WO1995024309A2

WO1995024309A2 - Non invasive error sensing dve method and apparatus

Info

Publication number: WO1995024309A2
Application number: PCT/US1995/002433
Authority: WO
Inventors: Dexter G. Smith
Original assignee: Noise Cancellation Technologies, Inc.
Priority date: 1994-03-02
Filing date: 1995-03-01
Publication date: 1995-09-14
Also published as: WO1995024309A3

Abstract

A method of cancelling pressure phenomena using only a single sensor (10) to sense the noise by: a) producing an attenuating noise; and b) measuring the tonal result of said original and said attenuated noise without sensing noise generated by local turbulence; via DVE (32), loudspeaker (33), and microphone (34).

Description

NON INVASIVE ERROR SENSING DVE METHOD AND APPARATUS

Field Of The Invention

The present invention relates to active cancellation systems for repetitive or non- repetitive phenomena, and, more specifically, to an active cancellation system that provides cancellation phenomena without requiring an external reference or timing signal.

Background Of The Invention

The simplest active cancellation for phenomena is an analog negative feedback system, also referred to as "a virtual earth" system. In such systems, phenomena is sent to an actuator, which provides cancellation phenomena into the area in which the phenomena is to be canceled.

A problem with analog feedback systems is that they become unstable and produce destructive positive feedback due to phase shifts. These phase shifts are typically due to delays such as that resulting from the distance between the sensor and the actuator, and also by echoes. The phase shifts vary by frequency and have not been amenable to a solution in the analog feedback systems except for over a very narrow range of frequencies or in a very confined environment, such as a headphone.

Digital cancellation systems have been provided for canceling repetitive noise. One such system has been described in "Adaptive Noise Canceling Principals and

Application" by Widrow et al in Proceedings IEEE Vol. 63, No. 12, December 1975, which describes two forms of active adaptive cancellers. The first, illustrated in Figure 5 of that paper, uses a multi-tap adaptive FIR filter with a reference signal correlated with the noise to be canceled. The reference signal is required to be within 90° in phase of the error signal. Consequently, the reference signal used by the adapter itself often requires filtering. This approach is referred to as the "filtered-x algorithm". This approach requires two sensors and does not account for variable delays in the system.

The second form described by Widrow et al is illustrated in Figure 6 of that paper, and provides a single frequency notch filter and requires only two single tap filters. Again, a reference signal correlated with the noise is used and is phase shift 90° for one of the filters. This approach cannot operate without the reference signal and does not account for variable delays in the system.

Therefore, there is a need for an active cancellation system for repetitive or non- repetitive phenomena which uses only a single sensor, solves the instabilities as described and extends the frequency range over that of the conventional analog virtual earth systems, accounts for environmentally produced delays, and is adaptable to maintain an appropriate phase relationship without the use of external reference signals. In U.S. Patent Number 5,105,377, hereby incorporated by reference herein, there is described an active cancellation system for repetitive or non-repetitive phenomena which estimates the noise signal by subtracting the predicted effects of the cancellation signal from the residual sensor signal. In this single input, single output system, an LMS filtered-x algorithm is employed to adapt the cancellation filter coefficients. The adaptive filter produces the cancellation signal by filtering the estimated noise with filter weights that are adapted using the residual signal and the estimated noise convolved with the system impulse response. The single channel version of the DVE algorithm is described by the equations: ^xk ^{= e}k " ∑ . yj- . i

Yk ⁼ ∑ ^Am,k'^xk - m m

^Am,k + l ^{= A}m,k - ^α * ^ek' ∑ Cι .^χ _k _ ι where: y_k is the cancellation signal value at sample k, βk is the error signal value at sample k,

C is the vector of coefficients of the impulse response from the controller output to the error sensor input,

A is the vector of coefficients of the cancellation filter, x_j- is the value of the estimated noise signal at sample k and α is the LMS convergence rate coefficient.

This form of the algorithm is only applicable to systems consisting of one channel, i.e., a single sensor and a single actuator.

Duct systems can take many shapes and forms but certain features are common to all. There is a prime energy converter to change primary energy (usually electricity) to mechanical energy. Next, the mechanical energy is used to move air about the structure where the system is installed. Ductwork is used to contain and direct the air to the end user location. Along with the air itself, noise from the energy conversion system propagates down the ductwork. Depending on the installation, this noise can be quite annoying to the end users especially if the fan chosen has a noticeable blade passage tonal which is the number of blades times the revolutions per second of the fan.

Many patents, articles and textbooks describe noise cancellation for ducts. "Active Control of Sound" by Nelson and Elliott, Academic Press, 1992, hereby incorporated by reference herein, is a good compilation of recent activity in the field. Regardless of the algorithm involved, they describe placing the error residual microphone in the duct or flush with the wall of the duct to sense the sound pressure in the duct. Implementation problems always arise with the addition of the input noise sensor (microphone) to the interior of the duct due to locally generated turbulence and environmental factors affecting the microphone. Linear velocities can be 10 m/sec or higher. Environmental factors may or may not be well controlled especially in the event or some sort of system failure. Microphones typically used for cost, simplicity and ability for remote mounting are capacitive electret types. These are made by exposing a charged plate to the environment to be sensed. A thin, acoustically transparent cover are over them, but this is by no means a robust environmental covering. Thus, the Nelson and Elliott solution is not satisfactory.

Summary Of The Invention

A digital virtual earth cancellation system for duct borne repetitive or non- repetitive noise is provided according to the present invention which receives a phenomena input signal from outside the duct system representing residual phenomena to be canceled and includes an adaptive filter for generating a cancellation signal. The adaptive filter adapts its filtering characteristics as a function of the difference between the residual signal and the estimated effects of the cancellation signal. A phase circuit maintains the adapting of the filtering characteristics and 90° phase of the phenomena signal. In the present invention, the impulse response of the entire cancellation system, which includes delays introduced by filters and other factors, is convolved with the output of the cancellation system, i.e., the cancellation signal. This value is subtracted from the externally sensed residual signal that is received by the cancellation system, to provide an estimate of the noise. This substantially eliminates the problems associated with destructive feedback due to phase shifts. The residual signal is used to control an adaptive filter that receives the estimated noise as an input. The adaptive filter produces the cancellation signal by filtering the estimated noise with filter weights that are adapted using the externally sensed residual signal and the estimated noise convolved with the system response. The error sensing microphone in a DVE duct cancellation system must sense the summation of noise plus anti-noise but not sense locally generated turbulence in the duct. It also must be placed so that an appropriate transfer function from speaker to microphone can be determined by the DVE algorithm and so that attenuation is maximized in the appropriate area, in this case the interior of a duct. Therefore, the microphone may be taken out of the duct if these conditions are met. If the duct wall is not very acoustically transparent, a small section of the duct may be replaced with a more acoustically transparent material and the microphone is then mounted outside the duct. This non-invasive error sensing DVE invention for duct cancellation of repetitive or non-repetitive noise takes the error residual noise sensor (microphone) out of the interior of the duct. The advantage is that direct contact with turbulent air flows and/or corrosive environments is avoided. By convoluting the estimate of the noise with the system impulse response to adapt the filter, the values sent to an adapter for the adaptive filter are kept within 90° phase of the residual signal to provide convergence of the adaptation.

An embodiment of the present invention measures the system impulse response from speaker mounted flush with the duct wall to the externally mounted microphone and includes a test signal generator for generating a test signal which is combined with the cancellation signal and provided to in the area to be monitored. An adaptive filter is provided that receives a random test signal and provides a filtered signal. A difference is produced between the filtered signal and the phenomena residual signal. An adapter adapts the filter weights of this adaptive filter as a function of the difference signal and a delay line of test signal values. The filter weights represent the measured impulse response of the system.

Other advantages and novel features of the present invention will become apparent from the following detailed description of the invention when considered in conjunction with the accompanying drawings in which Figure 1 is a block diagram of a virtual earth negative feedback system,

Figure 2 shows a typical duct system, and

Figure 3 shows a diagrammatic view of the invention.

Detailed Description Figure 1 shows a basic block diagram of a virtual earth, negative feedback system. Phenomena (such as noise) is detected by a sensor 10, which sends out a sensing signal. This sensor signal is affected by an anti-aliasing filter (not shown in Figure 1) and other factors which have an impulse response E. The sensor signal, as affected by the impulse response E, results in a residual signal r to a processor 11. From the residual signal r, the processor 11 produces an output signal y, the signal being the cancellation signal y. The cancellation signal y, used to form the canceling phenomena, is affected by filters, transit delays, and other factors which have an impulse response S. The output from the actuator 12, the cancellation phenomena, combines with the original phenomena and the residual is detected at a location 13 by the sensor 10. The signal values are given by r(t) = (y(t) * S + n(t)) * E where: n(t) is the noise at time t, r(t), y(t), S and E are as described above and a*b is the convolution of a and b.

An estimate of the noise itself can be obtained by rearranging the above equation as n(t) * E = r(t) - y(t) * (S * E)

The noise n(t) convolved with the impulse response E is found by subtracting the effects of the processor output y, as convolved by S and E, from the residual signal r received by the processor 11. The average power of r can be minimized by a gradient descent method, such as a known least mean square (LMS) algorithm as described in Patent No. 5,105,377.

This derivation places no restrictions on the location 13 of sensor 10 other than sensor 10 must here the training signal from processor 11 output from processor 12. A typical air movement system is seen in Figure 2. Here, fan 20 is the primary energy conversion system that moves air through duct 21 through air diffuser 26 to end user occupied space 25. In this embodiment, it is desired to cancel tonal noise from fan 20 before it reaches room 25. The DVE system is installed with loudspeaker 23 mounted flush in the duct along with microphone 24. Controller 22 senses the noise in the duct 21 and provides an output signal to actuator 23. The problem with this noise canceling system is that microphone 24 senses locally generated turbulence at its location. Much effort needs to be spent trying to find a good location for it to sense noise but not turbulence. It is also very difficult to move the microphone around inside a duct 21, reseal the duct 21 and then try the cancellation system. It is a very iterative process.

This invention is seen in Figure 3. Here, fan 30 is the primary energy conversion system that moves air through duct 31 through air diffuser 38 to end user occupied space 35. In this embodiment, it is desired to cancel tonal noise from fan 30 before it reaches room 35. The DVE system is installed with loudspeaker 33 mounted flush in the duct. Microphone 34 is mounted outside the duct 31 at location 36. (Note that location 36 may be just the duct wall itself or a small section of duct wall replaced by a more acoustically transparent material). Controller 32 senses the noise in the duct 31 and provides an output signal to actuator 33. The problem with the above noise canceling system is solved in that microphone 34 senses tonal noise propagated through the duct 31 at location 36 and does not sense noise generated by turbulence inside the duct 31. It is now a simple matter to manually move microphone 34 around outside the duct 31 to achieve maximum noise attenuation. Having described the invention it will become apparent to those of ordinary skill in the art that many changes and modifications can be made to the invention without departing from the scope of the appended claims.

Claims

1. The method of canceling repetitive or non-repetitive phenomena using only a single sensor which produces an impulse response to enhance the frequency range over conventional analog virtual earth systems, accounts for environmentally produced delays and maintains an appropriate phase relationship without the use of external reference signals, said method comprising sensing the noise to be canceled in a given system with a sensor means, producing a counter noise to attenuate said noise, measuring the tonal result of said noise and counter noise with said sensor means without sensing the noise generated by local turbulence within the system, evaluating the residual noise and causing a change in the counter noise to cancel said noise.

2. The method of claim 1 and including the step of moving said sensor means so as to maximize said attenuation.

3. The method of claim 1 which includes convolving the impulse response of the entire cancellation system with the cancellation signal.

4. The method of claim 3 which includes subtracting the value of the convolving from the sensed residual signal to provide an estimation of the noise to eliminate problems associated with destructive feedback due to phase shifts.

The method of claim 4 wherein the residual signal is used to control a filter step which produces a cancellation signal, the filtering step includes filtering the estimated noise with filter weights adapted by using the externally sensed residual noise convolved with impulse response.

6. A noise canceling system adapted to increase the frequency range over conventional analog virtual earth systems, account for local produced delays and maintains an appropriate phase relationship without the use of external reference signals, said system comprising a housing means adapted to provide for a fluid flow therethrough and having an inner surface, sensor means located adjacent said inner surface and adapted to sense the noise within the housing, actuator means located on said inner surface upstream from said sensor means, controller means operatively connected to said sensor means and actuator means so as to cancel tonal noise accompanying said fluid flow.

7. The system of claim 6 wherein said sensor means is adapted to sense only the tonal noise within the housing means and not noise assembled by turbulence within the housing means.

8. The system of claim 7 wherein the sensor means is adapted to be moved around on said inner surface to readily determine the optimum sensing location for cancellation.

9. The system of claim 7 wherein said sensor means is located on said housing means on the outer surface thereof.

10. The system of claim 7 wherein said controller means is adapted to convolve the impulse response of the entire system with the output of the cancellation system.

11. The system of claim 10 wherein said controller means is additionally adapted to subtract the convolved value from the externally sensed residual signal received by the system.