JP2007272008A - Active noise controller and active vibration controller - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To prevent occurrence of unnecessary variation in control output due to the influence of restrictions of a maximum value based upon the expression form of a parameter. <P>SOLUTION: An active noise controller 10 comprises a reference signal generator 12 which outputs a cosine wave and a sine wave from the frequency (f) of a keytone, an adaptive notch filter 14 which outputs a control signal y(n) for canceling noise, a speaker 16 which outputs a control sound, a microphone 18 which detects an error signal (e), a filter coefficient update means 22 of sequentially updating filter coefficients Wx and Wy of the adaptive notch filter 14 so that the error signal (e) becomes minimum, and a limit processing means 13 of finding limited filter coefficients Wxo and Wyo obtained by limiting the filter coefficients Wx and Wy by multiplication by the ratio Rate of a square root Aobj and a threshold Aobj when the square root Anow of the sum of squares of Wx as a real part signal and Wy as an imaginary part signal of the filter coefficients is larger than the threshold Aobj. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to an active noise control apparatus that actively controls noise generated from a noise source, and an active vibration control apparatus that actively controls transmission of vibration generated from a vibration source. The present invention relates to an active noise control apparatus suitable for application to a closed space such as a room of a moving body having a noise source and a vibration source. The moving body may be a vehicle such as an automobile, a ship, an amphibious vehicle, a pleasure boat, a helicopter, an airplane, or the like.

  Recently, there has been proposed an active noise control device that controls engine sound that can be heard in a passenger compartment of a vehicle with control sound output from a speaker, and reduces engine sound at the position of an occupant's ear (for example, Patent Document 1). , See Patent Document 2), which is put into practical use in some vehicles.

  In this active vibration noise control apparatus, attention is paid to the fact that the vibration noise in the vehicle interior is generated in synchronization with the rotation of the engine output shaft in the vehicle. The sound is muted using a filter to keep the passenger compartment quiet. An adaptive notch filter is a filter based on adaptive control (a control method in which one or more parameters are measured to meet the mark criteria and used to change the feedback signal).

  There are two control methods for adaptive control. The first is a method of updating the coefficient of an FIR (finite impulse response) adaptive filter that treats discrete data at regular intervals on the time axis as time axis data. A second method is to convert discrete data at regular intervals on the time axis into a frequency domain, perform a predetermined process, and reversely convert it into data in the real time domain at the time of output. In any control method, it is possible to prevent control divergence by limiting the filter coefficient value.

  Active noise control devices and active vibration control devices are often mounted on mobile bodies such as automobiles, and a compact and inexpensive microcomputer is used as a general control means because of restrictions on price, space, power consumption, etc. It is common to be used. In a small microcomputer, it is common to use a fixed-point type (or an integer type) that can perform simple and high-speed computation instead of an exponential expression type with a large computation load.

  On the other hand, a technology that actively suppresses vibration by using vibration instead of noise has been developed. Active type that detects vibration generated by the engine and reduces the transmission of this vibration to the passenger compartment. A vibration control device has been proposed (Patent Document 3).

  In this active vibration control device, an engine is supported on a vehicle body frame via a vibration sensor, an electromagnetic actuator, and an engine mount, and generates vibrations that cancel the engine vibration detected by the vibration sensor with the electromagnetic actuator. By driving, vibrations transmitted from the engine to the vehicle body frame are suppressed, and the interior of the vehicle is kept quiet while enhancing the ride comfort.

  Further, from the viewpoint of fuel efficiency during vehicle travel, a technique related to cylinder deactivation in which a cylinder supply / exhaust valve is stopped has been proposed in order to reduce resistance such as engine supply / exhaust resistance (see, for example, Patent Document 4). ).

JP 2000-99037 A JP 2004-361721 A Japanese Patent Laid-Open No. 5-99262 Japanese Patent No. 3415601

  When adaptive data is handled by converting discrete data to the frequency domain, a vector can be represented by a real part signal and an imaginary part signal that are orthogonal on the complex plane. When a microcomputer is used as a general control means, a fixed-point type is often used for the representation format of these real part signals and imaginary part signals.

  In the exponential phenotype, the range of numerical values that can be expressed is remarkably wide, and the range that can be expressed is substantially not limited by the maximum value that can be expressed. On the other hand, in the fixed-point type, the maximum value that can be expressed is limited by the number of bits and cannot represent an excessively large value. Therefore, the vector handled in adaptive control must be expressed in a rectangular area defined by the maximum value of the real part signal and the maximum value of the imaginary part signal. Since both the real part signal and the imaginary part signal are generally assigned the same number of bits as the expression format, the vector is represented in a square region.

  If there is a discrepancy between the control frequency indicated by the reference signal indicating the engine speed, etc. and the actual noise frequency, the noise vector is projected on the complex plane with the reference signal of the control frequency as the phase reference, and then rotated around the origin. To do. By the way, the complex plane for representing this vector is limited to a square defined by the maximum value on the fixed-point representation, so when the noise vector is large, it will rotate along the four sides of the square. Therefore, the magnitude of the vector will fluctuate.

  In addition, when controlling a plurality of frequencies simultaneously, preferable control characteristics differ for each frequency at a predetermined engine speed, and there is a case where priority is given to each frequency to limit the control output. However, the assignment to the dynamic range of the output is not clear on the complex plane limited to a square, and there is a possibility that unnecessary fluctuations occur in the control output.

  The present invention has been made in consideration of such problems, and in an active noise control device and an active vibration control device having a frequency domain type adaptive filter, unnecessary fluctuations occur in the control output of the adaptive filter. An object of the present invention is to provide an active noise control device and an active vibration control device that prevent this.

  An active noise control apparatus according to the present invention includes a reference signal generator that outputs a harmonic reference signal from a frequency of noise generated from a noise source, and a control for canceling the noise when the reference signal is input. An adaptive notch filter that outputs a signal; a sound output device that outputs the control signal as a control sound; a sound detector that detects a cancellation error between the noise and the control sound and outputs the error signal; and the sound output A correction filter that has a transfer function from a detector to the sound detector, outputs a reference signal when the reference signal is input, and receives the error signal and the reference signal, thereby minimizing the error signal. Filter coefficient updating means for sequentially updating the filter coefficient of the adaptive notch filter, and the filter coefficient is expressed in a frequency domain divided into a real part signal and an imaginary part signal, and the real part signal and the imaginary part signal Limit processing means for limiting the real part signal and the imaginary part signal by multiplying the ratio of the square root value and the predetermined threshold when the square root value of the sum of squares thereof is greater than the predetermined threshold. And

  As described above, when the square root value of the square sum of the real part signal and the imaginary part signal of the filter coefficient is excessive, the real part signal and the imaginary part signal are multiplied by the ratio of the square root value and the predetermined threshold value. By limiting, the vector indicated by the filter coefficient rotates with the amplitude based on the threshold around the origin on the complex plane, and it is possible to prevent unnecessary fluctuations from occurring in the control output of the adaptive filter.

  In this case, the noise source may include a rotating body, and the reference signal generator may specify harmonics of the noise frequency based on the number of rotations of the rotating body. Thereby, the harmonics of the noise frequency can be specified simply, quickly and appropriately.

  A square root table in which a square root corresponding to an argument is recorded, and the adaptive notch filter obtains the square root value by referring to the square root table with a square sum of the real part signal and the imaginary part signal as an argument. It may be. Thereby, the calculation of the square root can be performed easily and quickly, and the square root can be obtained with the fixed-point type or the integer type without converting the parameter into an exponential expression format or the like.

  An inverse square root table in which the inverse of the square root corresponding to the argument is recorded, and the adaptive notch filter refers to the square root inverse table with the square sum of the real part signal and the imaginary part signal as an argument, A numerical value may be obtained and used to limit the threshold. As a result, the reciprocal of the square root can be calculated simply and quickly, and the reciprocal of the square root can be obtained in the fixed-point type or integer type without converting the parameter into an exponential expression format or the like.

  The predetermined threshold value may be “0”, and the output of the control signal may be stopped. In this way, the control output may be stopped in order to suppress the occurrence of unnecessary fluctuations.

  In this case, if the square root is limited stepwise by the predetermined threshold value, a fade-out process is performed, and the generation of a “bot” sound (so-called pop-up noise) can be prevented.

  The limit processing means may change the predetermined threshold based on the number of rotations. That is, since the magnitude of noise and the peaks or valleys or phase of the sound field change depending on the rotational speed and control frequency of the rotating body of the noise source, an appropriate threshold value may be set based on the rotational speed.

  The noise source is a multi-cylinder internal combustion engine, and includes a cylinder deactivation function unit that stops operation of at least one of the plurality of cylinders, and the adaptive notch filter is configured to perform the predetermined operation based on the number of cylinders that are operating. The threshold value may be changed. That is, since the magnitude of noise and the peaks or valleys or phase of the sound field change depending on the number of operating cylinders, an appropriate threshold value may be set based on the number of operating cylinders.

  An active vibration control apparatus according to the present invention includes a reference signal generator that outputs a harmonic reference signal from a frequency of vibration generated from a vibration source, and a control for canceling the vibration when the reference signal is input. An adaptive notch filter that outputs a signal, an actuator that outputs the control signal as a control vibration, a vibration detector that detects an offset error between the vibration and the control vibration, and outputs the error signal, and the vibration from the actuator A correction filter having a transfer function up to a detector and receiving a reference signal and outputting a reference signal; and receiving the error signal and the reference signal so that the error signal is minimized. Filter coefficient updating means for sequentially updating the filter coefficient of the notch filter, and the filter coefficient is represented in a frequency domain divided into a real part signal and an imaginary part signal, Limiting the real part signal and the imaginary part signal by multiplying the ratio of the square root value and the predetermined threshold value when the square root value of the sum of squares of the partial signal and the imaginary part signal is greater than the predetermined threshold value It is characterized by.

  In this manner, when the parameter in the adaptive notch filter is excessive, the parameter is complex by limiting the real part signal and the imaginary part signal by multiplying the ratio of the square root value and the predetermined threshold. The rotation is performed with the amplitude based on the threshold around the origin on the plane, and it is possible to prevent unnecessary fluctuations from occurring in the control output of the adaptive filter.

  The vibration source may include a rotating body, and the reference signal generator may specify harmonics of the vibration frequency based on the number of rotations of the rotating body. Thereby, the harmonics of the noise frequency can be specified simply, quickly and appropriately.

  According to the active noise control device and the active vibration control device according to the present invention, by limiting the square root value of the square sum of the real part signal and the imaginary part signal of the parameter in the adaptive notch filter by a predetermined threshold, The parameter rotates with the amplitude based on the threshold around the origin on the complex plane, and it is possible to prevent unnecessary fluctuation from occurring in the control output of the adaptive filter.

  Hereinafter, an active noise control device and an active vibration control device according to the present invention will be described with reference to embodiments. First, an active noise control device 10 according to the present embodiment will be described with reference to FIGS.

  FIG. 1 is a block diagram showing a configuration of an active noise control apparatus 10 according to an embodiment of the present invention. The active noise control apparatus 10 according to this embodiment is basically composed of a microcomputer (control means) 1. The microcomputer 1 includes a CPU, a storage unit, an input / output port, a bus for connecting them, and the like. The CPU reads and executes a program stored in the storage unit to realize various means described later.

  FIG. 2 is a schematic diagram in which the active noise control device 10 shown in FIG. 1 is mounted on a vehicle 30 that is a moving body having an engine 28. The engine 28 is provided with a cylinder deactivation function for stopping the operation of at least one cylinder among a plurality of cylinders (for example, four cylinders) under the action of the engine controller 28a (cylinder deactivation function unit). This cylinder deactivation function is, for example, as described in Patent Document 4, in order to reduce the resistance such as the supply / exhaust resistance of the engine from the viewpoint of fuel efficiency during vehicle travel, This is a function for stopping the operation of the valve and the fuel injection.

  As shown in FIG. 1, this active noise control apparatus 10 basically includes a reference signal generator 12 that generates a harmonic reference signal from a frequency f of noise Nz generated from an engine 28 that is a noise source. And an adaptive notch filter 14 for outputting a control signal y (n) for canceling out the noise Nz at a time point n every sampling period. The adaptive notch filter 14 is a variable on the frequency domain in the noise frequency, updated based on a noise signal at regular intervals on the time axis, and converted to real-time data based on the adaptive notch filter variable and the reference signal at the time of output. Is.

  The active noise control device 10 further detects a speaker 16 which is a sound output device that outputs a control signal y (n) as a control sound, and an offset error sound between the noise Nz from the engine 28 and the control sound from the speaker 16. And a microphone 18 as a sound detector to be output as an error signal e (n), and a transfer function H of a sound field from the position of the speaker 16 to the position of the microphone 18, and the reference signal is input to output a reference signal. And a filter coefficient updating means (LMS algorithm computing unit) 22 that is supplied with the error signal e (n) and the reference signal and updates the filter coefficient W (n + 1) of the adaptive notch filter 14. The

  The filter coefficient updating unit 22 includes a first filter coefficient updating unit 22A and a second filter coefficient updating unit 22B.

  As schematically shown in FIG. 2, the active noise control device 10 is actually arranged and fixed under the dashboard, and detects the rotation of the main shaft of the engine 28 mounted on the chassis under the bonnet. The engine rotation pulse Ep and the error signal e (n) from the microphone 18 fixed to the roof lining on the driver's seat are input from the sensor, and the control sound is output from the speaker 16 disposed under the driver's seat. It is set as such. In this embodiment, the noise control only for the driver's seat will be described for easy understanding, but the present invention can be similarly applied to other seats such as a passenger seat and a rear seat.

  In FIG. 1, the frequency f of the noise Nz is detected by the frequency counter 32 from the engine rotation pulse Ep generated by detecting the rotation of the crankshaft (rotary body) of the engine 28, and the reference signal generator 12 and the reference signal generation circuit. 20 is supplied.

  The reference signal generator 12 generates a cosine wave cos (2π (f, n)), which is a harmonic reference signal, from the frequency f of the noise Nz, and a harmonic of the frequency f of the noise Nz. The sine wave generator 36 generates a sine wave sin (2π (f, n)) which is a reference signal. The magnitudes of the cosine wave generated by the cosine wave generator 34 and the sine wave generated by the sine wave generator 36 are each “1”.

  The adaptive notch filter 14 includes a limit processing means 13, a first pre-filter 14A, a second pre-filter 14B, a first adaptive notch filter 15A, a second adaptive notch filter 15B, and an adder 38. Yes.

  The first pre-stage filter 14A holds the filter coefficient (real part signal) Wx based on the signal supplied from the first filter coefficient update unit 22A, and supplies the latest filter coefficient Wx to the limit processing unit 13. Similarly, the second pre-stage filter 14B holds the filter coefficient (imaginary part signal) Wy based on the signal supplied from the second filter coefficient update means 22B and supplies the latest filter coefficient Wy to the limit processing means 13. .

  The limit processing means 13 calculates the sum of squares of the filter coefficient Wx and the filter coefficient Wy and its square root value, and performs limit processing described later. The limit processing means 13 can easily obtain the square root value and its inverse by referring to the square root table 13a in which the square root corresponding to the argument is recorded and the square root reciprocal number table 13b in which the inverse of the square root corresponding to the argument is recorded. it can.

  As shown in FIG. 3, the square root table 13a is a table indicating a function f (x) = √ (x), and returns a square root value based on an argument x. As shown in FIG. 4, the square root reciprocal number table 13b is a table indicating the function f (x) = 1 / √ (x), and returns an inverse value of the square root based on the argument x. Since these square root value and inverse square root value are calculated and recorded in advance, there is no need to perform another calculation, and the result is obtained instantaneously using an array or an address pointer. The square root table 13 a and the square root reciprocal table 13 b are recorded in a predetermined ROM of the storage unit in the microcomputer 1.

  Returning to FIG. 1, when the square root value of the square sum of the filter coefficient Wx and the filter coefficient Wy is larger than the threshold value Aobj, the limit processing means 13 multiplies the filter coefficient Wx by the ratio of the square root value and the threshold value Aobj. The limiting filter coefficient Wxo limited by is output. Further, the limit processing means 13 is limited by multiplying the filter coefficient Wy by the ratio of the square root value and the threshold value Aobj when the square root value of the square sum of the filter coefficient Wx and the filter coefficient Wy is larger than the threshold value Aobj. The limited filter coefficient Wyo is output.

  The threshold value Aobj is set by the threshold value switching means 13c, and the initial value is set to the maximum value of the data expression format of the filter coefficient Wx and the filter coefficient Wy.

  The first adaptive notch filter 15A receives a cosine wave cos (2π (f, n)) having a magnitude 1 and a limiting filter coefficient Wxo (n), and outputs a first control signal y1 (n) obtained by multiplying them. . The second adaptive notch filter 15B receives a sine wave sin (2π (f, n)) having a magnitude of 1 and a limiting filter coefficient Wyo (n), and outputs a second control signal y2 (n) obtained by multiplying them. .

  The adder 38 vector-adds the first control signal y1 (n) and the second control signal y2 (n) to generate a control signal y (n) having an arbitrary phase and amplitude. A control signal y (n) which is a digital signal is supplied to the speaker 16 via the D / A converter 211, the LPF (low-pass filter) 212, and the amplifier 213, and is output as a control sound via the speaker 16. The Note that the speaker 16 may also be used for audio by adding the output of the D / A converter 211 and the audio signal by a predetermined mixer.

  The reference signal generation circuit 20 includes four correction filters 41 to 44 and adders 46 and 48.

  The correction filters 41 and 43 have the characteristic part ReH (f) of the transfer function H of the sound field including the position of the microphone 18 from the position of the speaker 16, and the correction filters 42 and 44 are imaginary numbers of the transfer function H. Part characteristic ImH (f).

  The transfer function H is a signal transfer function from the position of the speaker 16 to the position of the microphone 18 in the passenger compartment. For the actual measurement of the transfer function, for example, a signal transfer characteristic measuring device comprising a Fourier transform device is activated. Connected between the input side of the D / A converter 211 (the output side of the adder 38) and the output side of the A / D converter 203 (the input side of the filter coefficient updating means 22) constituting the noise control apparatus 10. With this signal transfer characteristic measuring device, the signal transfer function is converted from the control signal y (n) output from the microcomputer 1 to the input side of the D / A converter 211 and from the microphone 18 through the A / D converter 203. It is measured based on the error signal e (n) input to the microcomputer 1.

  Therefore, according to the signal transfer function measurement method, the signal transfer function between the speaker 16 and the microphone 18 in the passenger compartment is converted into an analog electronic circuit inserted between the output and the input of the microcomputer 1, for example, In addition, transfer characteristics of the speaker 16, the microphone 18, the D / A converter 211, the LPF 212, the amplifier 213, the amplifier 201, the BPF 202, and the A / D converter 203 are also included.

  In other words, depending on the measurement method of the signal transfer characteristic, the transfer function H of the signal between the speaker 16 and the microphone 18 in the passenger compartment is the signal transfer characteristic from the output of the adaptive notch filter 14 to the input of the filter coefficient updating means 22. It becomes.

  Further, the characteristic values of the real part characteristic ReH (f) and the imaginary part ImH (f) change depending on the frequency f.

  A reference signal Cx (n) related to the cosine wave cos (2π (f, n)) is output from the adder 46 to the first filter coefficient update means 22A, and a sine wave sin (2π (f, n)) is output from the adder 48. The reference signal Cy (n) related to is output to the second filter coefficient updating means 22B.

  It can be seen that the reference signals Cx (n) and Cy (n) can be obtained by the following equations by referring to the circuit connection of the reference signal generation circuit 20.

Cx (n) = cos (2π (f, n)) · ReH (f)
-Sin (2π (f, n)) · ImH (f) (1)
Cy (n) = cos (2π (f, n)) · ImH (f)
+ Sin (2π (f, n)) · ReH (f) (2)

  Note that when both or one of the reference signals Cx (n) and Cy (n) is indicated, it is referred to as a reference signal C (n).

  In this case, the error signal e (n) and the reference signals Cx (n) and Cy (n) are input to the first filter coefficient updating unit 22A and the second filter coefficient updating unit 22B, respectively, and the error signal e ( The filter coefficient W (n) is sequentially updated by an adaptive algorithm (LMS algorithm) every time point n so that n) is minimized.

  That is, the filter coefficients Wx (n + 1) and Wy (n + 1) at the next sampling after the sampling period has elapsed from the sampling timing of the filter coefficients Wx (n) and Wy (n) are expressed by the following expressions (3) and (4). Ask based on.

Wy (n + 1) = Wy (n) −μ · e (n) · ry (n)
= Wy (n)-[mu] .e (n). (Cx.cos (2 [pi] ft (n) -Cy.sin (2 [pi] ft (n))) (3)
Wx (n + 1) = Wx (n) + μ · e (n) · rx (n)
= Wx (n) + μ · e (n) · (Cx · sin (2πft (n) + Cy · cos (2πft (n))) (4)

  Here, Wx (n + 1) and Wy (n + 1) indicate updated adaptive notch filter coefficients, and Wx (n) and Wy (n) indicate adaptive notch filter coefficients immediately before updating. Further, e (n) indicates an error signal immediately before update, r0 (n) and r1 (n) indicate addition output signals immediately before update, and μ indicates an update coefficient. Further, μ · e (n) · r0 (n) and μ · e (n) · r1 (n) indicate the update amount of the adaptive notch filter coefficient.

  Next, the operation procedure of the active noise control apparatus 10 configured as described above will be described with reference to FIG. The processing of FIG. 5 is continuously performed by the microcomputer 1 every minute time.

  In step S <b> 1, the error signal e is input from the microphone 18 and is input as a digital signal through the A / D converter 203.

  In step S 2, the reference frequency f of the noise Nz is detected from the engine rotation pulse Ep by the frequency counter 32 and supplied to the reference signal generator 12 and the reference signal generation circuit 20.

  In step S3, the cosine wave generator 34 and the sine wave generator 36 generate a sine wave sin (2π (f, n)) and a cosine wave cos (2π (f, n)), and the adaptive notch filter 14 and the reference signal. This is supplied to the generation circuit 20.

  In step S4, the filter coefficient updating means 22 reads the real part characteristic ReH (f) and the imaginary part characteristic ImH (f) of the transfer function H according to the frequency f serving as a control reference from the reference signal generation circuit 20. .

  In step S5, the filter coefficient updating means 22 updates the filter coefficient based on the above equations (3) and (4).

  In step S6, the adaptive filter limit process in the limit processing means 13 is performed.

  In step S7, the adaptive notch filter 14 performs a filtering process to obtain a control signal y (n). Details of the processes in steps S6 and S7 will be described later.

  In step S8, the control signal (n) is converted into an analog signal by the D / A converter 211, and then supplied to the speaker 16 via the LPF (low-pass filter) 212 and the amplifier 213, and output as a control sound. . Thereby, the noise transmitted from the engine 28 to the vehicle interior is effectively canceled.

  Next, the operation of the limit processing means 13 and the adaptive notch filter 14 in steps S6 and S7 will be described in detail with reference to FIG. In the following description, the process of step S6 and the process of step S7 will be described with no particular distinction in order to facilitate understanding.

  First, in step S101, the first pre-filter 14A updates the filter coefficient Wx supplied from the first filter coefficient update unit 22A to the latest one and supplies it to the limit processing unit 13. Further, the second pre-stage filter 14B updates the filter coefficient Wy supplied from the second filter coefficient update unit 22B to the latest one and supplies it to the limit processing unit 13.

  In step S102, the limit processing means 13 obtains a value “Anow” obtained by summing the square and root of the filter coefficient Wx and the filter coefficient Wy based on the following equation (5). The magnitude of this value Anow is set to such an extent that the noise Nz can be canceled.

Now = √ (Wx ² + Wy ² ) (5)

As shown in FIG. 7, “Anow” as a vector is expressed in a square complex plane region 90 specified by Wx and Wy orthogonal to each other with the origin O as a midpoint. The maximum values of Wx and Wy are equal to the threshold value Aobj in both the plus direction and the minus direction. Since the filter coefficient Wx and the filter coefficient Wy are expressed in a fixed-point format, for example, if the integer part (the number of remaining bits excluding the decimal part and the sign part) is 12 bits, the threshold value Aobj is Aobj = 4097. = 4096 (= 2 ¹² ) +1. The “+1” part is the maximum value represented by the decimal part.

  By the way, since the filter coefficient Wx and the filter coefficient Wy are 0 at the minimum of the absolute value and the threshold value Aobj at the maximum of the absolute value, for example, when the phase θ is θ = 0, Wx = Aobj and Wy = 0. The magnitude of the vector Now becomes | Anow | = Aobj, but as shown in FIG. 7, when the phase θ advances and θ = 45 °, Wx = Aobj and Wy = Aobj, the magnitude of the vector Now becomes | Anow | = √2 · Aobj. That is, when the original filter coefficient Wx and the filter coefficient Wy are large, “Anow” as a vector rotates along the four sides of the square complex plane region 90, and the magnitude of “Anow” as a vector varies. Will end up. For this reason, the following restriction process is performed.

For the calculation of √ (square root) in equation (5), the value of Now can be obtained by obtaining the value of Wx ² + Wy ² and referring to the square root table 13a using this value as an argument.

  In step S103, the magnitude of Anow is compared with the threshold value Aobj, and if Anow> Aobj, the process proceeds to step S104. If Anow ≦ threshold Aobj, the process proceeds to step S106 without performing the limiting process. In this case, the limiting filter coefficients Wxo and Wyo are set as Wxo = Wx and Wyo = Wy.

  In step S104, the limit processing means 13 obtains the ratio Rate used to limit the filter coefficient Wx and the filter coefficient Wy by the following equation (6).

Rate = Aobj / Anow = Aobj / √ (Wx ² + Wy ² ) (6)

In the calculation based on the equation (6), the division by Anow may be multiplied by the inverse of Anow. The reciprocal number of Now can be easily obtained by referring to the square root reciprocal table 13b with the value of Wx ² + Wy ² obtained in step S103 as an argument. Further, if the comparison between the magnitude of Annow and the threshold value Aobj in step S103 is determined by comparing the reciprocals, the square root table 13a may be omitted and only the square root reciprocal table 13b may be used.

  In step S105, limited filter coefficients Wxo and Wyo are obtained by limiting the filter coefficient Wx and the filter coefficient Wy based on the following expressions (7) and (8).

Wxo = Wx · Rate
= Wx · Aobj / Anow
= Wx · Aobj / √ (Wx ² + Wy ² ) (7)
Wyo = Wy · Rate
= Wy · Aobj / Anow
= Wy · Aobj / √ (Wx ² + Wy ² ) (8)

According to the equations (7) and (8), when the limiting filter coefficients Wxo and Wyo are added in vector, the magnitude becomes √ (Wxo ² + Wyo ² ) = Aobj, and is limited by the threshold value Aobj. For example, when the filter coefficient Wx and the filter coefficient Wy are Wx = Wy = Aobj, Now = √2 · Aobj, and Wxo = Wyo = Aobj / √2, and Wxo and Wyo are added in a vector form. It will be appreciated that the magnitude is equal to the threshold Aobj. As a result, the absolute amplitude of Wx (n) + i · Wy (n), (i is an imaginary number) is limited.

  The obtained limiting filter coefficient Wxo is supplied to the first adaptive notch filter 15A, and the limiting filter coefficient Wyo is supplied to the second adaptive notch filter 15B.

  In step S106, the first adaptive notch filter 15A adds the first control signal y1 (n) obtained by multiplying the cosine wave cos (2π (f, n)) and the limiting filter coefficient Wxo (n) to the adder 38. Supply. The first adaptive notch filter 15B supplies the adder 38 with the second control signal y2 (n) obtained by multiplying the sine wave sin (2π (f, n)) and the limiting filter coefficient Wyo (n). .

  In step S107, the adder 38 vector-adds the first control signal y1 (n) and the second control signal y2 (n) to generate a control signal y (n). Thereafter, the control signal y (n) generated by the process of step S8 is output from the speaker 16 via the D / A converter 211, the LPF 212, and the amplifier 213, and the noise Nz is suppressed.

  By the way, since the magnitude of the control signal y (n) is equal to the square root of the square sum of the first control signal y1 (n) and the second control signal y2 (n), the maximum value is equal to the threshold value Aobj. When represented on the complex plane area 90 with the origin as the center, the circle 90 rotates in a circle while maintaining its size constant regardless of the four sides forming the area 90. Therefore, the control performance is maintained without superimposing other unnecessary frequencies on the control frequency, and the convergence for canceling out the noise Nz is fast.

  As described above, in the active noise control apparatus 10 according to the present embodiment, the square root value of the square sum of the filter coefficient Wx that is the real part signal of the parameter and the filter coefficient Wy that is the imaginary part signal in the adaptive notch filter 14. By limiting Anow with the threshold value Aobj, the parameter rotates with an amplitude based on the threshold value Aobj with the origin on the complex plane as the center, preventing unnecessary fluctuations in the control output of the adaptive notch filter 14. it can. Thereby, the convergence property of noise is high and it becomes difficult to diverge.

  In addition, the adaptive notch filter 14 can easily and quickly calculate the square root value and its reciprocal by using the square root table 13a and the square root inverse table 13b. The reciprocal of the square root can be obtained with the type or integer type.

Further, since the filter coefficient Wx and the threshold value Aobj for limiting the filter coefficient Wy are set to the maximum value of the data expression format (for example, 2 ¹² + 1 = 4097), the frequency with which the parameters are limited is reduced as much as possible. Nz can be efficiently suppressed. On the other hand, the threshold Aobj may be switched to the maximum value of the data expression format such as the filter coefficients Wx and Wy according to the conditions.

  Next, threshold value Aobj switching processing performed by the threshold value switching means 13c will be described with reference to FIG. The threshold switching means 13c changes the threshold Aobj based on the combination of the rotational speed N of the engine 28 and the number of cylinders s operating (or the number of cylinders operating) among the plurality of cylinders of the engine 28.

  First, in step S201, the threshold switching means 13c obtains the rotational speed N from the frequency f obtained from the frequency counter 32, and from the signal supplied from the engine controller 28a, the number of cylinders s operating (that is, four cylinders). From the above, the cylinder deactivation function is used to obtain the deactivation number (excluding the number of cylinders).

  It is assumed that the engine 28 has four cylinders and the cylinder deactivation function deactivates three cylinders according to operating conditions. That is, the number of operating cylinders s is 4 or 1.

  In step S202, it is confirmed whether or not the all cylinder operation is performed based on the number of cylinders s. When the all cylinder operation is performed, the process proceeds to step S203, and when the cylinder is deactivated, the process proceeds to step S208. That is, when s = 4, the process proceeds to step S203, and when s = 1, the process proceeds to step S208.

  In step S203, it is confirmed whether the rotation speed N is less than the first threshold value N1. When N <N1, the process proceeds to step S205, and when N ≧ N1, the process proceeds to step S204.

  In step S204, it is confirmed whether or not the rotation speed N is less than the second threshold value N2. When N <N2, the process proceeds to step S206, and when N ≧ N2, the process proceeds to step S207.

  In step S205, the threshold setting parameter T is set as T = T1. In step S206, T = T2 is set, and in step S207, T = T3 is set. After steps S205 to S207, the process proceeds to step S213.

  On the other hand, in step S208, it is confirmed whether or not the rotational speed N is less than the third threshold value N3. When N <N3, the process proceeds to step S210, and when N ≧ N3, the process proceeds to step S209.

  In step S209, it is confirmed whether or not the rotation speed N is less than the fourth threshold value N4. When N <N4, the process proceeds to step S211, and when N ≧ N4, the process proceeds to step S212.

  In step S210, T is set as T = T4, T = T5 is set in step S211, and T = T6 is set in step S212. After steps S210 to S212, the process proceeds to step S213.

  In step S213, the value of the threshold Aobj is switched to the parameter T. This switching is performed step by step for every minute time and gradually changed. That is, when the threshold value Aobj is made smaller than the previous value, the minute value is subtracted every minute time until reaching the same value as the parameter T, or a constant smaller than “1” (for example, 0.9) is continued. Further, when the threshold value Aobj is made larger than the previous value, the minute value is added every minute time until reaching the same value as the parameter T, or a constant larger than “1” (for example, 1.1) is continued. Thus, by changing the threshold value Aobj stepwise, a sudden change in the control signal y (n) is prevented and fade-out processing is performed, and the generation of a “sound” (so-called pop-up noise) can be prevented.

  The values T1 to T6 set in the threshold setting parameter T are set to values equal to or less than the maximum value of the data expression format such as the filter coefficients Wx and Wy. At least one of the values T1 to T6 is set to the same value (for example, 4097) as the maximum value of the data expression format such as the filter coefficients Wx and Wy, and the threshold value Aobj is set to a default value (for example, 4097) under a predetermined traveling condition. Return.

  Further, at least one of the values T1 to T6 is “0”, and acts to stop the output of the control signal y (n). As described above, the output of the control signal y (n) may be stopped in order to suppress the occurrence of unnecessary fluctuations according to the operating conditions.

  The magnitude of the noise Nz and the peaks or valleys or phases of the sound field vary depending on the rotational speed N of the engine 28, which is a noise source, the control reference frequency f, and the number of cylinders s being operated. According to this switching process, the threshold value Aobj can be appropriately switched based on the combination of the rotational speed N and the cylinder number s.

  Further, according to the processing by the threshold value switching means 13c as shown in FIG. 8, the limit value is set in the vehicle running state (the rotational speed N, the driving speed) in the amplitude limit that is performed while the phase change is abrupt and the control is unstable in the acoustic transfer characteristic. The frequency is switched depending on the number of cylinders), and the instability is eliminated by reducing the limit value at frequencies where the phase change in the acoustic transfer characteristics is sudden and the control becomes unstable. By increasing the value, it is possible to adjust the control performance, such as making the best use of the silencing capability. The threshold switching parameter may be switched based on, for example, the vehicle speed (or a combination of the vehicle speed and the rotational speed N and the operating cylinder number s) in addition to the rotational speed N and the operating cylinder number s. The vehicle speed is determined based on, for example, the rotational speed N and a gear ratio in a transmission (not shown).

  Next, the active vibration control apparatus 100 according to the present embodiment will be described with reference to FIGS.

  The active vibration control device 100 uses a self-expanding engine mount 102 to cancel the vibration of the engine 28. Further, noise can be reduced by canceling vibration of the engine 28. In the active vibration control device 100, the same components as those in the active noise control device 10 are denoted by the same reference numerals, and detailed description thereof is omitted. In general, a plurality of engine mounts 102 are provided, but the engine 28 is assumed to be supported by a single engine mount 102 for easy understanding.

  As shown in FIGS. 9 and 10, the active vibration control apparatus 100 cancels vibrations with an adaptive notch filter 14 in the microcomputer 1 based on the frequency f of vibration generated from the engine 28 as a vibration source. A control signal y (n) is generated. The control signal y (n) is driven to extend and retract the engine mount 102 via the D / A converter 211, the LPF 212, and the amplifier 213, thereby canceling the vibration of the engine 28. In this case, if the phase for expanding and contracting the engine mount 102 and the phase of the vibration of the engine 28 are the same, the vibration is absorbed, and the vibration is easily canceled.

  As the engine mount 102, for example, a self-expandable mount described in Patent Document 3 may be used. In addition to the microcomputer 1, the engine mount 102 performs an expansion / contraction operation under the action of a predetermined ECU in order to cancel vibration received from the road surface, and the output of the microcomputer 1 and the output of the ECU are added together by a mixer. May be driven.

  The engine 28 is provided with a vibration sensor 104 that detects vibration generated as an offset error, and the vibration sensor 104 detects vibration in the same direction as the direction of expansion and contraction of the engine mount 102. As the vibration sensor 104, for example, a displacement sensor, an acceleration sensor, a load cell, or the like can be used. The error vibration detected by the vibration sensor 104 is converted into an error signal e (n) which is a digital signal via the amplifier 201, the BPF 202 and the A / D converter 203, and the first filter coefficient updating means 22A and the second filter coefficient updating means 22A and second This is supplied to the filter coefficient updating means 22B.

  That is, in the active vibration control apparatus 100, the speaker 16 and the microphone 18 in the active noise control apparatus 10 are replaced with the engine mount 102 and the vibration sensor 104. Further, the transfer function H is replaced as a transfer characteristic of the gain and phase of vibration between the engine mount 102 and the vibration sensor 104 (or between D / A 211 and A / D 233). In the microcomputer 1, the reference signal generator 12, the adaptive notch filter 14, the reference signal generation circuit 20, the first filter coefficient update means 22A and the second filter coefficient update means 22B perform the above processing.

  In the active vibration control device 100 as well, as with the active noise control device 10 described above, the limit processing means 13 in the adaptive notch filter 14 uses the filter coefficient Wx, which is a parameter related to the real part, and the parameter related to the imaginary part. When the square root of the sum of squares of a certain filter coefficient Wy (that is, the absolute amplitude of Wx (n) + i · Wy (n)) is larger than the threshold value Aobj, these filter coefficient Wx and filter coefficient Wy are multiplied by the ratio Rate. The restriction process is performed. Therefore, by limiting the square root value of the square sum of the real part signal and the imaginary part signal of the adaptive notch filter 14 with the threshold value Aobj, the synthesized vector is rotated with the amplitude based on the threshold value around the origin on the complex plane. Therefore, it is possible to prevent unnecessary fluctuations from occurring in the control output y (n) of the adaptive notch filter 14. Thereby, the convergence of vibration becomes high and it becomes difficult to diverge. Note that the control target for canceling the vibration by the active vibration control device 100 is not limited to the engine 28 but may be, for example, a pipe of an intake / exhaust system.

  It goes without saying that the active noise control device and the active vibration control device according to the present invention are not limited to the above-described embodiments, and can adopt various configurations without departing from the gist of the present invention.

It is a block block diagram of the active noise control apparatus which concerns on this Embodiment. 1 is a schematic plan view of a vehicle equipped with an active noise control device. It is a graph which shows the contents of a square root table. It is a graph which shows the contents of a parallel root reciprocal number table. It is a flowchart which shows the noise control procedure by an active noise control apparatus. It is a flowchart which shows the procedure of the limit process in an adaptive notch filter. It is a figure which shows the area | region restrict | limited to the square by the parameter expression format on the complex plane. It is a flowchart which shows the switching process procedure of the threshold value performed by a threshold value switching means. It is a block block diagram of the active vibration control apparatus which concerns on this Embodiment. 1 is a schematic side view of a vehicle equipped with an active noise control device.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 ... Active noise control apparatus 12 ... Reference signal generator 13 ... Limit processing means 13a ... Square root table 13b ... Square root reciprocal table 13c ... Threshold switching means 14 ... Adaptive notch filter 14A ... First pre-filter 14B ... Second pre-filter 15A First adaptive notch filter 15B Second adaptive notch filter 16 Speaker 18 Microphone 20 Reference signal generation circuit 22, 22A, 22B Filter coefficient updating means 28 Engine 30 Vehicle 32 Frequency counter 34 Cosine wave generation 36: Sine wave generator 38, 46, 48 ... Adders 41-44 ... Correction filter 90 ... Square area 100 ... Active vibration controller 102 ... Engine mount 104 ... Vibration sensor Ep ... Engine rotation pulse H ... Transfer function N ... Number of revolutions Nz ... Noise Wx ... Filter coefficient (real part) No.) Wy ... filter coefficient (imaginary part signal)
e ... error signal f ... frequency y ... control signal

Claims (10)

A reference signal generator for outputting a harmonic reference signal from the frequency of noise generated from a noise source;
An adaptive notch filter that receives the reference signal and outputs a control signal for canceling the noise;
A sound output device for outputting the control signal as a control sound;
A sound detector that detects an offset error between the noise and the control sound and outputs an error signal;
A correction filter having a transfer function from the sound output device to the sound detector, and receiving the reference signal and outputting a reference signal;
Filter coefficient updating means for sequentially updating filter coefficients of the adaptive notch filter so that the error signal is minimized when the error signal and the reference signal are input;
The filter coefficient is expressed in a frequency domain divided into a real part signal and an imaginary part signal, and when the square root value of the square sum of the real part signal and the imaginary part signal is larger than a predetermined threshold, the square root value and the Limit processing means for limiting the real part signal and the imaginary part signal by multiplying by a ratio with a predetermined threshold;
An active noise control device comprising: The active noise control device according to claim 1,
The noise source includes a rotating body,
The active noise control device, wherein the reference signal generator identifies harmonics of the noise frequency based on the number of rotations of the rotating body. The active noise control device according to claim 1 or 2,
A square root table in which the square root corresponding to the argument is recorded,
The adaptive notch filter obtains the square root value by referring to the square root table with a square sum of the real part signal and the imaginary part signal as an argument. The active noise control device according to claim 1 or 2,
A square root reciprocal table in which the reciprocal of the square root corresponding to the argument is recorded;
The adaptive notch filter, wherein the reciprocal number is obtained by referring to the square root reciprocal table with a square sum of the real part signal and the imaginary part signal as an argument. In the active noise control device according to any one of claims 1 to 4,
An active noise control apparatus, wherein the predetermined threshold value is “0”, and the output of the control signal is stopped. The active noise control device according to claim 5,
An active noise control device, wherein the square root is limited stepwise by the predetermined threshold. The active noise control device according to claim 2,
The adaptive noise control device, wherein the adaptive notch filter changes the predetermined threshold based on the rotation speed. The active noise control device according to claim 1 or 2,
The noise source is a multi-cylinder internal combustion engine,
A cylinder deactivation function unit for stopping the operation of at least one of the plurality of cylinders;
The adaptive noise control device, wherein the adaptive notch filter changes the predetermined threshold based on the number of cylinders in operation. A reference signal generator that outputs a harmonic reference signal from the frequency of vibration generated from the vibration source;
An adaptive notch filter that receives the reference signal and outputs a control signal for canceling the vibration;
An actuator for outputting the control signal as a control vibration;
A vibration detector that detects an offset error between the vibration and the control vibration and outputs an error signal;
A correction filter that has a transfer function from the actuator to the vibration detector, and that receives the reference signal and outputs a reference signal;
Filter coefficient updating means for sequentially updating filter coefficients of the adaptive notch filter so that the error signal is minimized when the error signal and the reference signal are input;
The filter coefficient is expressed in a frequency domain divided into a real part signal and an imaginary part signal, and when the square root value of the square sum of the real part signal and the imaginary part signal is larger than a predetermined threshold, the square root value and the Limit processing means for limiting the real part signal and the imaginary part signal by multiplying a ratio with a predetermined threshold;
An active vibration control device comprising: The active vibration control device according to claim 9,
The vibration source includes a rotating body,
The active vibration control device characterized in that the reference signal generator specifies harmonics of the vibration frequency based on the number of rotations of the rotating body.

Images