Note: Descriptions are shown in the official language in which they were submitted.
WO93/11529 PCT/US91/08772
'3
2 1 2 4 1 8 3
ACTIVE NOISE CONTROL OF AN ENCLOSURE WITH MULTIPLE
TRANSDUCERS
RAckgrolln~ of the Inventlon
The present invention relates to the development of an
improved arrangement for controlling repetitive phenomena
cancellation in an arrangement wherein a plurality of
residual repetitive phenomena sensors and a plurality of
cancelling actuators are provided. The repetitive phenomena
being cancelled in certain cases may be unwanted noise, with
microphone sensors and loudspeaker as the repetitive
phenomena sensors and cancelling actuators, respectively.
The repetitive phenomena being cancelled in certain other
cases may be unwanted physical vibrations, with vibration
sensors and counter vibration actuators as the repetitive
phenomena sensors and cancelling actuators, respectively.
A time domain approach to the noise cancellation
problem is presented in a paper by S. J. Elliott, I.M
Strothers, and P.A. Nelson, "A Multiple Error LMS Algorithm
and Its Application to the Active control of Sound and
Vibration," IEEE Transactions on Acoustics, Speech, and
Signal Processing, VOL. ASSP-35, No. 10, October 1987, pp.
1423-1434.
The approach taught in the above paper generates
cancellation actuator signals by passing a single ref^rence
signal derived from the noise signal through Na FIR filters
whose taps are adjusted by a modified version of the LMS
algorithm. The assumption that the signals are sampled
synchronously with the noise period is not required. In
fact, the above approach does not assume that the noise
signal has to be periodic in the first p~t of the paper.
However, the above approach does assume that the matrix of
impulse responses relating the actuator and sensor signals
is known. No suggestions on how to estimate the impulse
responses are made.
WO93/11529 PCT/US91/08772
2124183
The frequency domain approach to the interpretation of
the problem is presented as follows, as shown in Figure 5
herein which is a block diagram of the system:
The system consists of a set of Na actuators driven by
a controller that produces a signal C which is a Na x 1
column vector of complex numbers. A set of Ns sensors
measures the sum of the actuator signals and undesired
noise. The sensor output is the Ns x 1 residual vector R
which at each harmonic has the form
R = V + HC
where
V is a Ns x 1 column vector of noise components
and
H is the Ns x Na transfer function matrix between the
actuators and sensors at the harmonic of interest.
The problem addressed by the present invention is to
choose the actuator signals to m~m~ze the sum of the
squared magnitudes of the residual components. Suppose that
the actuator signals are currently set to the value C which
is not necessarily optimum and that the optimum value is
Copt = C + dC. The residual with Copt would be
Ro = H (C + dC) + V = (HC + V) + H dC = R + H dC
The problem is to find dC to minimize the sum squared
residual
Ro@Ro
where @ denotes con~ugate transpose. An equivalent
statement of the problem is: Find dC so that H dC is the
least squares approximation to -R. This problem will be
represented by the notation
-R == H dC
The solution to the least squares problem has been
studied extensively. One approach is to set the derivatives
of the sum squared error with respect to the real and
W O 93/11529 PC~r/US91/08772
21 241 83
imaginary parts of the components of dC equal to 0. This
leads to the "normal equations"
H@ H dC = -H@R
S If the columns of H are linearly independent, the
closed form solution for the required change in C is
dC = - [H@H]-1H@R
The present invention provides methods and arrangements
for accommodating the interaction between the respective
actuators and sensors without requiring a specific pairing
of the sensors and actuators as in prior art single point
cancellation techniques such as exemplified by U.S. Patent
4,473,906 to W~rnAk~, U.S. Patents 4,677,676 and 4,677,677
to Friksson, and U.S. Patents 4,153,815, 4,417,098 and
lS 4,490,841 to Ch~pl' n. The present invention is also a
departure from prior art techniques such as described in the
above-mentioned ~ll;ott et ~l. article and U.S. Patent
4,562,589 to W~rn~k~ which handle interactions between
multiple sensors and actuators by using time domain filters
which to not provide means to cancel selected harmonics of a
repetitive phenomena.
Active noise control has been shown to be effective in
reducing low frequency noise in applications such as
mufflers, headsets, engine mounts, fans, etc. Adaptive
control has proved to be an effective technique for lhe
implementation of active noise attenuation. Most
applications have focused on single channel adaptive control
(one sensor and actuator pair) or multiple channels when the
interactions between channels is negligible. Applications
such as cabin quieting and active enclosures have i..ade
apparent the need for multiple channel control algorithms.
In these applications because of the complexity of the noise
source, a single transducer will not be able to provide
attenuation at the required regions. Additionally,
35 interactions on the multiple t ansducers can cause adaptive
algorithms to become unstable unless the interactions are
WO93/11529 PCT/US91/08772
4 21 24t 83
accounted for in the control process. University of
Maryland has developed the MISACT algorithm. This invention
considers the problem of controlling noise radiating from an
enclosure with multiple openings and multiple transducers.
Performance of the MISACT Algorithm will be shown both
experimentally and using a simulation model.
In the past active noise control has been used to
reduce low frequency noise in applications such as mufflers,
headsets, etc. These attempts were noted in the article by
G. Eatwell, M.J. Burke, Kh. Eghtesadi and W. E. Gossman
entitled "The Application of Active Cancellation to Vehicle
Noise and Vibration" presented at the 1990 International
Conference on Quiet Revolutions. The use of adaptive
control has been shown to be effective in implementing
active noise attenuation. Various papers on this phenomena
have been presented including
B Widrow, ~Adaptive Noise Cancelling: Principles and
Applications", Proceedings of the IEEE, 63(12), 1962-
1716, 1975 and,
S.J. Elliott, I Stother and P. Nelson, "A Multiple
Error LMS Algorithm and Its Application to the Control
of Sound and Vibration", IEEE Transactions on
Acoustics, Speech and Signal Processing, 35(10), 1423-
1434, 1987.
The use of the MISACT algorithm [Multiple Interacting
Actuators as Sensors] is discussed by Kh. Eghtesadi, M.P.
McLoughlin and E.W. Ziegler, Jr., "Development of the
Simulation Model of the Multiple Interacting Sensors and
Actuators (MISACT) for an Active Control System",
Proceedings of the Conference on Recent Advances ln Active
control of Sound and Vibration, 246-257, 1991.
The algorithm is also discussed in co-pending
Application, PCT/GB90/2021 which is hereby incorporated by
reference herein.
Others have devised single channel systems such as in
U.S. Patent No. 4,989,252 to Toshiba in which the enclosure
2 1 2 4 1 8 3
must have a dimensional correlation with itself and a
standing wave of sound to be attenuated. However, these
systems do not allow for "cross talk" between multiple
units operating at the same frequency and allows for two
or more holes to be located in the enclosure walls. The
aforementioned Toshiba patent allows for only one hole.
In the Toshiba system the limitation to one opening means
that the opening must be relatively large to provide
adequate airflow and therefore limiting the frequency
range that can be cancelled. The Toshiba system also has
- trouble operating in the presence of loud interfering
noise.
The basic active noise cancellation employed here is
described in U.S. Patent No. 4,417,098.
Summarv of the Invention
A primary object of this invention is to provide an
enclosure having multiple means to quiet sounds therein.
Another object of this invention is the provision of
multiple transducers within an enclosure to allow for
multiple openings in the enclosure so as to facilitate
fluid flow.
Still another object of this invention is the
application of a multiple interacting actuators and
sensors algorithm to the task of quieting an enclosure.
Yet another object of this invention is the use of a
multi-channel active noise cancellation system in the
quieting by a noise cancellation system of an enclosure.
In accordance with one aspect of the invention there
is provided a system for actively controlling and quieting
repetitive noise arising from one or more noise sources
comprising: an enclosure means surrounding said one or
more noise sources; multiple opening in said enclosure
means; speaker means mounted in near proximity to said
multiple openings and adapted to cause deadening of sound
generated from said noise sources; residual microphone
, 5a 21 241 83
means adapted to receive noise and to generate electrical
signals in response thereto; controller means connected to
said speaker means and adapted to receive signals from
said residual microphone means to thereby generate counter
sound in the proximity of said multiple opening means to
thereby deaden the sound emanating from said noise sources
within said enclosure; and synchronous signal means
whereby an electronic signal is fed from said noise source
to said controller means.
In accordance with another aspect of the invention
there is provided an active noise cancellation system for
actively cancelling noise emitted by an enclosure with
multiple openings containing at least one noise source,
said system comprising: multiple speaker means affixed
adjacent said multiple openings and adapted to generate
sound of opposite polarity to that emanating from said
noise source so as to cancel it; microphone means mounted
in the proximity of said enclosure means and to received
sound from said noise source and enclosure and transmit an
electrical impulse in response thereto; controller means
operatively connected to said speaker means and said
microphone means so as to cause signals to be generated to
said speaker means to generate said opposite polarity
sound; and synchronous signal means connecting said noise
source and said controller means.
These and other objects will become apparent when
reference is had to the accompanying specification and
drawings in which:
Fig 1 is a diagrammatic view of a multi-channel
active noise cancellation system showing the overall
system, and
Fig 2 shows a noise spectrum plot of frequency versus
sound level, and
Fig 3 shows a similar noise spectrum plot of
frequency versus sound level at the residual microphone
location, and
WO93/11529 PCT/US91/08772
~ 6 2124183
Fig 4 shows a specific application of the multiple
channel approach to silencing a refrigerator compressor, and
Fig 5 is a block diagram of the algorithm frequency
time domain approach to controlling the instant system.
The application of active noise quieting suggests
itself to the area of home appliances.
Home appliances do not produce noise levels that are
dangerous or very obnoxious. The purpose of quieting
appliances falls under the goal of providing a kitchen with
an overall sound power level not greater than 40 dBA or to
provide an appliance quieter than the competition. In a
kitchen, for example, refrigerators, microwaves, rangehoods
and dishwashers are all candidates for quieting techniques.
Fans, motors and fluid noise are all present in the
lS appliances mentioned above. Quieting techniques will have
to address all the noise sources in a particular appliance.
Quieting one source may well make a previously masked source
annoying.
A typical kitchen refrigerator has two main sources of
noise, the compressor and the freezer compartment fan. The
compressor consists of an electric motor and compressor
device such as a piston. The fan is usually an axial type
mounted in the freezer compartment. The enclosure referred
to in this application can be either of these compartments.
To develop torque, the compressor motor turns at a rate
slightly slower than the line frequency, e.g. 58.5 Hz
instead of 60 Hz. This frequency is the fundamental rate of
the noise heard from the compressor. Harmonics of this
fundamental at varying amplitudes are then heard. In
addition, there is a low pressure inlet and high pressure
outlet valve that open and close each cycle that produce the
"clicking" type noise heard from the compressor.
The refrigeration system has other noise sources also.
There is a tone around 1500 Hz that varies in amplitude 2S a
function of time that is produced in the piston part cf tne
compressor and is fluid borne into the cooling coils where
W O 93/11529 ~ PC~r/US91/08772
2 ~ 24 1 83
it can be heard. The expansion valve in the freezer
compartment produces noisy turbulent fluid flow in the
return line to the compressor. This noise varies with time
and is proportional to the amount of Freon in the system
that is being moved by the compressor (much less noticeable
if the Freon level is low).
A natural question at this point is why active control
of noise? Why not just a foam lined box around the noise?
The problem is heat. A refrigerator collects heat from the
freezer compartment and dumps it into the room by way of the
cooling coils. Enclosing the compressor and coils will
necessitate an inlet and outlet and fan possibility for heat
transfer and you are back where you started. Enclosing the
compressor only is a better idea because the heat transfer
to the room is not compromised and the source of the most
noise is directly addressed. An inlet and outlet for heat
is necessary because the compressor does get hot; however,
no fan is needed because the flow of Freon through the
compressor helps to cool it.
Enclosing the compressor and adding active control is
good for several reasons. Passive materials lower higher
frequencies where active alone is not as effective. Active
control of lower frequencies is very effective where passive
materials are not. The low frequency noise that can be
2~ heard all through the house is cancelled at the source
giving good global cancellation. Heat transfer from the
coils is not interfered with. This technology allows all
the active components to fit within the enclosure and in
hign volumes is very cost effective.
Specific Descr;ption
A simulation of the MISACT algorithm was developed to
assist in predict~ng the performance of the attenuator
system. The simulator uses a model of the operating
environment to reproduce the interactions between actuators
anc sensors c~d can either be user defined or experimentally
WO93/11529 PCT/US91/08772
8 21 241 83
measured. The simulator will accept up to four actuators
and four sensors, along with user inputs for the noise type
and frequency range, transfer functions type tmeasured or
user defined) and sample rates. For this paper the transfer
functions between all speakers and microphones were measured
in real time by the NCT 2010 controller. Simulation runs
were made with noise frequencies of 228 and 456 Hz. To help
illustrate the importance of interactions simulations were
also run for the case of a diagonal transfer function matrix
(no interactions). Results are tabulated in Table l for
both noise frequencies. The simulation results show that
when full interaction between sensors is used, the overall
noise level of both noise frequencies is reduced. However,
when the interactions are not used the algorithm becomes
unstable at around 228 Hz.
For experimental measurements an enclosure as in Figure
l was constructed. The sides of the enclosure were
constructed of plywood. A speaker inside the enclosure
served as the noise source. Two 6 inch ports were used as
outlets. Speakers were mounted to inject the cancelling
signal at each port. Microphone elements were mounted at
the top of the cabinet to provide the feedback signal. Both
microphones and speakers were connected to a NCT 2010 MISACT
controller. A synthetic noise source was connected to the
speaker inside the enclosure and provided a speed signal to
the controller. A B&K type 2230 sound level meter was used
to measure the overall noise reduction along with a
Tektronix 2630 spectrum analyzer.
The controller performed a calibration to determine the
system response between the full matrix of speakers and
microphones. A noise signal consisting of 228 and 456 Hz
was then generated in the enclosure. The controller was
then enabled and allowed to reach steady state operating
condition. The noise spectrum measured at the monitor
microphone is shown in Figure 2 for the controller both on
WO93/11529 PCT/US91/08772
9 2124183
and off. The overall noise level was reduced from 59.3 dBA
to 45.5 dBA. To illustrate the importance of accounting for
the interactions, the control was implemented with two
independent channels. The response at the two error
microphones is shown in Figure 3. While the 456 Hz tone is
reduced, the level of the 228 Hz torl~ is increased. It
should be noted that the signal level was prevented from
increasing further by limiting logic in the software.
Table 1, following, shows the simulation results for
interacting versus independent control.
Noise Fre~uency Control Mo~e Sensor 1 Sensor 2 Over~ll
228 HZ interacting -3.8 dB -8 .1 dB -5.4 ds
228 Hz Non-Interacting 15.8 dB 20.0 dB 18.4 dB
456 Hz Interacting -4.2 dB -12.7 dB -5.6 dB
456 Hz Non-Interacting -11.7 dB -3.1 dB -5.6 dB
Table 1. Simulation results for interacting cs. independent
control.
The instant invention solution to refrigerator quieting
is seen in the attached drawings, Fig 4. It is a
combination active and passive approaoh.
A shell was constructed around tne compresso
compartment to take out the high frequency tones and random
noise. This shell also hac an input slit and output port to
allow heat from the compressor to es--oe. The shell and
port were designed to resonate at a~ roximately 60 Hz.
Testing showed that the compressor s~ ll temperature stayed
under manufacturer's guidelines ev-:~ when subjected to an
ambient temperature of 110 degrees L'.
An accelerometer placed on the compressor shel served
as the n~_se sync signal. The antinoise actuator was a
speaker _nside the shell. The speaker cabinet was optimized
for the low frequency fundamental, 58.7 Hz, of the
compressor. A microphone on the outside of the shell served
as the residual sensor. The NCT control algorithm then
WO93/11529 PCT/US91/08772
21 241 83
calculated the impulse response from the speaker to the
residual microphone and produced the correct cancelling
signal. A reduction of lO dBA rear and 5 dBA front over an
identical unmodified refrigerator was obtained.
The uniqueness of the solution is as follows:
l) Combination passive and active approach for a system
solution.
2) Tuning the shell at the fundamental frequency of
interest.
3) Heat transfer of the compressor taken into account.
4) Flexible residual microphone location for optimum
cancellation when installed.
5) Speaker location flexible inside the shell.
6) Very low (milliwatts) power needed for speaker.
7) Non-sensitive to loud transients.
This invention shows control of low frequency noise
from an enclosure with multiple transducers. It was shown
that simulation can provide insight about performance of an
active control system. The simulation model has been used
as an analytical tool on electronic mufflers and vibration
mounts. Furthermore it was demonstrated both through
simulation and experiment that the MISACT algorithm will
remain stable in situations where independent control
channel may become unstable.
