Active noise control method and system using variable actuator and sensor participation

Abstract

A method for reducing noise in at least one monitor position in a vehicle compartment by actively controlling the power of a primary noise (d.sub.m(t)) as sensed at two or more control positions in the vehicle compartment, the method comprising the updating of filter coefficient(s) of (an) adaptive filter(s) (w(n)) based on variable contribution of error sensors and actuator(s) for different noise source operating conditions.

Claims

1. A method for reducing noise in at least one monitor position in a vehicle compartment by actively controlling the power of primary noise (d.sub.m(t)) as sensed at two or more control positions in said vehicle compartment, the primary noise originating from a noise source transmitting noise (x(t)) through a respective primary path (P.sub.m) to the respective control position, the method comprising: arranging at least one actuator in the compartment, arranging an error sensor in each control position, arranging at least one adaptive filter (w.sub.k(n)) per actuator, arranging an adaptive algorithm unit providing updated filter coefficients to the at least one adaptive filter (w.sub.k(n)), arranging at least one reference sensor providing a reference signal x(n), coherent with the noise (x(t)) from the noise source, to the at least one adaptive filter (w.sub.k(n)) and to the adaptive algorithm unit, applying the at least one adaptive filter (w.sub.k(n)) to the reference signal (x(n)) to provide and transmit a drive signal (y.sub.k(n)) to its respective actuator, arranging the at least one actuator to, as a response to the drive signal (y.sub.k(n)), provide and transmit a respective secondary noise (y.sub.k(t)) through a respective secondary path (S.sub.km) between the actuator and the respective control position, arriving at the respective control position as a respective secondary anti-noise (y′.sub.m(t)), arranging the error sensors to provide and transmit a respective error signal (e.sub.m(n)), representing a sensed residual noise (e.sub.m(t)) of the sensed primary noise and sensed secondary anti-noise, to the adaptive algorithm unit, arranging an actuator and error sensor weighting device to receive signal(s) (c(n)) representing noise source operating condition(s), to determine a set of weighting factors (mp.sub.m(n), kp.sub.k(n)) for each actuator and error sensor, respectively, based on the signal(s) (c(n)) representing the noise source operating condition(s), and to transmit the determined set of weighting factors to the adaptive algorithm unit, and arranging the adaptive algorithm unit to, based on the received set of weighting factors, provide updated filter coefficients to the at least one adaptive filter (w.sub.k(n)) to reduce the power of the residual noise (e.sub.m(t)) sensed in at least one of the control positions, Wherein the adaptive algorithm unit comprises: a filter update device, a filtering and weighting device arranged to filter the reference signal (x(n)) with a respective secondary path digital model (Ŝ′.sub.km) of the respective secondary path (S.sub.km), update the filtered reference signal based on the received set of weighting factors, and to transmit the filtered and weighted reference signal (x′.sub.km(n)) to the filter update device, an error sensor weighting device (10) arranged to determine respective weighted error signals (e′.sub.m(n)) by applying respective error sensor weighting factors (mp.sub.m(n)) to the respective error sensor signal (e.sub.m(n)), and to transmit the weighted error signal(s) (e′.sub.m(n)) the to the filter update device, wherein the filter update device is arranged to update the filter coefficients of the adaptive filter step wise by an iterative process using the expression: $w_k\left(n+1\right)=\left(1-\mu {\gamma }_k\right)w_k\left(n\right)-\mu \underset{m=1}{\overset{M}{.Math.}}{x}_{\mathrm{km}}^{\prime }\left(n\right)e_m^{\prime }\left(n\right)$ wherein μis the step size k represents the k.sup.th actuator m represents the m.sup.th error sensor w.sub.k(n) is a vector containing the current set of filter coefficients w.sub.k(n+1) is a vector containing the updated set of filter coefficients x′.sub.km(n) is a vector containing a time history of the weighted and filtered reference signal x(n) e′.sub.m(n) is the weighted error signal from the m.sup.th error sensor and μλ.sub.k is the leakage factor.

2. The method of claim 1, wherein the weighting factors for the error sensors and actuator(s) for a certain noise source operating condition are determined from a set of predetermined relationships between signals (c(n)) representing different noise source operating conditions and corresponding predetermined weighting factors.

3. The method of claim 2, wherein the predetermined weighting factors for the error sensors and actuator(s) for a certain noise source operating condition are determined from predetermined spatial characteristics of a primary noise field in the compartment and predetermined spatial characteristics of a secondary anti-noise field in the compartment corresponding to a minimal residual noise level in at least one of the monitor positions.

4. The method of claim 2, wherein the predetermined weighting factors and signals (c(n)) representing different noise source operating conditions are stored as a lookup table.

5. The method of claim 2, wherein the weighting factors for error sensors and actuator(s) for a certain operating condition are determined by interpolation of stored weighting factors.

6. The method of claim 1, wherein the weighting factors for the error sensors and actuator(s) for a certain noise source operating condition are determined as a function of predetermined weighting factors and a variable that represents a change of the noise source operating condition(s).

7. The method of claim 1, wherein the signal(s) (c(n)) representing (a) noise source operating condition(s) are extracted from a computer bus/network of the vehicle, from one or more error sensors, from a tachometer signal, from one or more vibration sensors, or from the reference sensor used in the method.

8. The method of claim 1, wherein the adaptive algorithm unit applies the weighting factors to an LMS algorithm selected from a group comprising filtered-reference-LMS, leaky-filtered-reference-LMS, filtered-error-LMS, leaky-filtered-error-LMS, normalized-filtered-reference-LMS and normalized-leaky-filtered-reference-LMS.

9. The method of claim 1, wherein the adaptive algorithm unit applies the weighting factors to an RLS algorithm selected from a group comprising filtered-reference-RLS, leaky-filtered-reference-RLS, normalized-filtered-reference-RLS and normalized-leaky-filtered-reference-RLS.

10. The method of claim 1, wherein the reference signal (x(n)) is filtered with an adaptive FIR-filter w.sub.k as:
y.sub.k(n)=w.sub.k.sup.T(n)×(n) where
x(n)=[x(n) x(n−1) . . . x(n−L.sub.w+1)].sup.T
w.sub.k(n)=[w.sub.k,0(n) w.sub.k,1(n) . . . w.sub.k,L.sub.w.sub.−1(n)].sup.T where L.sub.w is the number of coefficients of the adaptive filter and n is the current time step.

11. An active noise control system for reducing noise in at least one monitor position in a vehicle compartment by active control of the power of primary noise (d.sub.m(t) as sensed at two or more control positions in said vehicle compartment, the primary noise originating from a noise source transmitting noise (x(t)) through a respective primary path (P.sub.m) to the respective control position, the system comprising: at least one actuator arranged in the compartment, an error sensor arranged in each control position, at least one adaptive filter (w.sub.k(n)) arranged per actuator, an adaptive algorithm unit arranged to provide updated filter coefficients to the at least one adaptive filter (w.sub.k(n)), at least one reference sensor arranged to provide a reference signal x(n), coherent with the noise (x(t)) from the noise source, to the at least one adaptive filter (w.sub.k(n)) and to the adaptive algorithm unit, wherein the at least one adaptive filter (w.sub.k(n)) is arranged to be applied to the reference signal (x(n)) to provide and transmit a drive signal (y.sub.k(n)) to its respective actuator; wherein the at least one actuator is arranged to, as a response to the drive signal (y.sub.k(n)), provide and transmit a respective secondary noise (y.sub.k(t)) through a respective secondary path (S.sub.km) between the actuator and the respective control position, arriving at the respective control position as a respective secondary anti-noise (y′.sub.m(t)), and wherein the error sensors are arranged to provide and transmit a respective error signal (e(n)), representing a sensed residual noise (e.sub.m(t)) of the sensed primary noise and sensed secondary anti-noise, to the adaptive algorithm unit; an actuator and error sensor weighting device arranged to receive signal(s) (c(n)) representing noise source operating condition(s), to determine a set of weighting factors (mp.sub.m(n), kp.sub.k(n)) for each actuator and error sensor, respectively, based on the signal(s) (c(n)) representing the noise source operating condition(s), and to transmit the determined set of weighting factors to the adaptive algorithm unit, wherein the adaptive algorithm unit is arranged to, based on the received set of weighting factors, provide updated filter coefficients to the at least one adaptive filter (w.sub.k(n)) to reduce the power of the residual noise (e.sub.m(t)) sensed in at least one of the control positions, Wherein the adaptive algorithm unit comprises: a filter update device, a filtering and weighting device arranged to filter the reference signal (x(n)) with a respective secondary path digital model (Ŝ′.sub.km)of the respective secondary path (S.sub.km), update the filtered reference signal based on the received set of weighting factors, and to transmit the filtered and weighted reference signal (x′.sub.km(n)) to the filter update device, an error sensor weighting device (10) arranged to determine respective weighted error signals (e′.sub.m(n)) by applying respective error sensor weighting factors (mp.sub.m(n)) to the respective error sensor signal (e.sub.m(n)), and to transmit the weighted error signal(s) (e′.sub.m(n)) to the filter update device, wherein the filter update device is arranged to update the filter coefficients of the adaptive filter step wise by an iterative process using the expression: $w_k\left(n+1\right)=\left(1-\mu {\gamma }_k\right)w_k\left(n\right)-\mu \underset{m=1}{\overset{M}{.Math.}}{x}_{\mathrm{km}}^{\prime }\left(n\right)e_m^{\prime }\left(n\right)$ wherein μis the step size k represents the k.sup.th actuator m represents the m.sup.th error sensor w.sub.k(n) is a vector containing the current set of filter coefficients w.sub.k(n+1) is a vector containing the updated set of filter coefficients x′.sub.km(n) is a vector containing a time history of the weighted and filtered reference signal x(n) e′.sub.m(n) is the weighted error signal from the m.sup.th error sensor and μλ.sub.k is the leakage factor.

12. A method of reducing the power of residual noise (e.sub.m(t)) sensed in at least one control position arranged in a compartment of a motor vehicle, comprising operating an active noise control system of claim 11 in the motor vehicle.

13. The method of claim 12, wherein the motor vehicle is a road vehicle.

14. The method of claim 13, wherein the road vehicle is a car.

15. The method of claim 12, wherein the motor vehicle is an aircraft.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) FIG. 1 illustrates an ANC system in a car.

(2) FIG. 2 shows a block scheme of a modified ANC system/method.

(3) FIG. 3 is a schematic view of the modified ANC system/method of FIG. 2.

(4) FIG. 4 is a detailed schematic view of the modified ANC system/method of FIG. 2

(5) FIG. 5 shows a block scheme of a prior art ANC system/method.

DETAILED DESCRIPTION OF THE DRAWINGS

(6) In FIG. 1 is illustrated an active noise control (ANC) system 1 arranged in a car. The system comprises actuators 2, e.g. loudspeakers, and error sensors 3, e.g. microphones, arranged in the car compartment and an ANC controller 100. Such ANC systems may also be arranged in other vehicles such as aircrafts, busses, trains, boats etc.

(7) The ANC system 1/method shown in FIGS. 1-4, may be used to reduce disturbing sound (noise) in at least one monitor position 13 in the compartment by actively controlling the power of primary noise d.sub.m(t) as sensed at two or more control positions in the vehicle compartment. The monitor positions 13 are typically positions in the compartment where the passenger's ears can be located. Control positions are positions in which error sensors 3 may be installed. The primary noise d.sub.m(t) may originate from a noise source 5 transmitting noise x(t) through a respective primary path P.sub.m to the respective control position. Such primary noise may be generated by mechanical vibrations of an engine 5 (as illustrated in FIG. 1) and/or components mechanically or acoustically coupled thereto (e.g., a fan), wind passing over and around the vehicle, and/or tires contacting, for example, a paved surface, propeller noise waves exciting the walls of the passenger compartment.

(8) With the ANC system 1/method the power of primary noise d.sub.m(t) as sensed at the two or more control positions in the vehicle compartment may be actively controlled. The control positions may be positions in the compartment where it is possible to install error sensors 3 and in which the primary noise d.sub.m(t) is controlled, e.g. eliminated or at least reduced.

(9) The ANC system 1/method comprises M error sensors 3, arranged in respective control positions in the compartment. The ANC system should comprise at least two error sensors 3. The system 1 comprises K actuators 2. The number of actuators 2 and error sensors 3 used in the system/method depends on the application and size of the compartment. Preferably, the number of error sensors 3 used should not be less than the number of actuators 2 used.

(10) At least one adaptive filter w.sub.k(n) may be arranged for each actuator 2 and there may be an adaptive algorithm unit 6 arranged to provide updated filter coefficients to the at least one adaptive filter w.sub.k(n).

(11) A reference sensor 4 may be arranged to provide a reference signal x(n), which signal is coherent with the noise x(t) from the noise source 5, to the adaptive filter(s) w.sub.k(n) and to the adaptive algorithm unit 6. The variable n here represents the latest sample of the signals, i.e. x(n) is the latest sample of the time continuous x(t).

(12) The adaptive filter w.sub.k(n) is applied to the reference signal x(n) to provide and transmit a drive signal y.sub.k(n) to its respective actuator 2.

(13) The actuator(s) 2 may be arranged to, as a response to the drive signal y.sub.k(n), provide and transmit a respective secondary noise y.sub.k(t) through a respective secondary path S.sub.km between the actuator 2 and the respective control position arriving at the respective control position as a respective secondary anti-noise y′.sub.m(t). The error sensors 3 may be arranged to provide and transmit a respective error signal e.sub.m(n), representing a sensed residual noise e.sub.m(t) of the sensed primary noise and sensed secondary anti-noise, to the adaptive algorithm unit 6.

(14) The aim of the secondary anti-noise y′.sub.m(t) is to be an opposite-phase image of the sensed primary noise d.sub.m(t). The degree to which the secondary anti-noise y′.sub.m(t) matches the primary noise d.sub.m(t) determines the sensed residual noise e.sub.m(t) and the corresponding error signal e.sub.m(n). If the primary noise and the secondary anti-noise were matched exactly, both in space and time, the primary noise would be completely eliminated at the control position and the error signal e.sub.m(n) would be zero in a control position.

(15) An actuator and error sensor weighting device 7 may be arranged to receive signal(s) c(n) representing noise source 5 operating condition(s), e.g. rotational speed of the motor, propeller speed, vehicle speed, engine power setting or combinations thereof. The actuator and error sensor weighting device 7 may also be arranged to determine a set of weighting factors mp.sub.m(n), kp.sub.k(n) for each actuator 2 and error sensor 3 used in the system/method, based on the signal(s) c(n) representing the noise source operating condition(s).

(16) A weighting factor is a factor that determines the contribution of an actuator 2 or error sensor 3 in the system/method. Some of the actuators/error sensors may be adjusted to contribute more than others in the reduction of residual noise e.sub.m(t) sensed in a control position. Sometimes actuators/error sensors may be switched off and not used at all. The weighting factors are variable depending on different noise source operating conditions.

(17) The error sensor weighting device 7 may transmit the determined set of weighting factors to the adaptive algorithm unit 6. The adaptive algorithm unit 6 may be arranged to, based on the received set of weighting factors, provide updated filter coefficients to the at least one adaptive filter w.sub.k(n) to reduce the power of the residual noise e.sub.m(t) sensed in at least one of the control positions.

(18) The updating of the filter coefficient(s) of the at least one adaptive filter w.sub.k(n) may be a continuous and iterative process where the update is performed in steps and where the update is based on the variable weighting factors. The updating of the filter coefficient(s) is, hence, based on variable contribution of error sensors 3 and actuator(s) 2 for different noise source operating conditions. Thereby achieving optimal spatial alignment of actuators/error sensors for different noise source operating conditions.

(19) A distribution of actuators 2 and error sensors 3 in the compartment may be spatially optimal for a given noise disturbance, but may not be adapted when the noise disturbance changes, such as when noise source operating conditions change. In such case, using a different spatial distribution of actuators 2 and error sensors 3 may improve the performance of the method.

(20) To achieve spatial alignment which results in global sound control the error sensors 3 and actuators 2 must be carefully selected. This is done by measurements and optimizations in the design phase of the ANC system/method for a specific vehicle. To determine which sensors 3 and actuators 2 give the best global control steady-state simulations are performed for many different combinations of actuators 2 and sensors 3. The number of possible combinations in such optimization depends on selected system size and number of possible locations to select among. For cavities like in a car or similar it is possible to try all combinations and select the one which gives best global control, i.e. that minimizes the error signals in the whole compartment. However, for cavities in e.g. busses or aircrafts the number of combinations is far too big to be able to try all combinations. In that case optimization algorithms like “random walk” or “simulated annealing” may be used.

(21) Such optimizations will typically find different set of optimal actuators 2 and error sensors 3 depending on type of noise source and operating conditions. For example, when controlling engine noise the optimal set of actuators 2 and sensors 3 depends on engine rotational speed and which engine order that dominates. Traditionally the optimizations have been set up to find one set of actuators 2 and sensors 3 that performs reasonably well at all conditions.

(22) In FIG. 2 is shown a simplified block scheme of the ANC system/method discussed above. FIG. 2 illustrates that the method/system uses signal(s) c(t) representing noise source 5 operating condition(s), to update drive signals y.sub.k(t) to the actuators 2, such that the updating thereof is based on variable contribution of error sensors 3 and actuator(s) 2 for different noise source operating conditions. Thereby achieving optimal spatial alignment of actuators/error sensors for different noise source operating conditions.

(23) In FIG. 5 is shown a block scheme of common prior art ANC systems/methods in which drive signals y.sub.k(t) to the actuators 2 are only updated based on signals from the reference sensor 4 and error sensors 3.

(24) In FIG. 4 is a more detailed description of the system/method in FIG. 3. As seen in FIG. 4, the adaptive algorithm unit 6 may comprise a filter update device 8, a filtering and weighting device 9 and an error sensor weighting device 10. The filtering and weighting device 9 may be arranged to filter the reference signal x(n) with a respective secondary path digital model Ŝ.sub.km of the respective secondary path S.sub.km and to update the filtered reference signal based on the received set of weighting factors.

(25) The secondary path digital model Ŝ.sub.km represents a transfer function between an actuator 2 and an error sensor 3. It may be determined offline (when there is no disturbing noise) in a calibration step, or online (in presence of the disturbing noise), through so-called online secondary path modelling techniques.

(26) The transfer functions may as below be represented by FIR filters and the filtered and weighted reference signal x′.sub.km(n) may then be determined by a dot-product as
x′.sub.km(n)=ŝ′.sub.mk.sup.T(n)x(n)
where
ŝ′.sub.mk(n)=mp.sub.m(n).Math.kp.sub.k(n).Math.[ŝ.sub.mk,0(n) ŝ.sub.mk,1(n) . . . ŝ.sub.k,L.sub.s.sub.-1(n)].sup.T
x(n)=[x(n) x(n−1) . . . x(n−L.sub.s+1)].sup.T x′.sub.km(n) is the weighted and filtered reference signal x(n), n is the current time step, x(n) is a vector containing a time history of the reference signal x(n). Ŝ′.sub.mk(n) is a vector containing L.sub.s coefficients of the weighted and filtered FIR filter representing the secondary path Ŝ.sub.mk between actuator k and error sensor m.

(27) The filtered and weighted reference signal x′.sub.km(n) may be transmitted from the filtering and weighting device 9 to the filter update device 8. The error sensor weighting device 10 may be arranged to determine respective weighted error signals e′.sub.m(n) by applying respective error sensor weighting factors mp.sub.m(n) to the respective error sensor signal e.sub.m(n), and to transmit the weighted error signal(s) e′.sub.m(n) the to the filter update device 8. The filter update device 8 may be arranged to update the filter coefficients of the adaptive filter step wise by an iterative process using the expression:

(28) $w_k\left(n+1\right)=\left(1-\mu {\gamma }_k\right)w_k\left(n\right)-\mu \underset{m=1}{\overset{M}{.Math.}}{x}_{\mathrm{km}}^{\prime }\left(n\right)e_m^{\prime }\left(n\right)$ wherein μ is the step size k represents the k.sup.th actuator m represents the m.sup.th error sensor w.sub.k(n) is a vector containing the current set of filter coefficients w.sub.k(n+1) is a vector containing the updated set of filter coefficients x′.sub.km(n) is a vector containing a time history of the weighted and filtered reference signal x(n). e′.sub.m(n) is the weighted error signal from the m.sup.th error sensor. μγ.sub.k is the leakage factor.

(29) The adaptive algorithm unit 6 may apply the weighting factors to an LMS algorithm, an RLS algorithm or any other suitable algorithm.

(30) A weighting factor for an error sensor 3/actuator 2 for a certain noise source operating condition may be determined from a set of predetermined relationships between signals c(n) representing different noise source operating conditions and corresponding predetermined weighting factors. Predetermined weighting factors may be determined by optimizing the method for various noise source operating conditions. The predetermined weighting factors may be stored together with the corresponding signals representing different noise source operating conditions, e.g. as a lookup table.

(31) The pre-determined weighting factors for the actuator(s) 2 and error sensor(s) 3 may be saved as weighting matrices.

(32) $M_p\left(c\right)=\left[\begin{array}{ccc}{\mathrm{mp}}_1\left(c\right)& \mathrm{.Math.}& 0\\ \mathrm{.Math.}& \ddots & \mathrm{.Math.}\\ 0& \mathrm{.Math.}& {\mathrm{mp}}_m\left(c\right)\end{array}\right]$$K_p\left(c\right)=\left[\begin{array}{ccc}{\mathrm{kp}}_1\left(c\right)& \mathrm{.Math.}& 0\\ \mathrm{.Math.}& \ddots & \mathrm{.Math.}\\ 0& \mathrm{.Math.}& {\mathrm{kp}}_k\left(c\right)\end{array}\right]$

(33) wherein mp and kp are participation factors for error sensors 3 and actuators 2 respectively and c represents a variable vehicle operating condition, e.g. rpm, wheel speed or similar.

(34) A weighting factor for an error sensor 3/actuator 2 for a certain noise source operating condition may be determined by interpolation, such as linear interpolation or other curve fitting technique, of stored weighting factors.

(35) Below is an example with linear interpolation where the operating condition c(n) is between condition c0 and c1 and the actuators 2 and error sensors 3 have been updated with weighting (participation factors).

(36) ${\mathrm{mp}}_m\left(n\right)={\mathrm{mp}}_m\left(c0\right).Math.\left(1-\frac{c\left(n\right)-c0}{c1-c0}\right)+{\mathrm{mp}}_m\left(c1\right).Math.\left(1-\frac{c1-c\left(n\right)}{c1-c0}\right)$${\mathrm{kp}}_k\left(n\right)={\mathrm{kp}}_k\left(c0\right).Math.\left(1-\frac{c\left(n\right)-c0}{c1-c0}\right)+{\mathrm{kp}}_k\left(c1\right).Math.\left(1-\frac{c1-c\left(n\right)}{c1-c0}\right)$

(37) A weighting factor for an error sensor 3/actuator 2 may be determined as a function of a predetermined weighting factor and a variable that represents the change of the vehicle operating condition.