary sound (road noise). It will be updated every moment.

In the system of FIG. 1, the adjustment of the filter coefficient of the adaptive digital filter is performed by the Multiple Error Filter.
The red-x LMS algorithm (abbreviated as MEFX-LMS algorithm) is used. Where MEFX-
For details of the LMS algorithm, see "Signal
Processing for Active Control -AdaptiveSignal Pro
cessing- ", Hareo HAMADA, International Symposium o
n ActiveControl of Sound and Vibration, ASJ Proc. '
91, Tokyo, April 9-11, 1991, pp.33-4
It is as explained in detail in 4th grade.

FIG. 2 shows a digital active noise control system including K sensors (reference signals), M speakers (secondary sound output), and L microphones (error signals) using the MEFX-LMS algorithm. The block diagram is shown. From the figure, the secondary sound control output y _m (n) at the time of the n-th sample is represented by the following equation: the reference signal x _k (n) and the adaptive filter w _mk (i) (i = 0 to I-1). ) Is given by the convolution of.

[0024]

## EQU1 ## y _m (n) = Σ _i w _mk (i) x _k (n−i) where m = 0 to (M−1), k = 0 to (K−1), i = 0 to (I
-1) M: the number of speakers, K: the number of sensors, I: the number of taps of the adaptive filter Further, the update formula of the adaptive filter w _mk (i) is expressed by the following formula.

[0025]

[Number 2] _{w mk (i) (n +} 1) = λ mk · w mk (i) (n) -α mk · Σ l e l (n) r lmk (n-i) r lmk (n) = Σ _j C ^ _lm (j) x (n-j) where C ^ _lm (j) is a model function of the acoustic transfer system between the m-th speaker and the l-th microphone and has a coefficient of J digital FIs.
It is represented by an R filter.

In the adaptive filter update equation of the equation 2, the coefficients α _mk and λ _mk are called the convergence coefficient and the leaky parameter, respectively. If the convergence coefficient α _mk is large, the update amount of the adaptive filter w _mk for each time becomes large and the adaptation time until it converges to the optimum value becomes short, but if it is too large, the control becomes unstable and the sound increase and divergence phenomena occur. It is easy to cause. On the other hand, the leaky parameter λ _mk usually takes a value of 1 or less, and the absolute value of each coefficient is reduced each time the adaptive filter is updated. That is,
The function of the leaky parameter can suppress the excessive growth of the adaptive filter. However, if the function of the leaky parameter is too large compared to the update amount, the adaptive filter cannot grow and a sufficient control effect cannot be obtained.

FIG. 3 shows a case in which low frequency noise and high frequency noise are controlled separately for noise in which low frequency noise is predominant such as the road noise shown in FIG. Here, in the examples shown in FIGS. 1 and 2, a multi-input multi-output system in which the number of reference signals, the number of sound pressure detections, the number of secondary sound outputs, and a plurality of both are provided. However, since there is no essential change, in the following description of the embodiments, the description will be given as a 1-input 1-output system.

First, the signal after passing through the low-pass filter 41 (anti-alias filter) in FIG. 1 is divided into two systems,
One of them is passed through the high-pass filter 410 of low order (about 1st to 2nd order) and only the high frequency component is passed through, and the other is left as it is, and the digital signal 20 is passed through the A / D converter 142.
Converted to 2.

Next, the vehicle interior sound pressure signal 106 obtained from the microphone 2 is similarly converted to the analog low-pass filter 4.
After passing through one, it is divided into two systems, one of which is passed by the high-pass filter 411 so that only the high-frequency component is passed through, and the other is left unchanged, and the digital signal 2 is passed through the A / D converter 142.
It is converted into 06 and input to the microprocessor 43.
Here, the high-pass filter 411 may use a filter steeper (higher order) than 410.

Next, by the adaptive control algorithm, adaptive filters w _H (i) in the high frequency control system and the low frequency control system,
w _L (i) is a sound pressure signal e _H (n), e _L (n), a reference signal x
_H (n), x _L (n) and the model function C _H ^ (j) of the acoustic transfer system,
It is updated as follows using C _L ^ (j).

[0031]

[Mathematical formula-see original document] w _H (i) = λ _H · w _H (i) + α _H · e _H (n) · r _H (n−i) r _H (n) = ΣC _H ^ (n−i) · x _{H (n) w L (i} ) = λ L · w L (i) + α L · e L (n) · r L (n-i) r L (n) = ΣC L ^ (n-j) · x _L (n) where α _H , λ _H , α _L and λ _L are the convergence coefficient and leaky parameter when updating w _H (i) and w _L (i).

On the other hand, the sensor detection signal 10 serving as a reference signal
1 and the vehicle interior sound pressure signal 106 are used as they are.
That is, the secondary sound control signals y _H (n) and y _L (n) in the high frequency control system and the low frequency control system are

[0033]

## EQU4 ## y _H (n) = Σw _H (i) · x _H (n−i) y _L (n) = Σw _L (i) · x _L (n−i) Then, they are respectively input to the high-frequency speaker 30 and the low-frequency speaker 31 and become the secondary sound output.

The adaptive convergence time (secondary sound control signal generation speed) of the high frequency control system and the low frequency control system can be freely adjusted by setting the respective convergence coefficients. FIG. 5 shows that when the adaptive control is started from the initial state, the low-frequency control system is controlled with priority over the high-frequency control system (control is performed so as to converge early), and the convergence coefficients are changed every moment. The setting table is shown.

Now, considering the case where the adaptive filter starts the adaptation from the initial value 0 when the vehicle starts traveling,
First, the convergence coefficient α _L is set to be larger so that the low frequency control system converges in a short time. After the lapse of time T _c , α _L is gradually decreased to shift to a stable control state of only fine adjustment.

Next, the convergence coefficient α _H , which was initially set to 0 or a small value, is increased to control the low frequency noise. At this point in time, as a result of the control by the low frequency control system, the component having the coherence with the original reference signal in the low frequency noise included in the primary sound is removed by the interference of the secondary sound and detected by the microphone 2. Of the noise, the only silenceable component is the high frequency spectrum component. Therefore, the adaptive filter of the high frequency control system does not react to the low frequency noise and adapts so as to mute the coherent component of the high frequency component.

The timing of control switching (or adaptive speed adjustment) between the high frequency control system and the low frequency control system can also be performed by using some index indicating the control state of the low frequency control system. For example, the band-limited sound pressure signal e _L (n),
Alternatively, the squared value e _L (n) ² or the mean square value <e _L (n) ² > during a certain period of time is monitored, and if this value falls below a certain value, the high frequency It is conceivable to start execution of the control system. Also, since the output sound pressure level of the secondary sound can be estimated from the size of the adaptive filter,
For example, an index as shown in the following equation,

[0038]

## EQU5 _## It is also possible to calculate W _L = Σ _i | w _L (i) | at regular sample intervals and use it as an index for control switching.

FIG. 6 shows the output power W _{L of} this adaptive filter.
An example is shown in which a threshold value Wp is provided for the value of and the values of the convergence coefficients α _H and α _L are tabulated and adjusted according to the magnitude thereof. That is, in the initial adaptation of W _L = 0, α _L has a large value and α _{H has} a value of 0. Next, as the adaptive filter grows and W _H increases, α _L decreases. When W _L exceeds a certain threshold Wp, α _H ≠
The execution of the high frequency control system is started as 0.

The device for actively muting sound by using the secondary sound with respect to the noise (primary sound) has been described above. However, due to the nature of the present invention, for example, low frequency vibration and high frequency vibration are mixed. It can also be applied to the case where active vibration control is performed for wide band vibration.

[0041]

According to the present invention, the noise signal and the reference signal are reduced even with respect to random noise having a spectral structure which is distributed over a wide band and whose absolute sound pressure level is superior in a low frequency region. By separating the low frequency sound control and the high frequency sound control separately,
There is an advantage that the control is not concentrated on the low frequency component and can be stably controlled over a wide frequency band from the low frequency to the high frequency.

[Brief description of drawings]

FIG. 1 is a configuration diagram of an active silencer for road noise.

FIG. 2 is a diagram showing a digital active noise control system.

FIG. 3 is a low-frequency / high-frequency control system parallel control block diagram.

FIG. 4 is a diagram showing an example of a low-frequency / high-frequency control system convergence coefficient setting method.

FIG. 5 is a diagram showing an example of a low-frequency / high-frequency control system convergence coefficient setting method.

FIG. 6 is a diagram showing an example of a vehicle interior sound pressure spectrum.

FIG. 7 is an explanatory diagram of a conventional active noise control device.

[Explanation of symbols]

1 ... Acceleration sensor, 2 ... Microphone, 3 ... Speaker, 4 ... Controller, 40, 41 ... Low pass filter, 42, 50, 142 ... A / D converter, 43 ... Microprocessor, 101 ... Sensor detection signal, 105 ... 2 Next sound, 106 ... Sound pressure signal in vehicle compartment, 410, 411 ... High pass filter.

Claims (1)

[Claims] 1. Random noise having a spectral structure distributed over a wide band and having an absolute sound pressure level predominant in a lower frequency range, has one or a plurality of noise detecting means and its noise detecting position. At one or a plurality of secondary sound output means for outputting a secondary sound for canceling the noise (primary sound) by actively interfering with sound waves and a noise source of the detected noise and Means for detecting one or a plurality of signals having high correlation with the noise as a reference signal, and generating a secondary sound control signal from the reference signal obtained by the reference signal detecting means so as to minimize a certain evaluation function. In the active noise control device comprising adaptive signal processing means having one or a plurality of adaptive filters, the noise signal and the reference signal are separated into a low frequency component and a high frequency component, and the adaptive signal is separated. The low-frequency sound control and the high-frequency sound control are separately provided by supplying the secondary sound control signal to the processing means, separately generating the low-frequency sound control signal and the high-frequency sound control signal, and supplying the generated signals to the secondary sound output means. Active noise that can be applied to the high-frequency sound control at a higher speed than the high-frequency sound control in the initial stage of adaptation to achieve a certain control effect. Control device.