JPH07133842A - Active vibration control device - Google Patents

Abstract

PURPOSE:To reduce memory quantity and computation time by carrying out DFT in accordance with an oscillation cycle desired to erase, finding a system transfer constant at its frequency and correcting an IDFT coefficient to compute a control output. CONSTITUTION:Oscillation of an engine O is transmitted to a point of evaluation as d(n), and a reference signal X(n) which is high in correlation to periodic oscillation is provided by a reference signal generation means 9. A reference sine wave generation means 10 outputs sine and cosine waves with a reference signal as a trigger, and a cycle is made to correspond to a frequency of oscillation desired to erase. A coefficient value of the sine wave corresponding with a frequency is in an IDFT coefficient memory 1, and a sum of products of the reference sine wave and its coefficient is output to an actuator 3 by way of computing it by an IDFT computation means 2. This device is made to actuate to eliminate residual oscillation e(n) as a controller, but in consideration of a transmission delay from an output point of the computation means 2 to the point of evaluation, the coefficient of the coefficient memory 1 is corrected by a coefficient correction means 8. Consequently, it is possible to control oscillation by way of reducing computation time and memory quantity.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention positively controls the noise on the left by a vibration source for control or a sound source for control when noise synchronized with the engine is remarkable, such as vibration (or noise) in the passenger compartment of an automobile. The present invention relates to an active vibration control device used for mechanically driving and erasing.

[0002]

2. Description of the Related Art In recent years, active vibration control devices have attracted attention in the fields of automobiles, aircrafts, etc. as control devices for reducing vibrations such as mechanical noise and noise while adapting to the surrounding environment.

An example of the above-described conventional active vibration control device will be described below with reference to the drawings.

FIG. 3 is a schematic view of a conventional active vibration control device. In FIG. 3, reference numeral 30 is an automobile engine, 31 is an FIR filter coefficient memory, and 32 is FI.
R filter product-sum means, 33 is an actuator, 34 is a system delay, 35 is a system transfer characteristic memory, 36 is a system transfer characteristic product-sum means, 37 is residual vibration detection means, 3
Reference numeral 8 is a coefficient correcting means, and 39 is a reference signal generating means.

The operation of the active vibration control device configured as described above will be described below.

[0006] The engine 30 of an automobile is considered to be a rotary device that converts energy during fuel combustion into rotary motion. However, due to temporal discontinuity when extracting energy from combustion and imbalance of a rotating object, it has periodicity. Strong vibration and noise are generated. Now, the vibration is d (n) at the evaluation point.
Will be transmitted as.

On the other hand, a reference signal (impulse signal sequence) X (n) having a high correlation with the periodic vibration can be obtained from the engine 30 by the reference signal generating means 39. For example, by cutting N teeth on the engine shaft and detecting the behavior of the teeth by a Hall element provided outside the shaft, N impulse trains can be generated for one rotation of the engine shaft.

Next, an FIR filter having a delay time of the FIR filter coefficient memory 31 is constructed, and the product sum of the FIR filter product sum means 32 and the reference signal X (n) obtained by the reference signal generation means 39 is calculated. To do. The output of the sum of products is output to the actuator 33 as a control output.

Now, the active vibration control device controls the difference between d (n) and the control output y (n) at the evaluation point so as to reduce the engine vibration d (n) at the evaluation point, and the residual. The operation is performed so as to eliminate the vibration e (n), but from the output point of the FIR filter sum-of-products means 32 to the evaluation point, for example,
There is a time delay and a phase delay due to the actuator 33 and the system delay 34. (For example, an actuator composed of a spring and a damper can be approximated to a second-order low-pass filter). Therefore, the coefficient correction means 38 corrects each coefficient of the FIR filter coefficient memory 31 so as to minimize the error detected by the residual vibration detection means 37 while considering the system transmission delay. An example of this modified algorithm is the Filtered-X LMS algorithm. Fi
The lterd-XLMS algorithm is the filter coefficient at discrete time n Wk (n) (the number of filter taps is m), the residual vibration (error) e (n), and the system transfer characteristic expressed by FIR as C ( n), where X (n) is the reference signal vector at time n consisting of the past m elements, and μ is the step size parameter (constant), this is an algorithm that corrects the filter coefficient by Equation 1.

[0010]

[Equation 1]

(For example, EA, Acoustical Society of Japan Material EA
90-2 Realization of active noise control chair (see Haruo Hamada et al.) In order to realize the above expression 1, the transfer characteristic from the output point of the FIR filter product-sum means 32 to the evaluation point is known in advance by system identification. The FIR filter having the number of taps P that can sufficiently include the characteristic is generated and stored in the system transfer characteristic memory 35. Next, the system transfer characteristic sum of products means 36 calculates the harmony between the system transfer characteristic and the reference signal X (n).

By the above method, it is possible to construct an active vibration control device capable of reducing engine vibration at an evaluation point even if there is a system transmission delay.

[0013]

However, in the above-mentioned configuration, in order to realize flexible control in which the control system follows fluctuations in the engine speed of the automobile, the reference signal is fetched without error for control calculation. FIR filter 1
The delay time must be small and therefore the FI
The number of taps of the R filter increases. In particular, the system transfer characteristic configured in the system transfer characteristic memory increases the number of taps when trying to cope with a low frequency. (For example, if the impulse response of 0.1 sec is fs =
If you try to realize a FIR filter at 1000Hz,
Requires 100 taps).

Then, there are problems that it takes time to calculate the sum of products, it is difficult to expand the control band, the calculation load is large, and the memory amount increases.

Alternatively, a configuration may be considered in which only the periodic signal is efficiently dropped by matching the delay time of the filter with the impulse period of the reference signal. In this case, an FIR filter (1 delay representing the system transfer characteristic that has been set in advance is used. (Time is determined by sampling during system identification), but the system transfer characteristic filter corresponding to the current engine vibration frequency must be calculated by sampling conversion, but this decimation process takes time, which also increases the calculation time. Connect

In view of the above problems, the present invention provides an active vibration control system which has a small calculation load and can efficiently reduce engine vibration.

[0017]

In order to solve the above problems, the active vibration control system of the present invention focuses on the fact that the engine vibration has a high periodicity. The system transfer characteristic is extracted by FFT only for the harmonic component, the filter error is corrected as it is in the frequency domain, and finally IDFT is performed to calculate the output to obtain the control output.

[0018]

According to the present invention, with the above configuration, the configuration of the control system is simplified, the calculation time is short, and the vibration at the evaluation point can be actively reduced.

[0019]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An active vibration control system according to a first embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 is a schematic view of an active vibration control system in an embodiment of the invention of claim 1 of the present invention. In FIG. 1, 0 is an automobile engine and 1 is an IDF.
T (inverse discrete Fourier transform) coefficient memory, 2 is IDFT
Calculation means, 3 is an actuator, 4 is a system delay, 5
Is a system transfer characteristic memory, 6 is a DFT (discrete Fourier transform) calculation means, 7 is residual vibration detection means, 8 is coefficient correction means, 9 is reference signal generation means, 10 is reference sine wave generation means, and 11 is a measurement window. It is a calculation means.

Next, the operation of the active vibration control device will be described below with reference to FIG. Car engine 0
Is considered to be a rotating device that converts energy during fuel combustion into rotary motion. However, due to temporal discontinuity when extracting energy from combustion and unbalance of rotating objects, strong periodic vibration and noise are generated. . Now, it is assumed that the vibration is transmitted to the evaluation point as d (n).

On the other hand, the reference signal (impulse signal train) Xn having a high correlation with the periodic vibration can be obtained from the engine 0 by the reference signal generating means 9. For example, by cutting N teeth on the engine shaft and detecting the behavior of the teeth by a Hall element provided outside the shaft, N impulse trains can be generated for one rotation of the engine shaft.

The reference sine wave generating means 10 outputs a sine wave and a cosine wave by using the impulse train of the reference signal as a trigger. The cycle corresponds to the frequency of the vibration you want to erase. Now, since it is desired to eliminate the vibration synchronized with the engine vibration, its cycle is N, N / 2, as the number of the impulse trains.
N / 3, ..., N / k (k is an integer). For the sake of explanation, the vibrations to be eliminated will be described below in the order of primary, secondary, tertiary,
Let k order.

The IDFT coefficient memory 1 stores coefficient values of sinusoidal waves sin and cos corresponding to the frequencies of the above orders. Therefore, the IDFT calculation means 2 calculates the sum of products of the reference sine wave and its coefficient as shown in Equation 2 to obtain the control output. Just this is IDFT (Invert Discrete Fo
The IDFT calculation means 2 can be used because it corresponds to a urier transform (ie, an inverse discrete Fourier transform). Actuator 4 with the output of the product sum as the control output
Output to 3. As is clear from the expression (2), in the active vibration control device of the present invention, only two taps are required per vibration of one frequency.

[0025]

[Equation 2]

However, S represents the sum of i = 1, ... K, and j is n mod Ni and is cyclically repeated. Also, N
i is the number of taps corresponding to one cycle of the i-th vibration.

Now, the active vibration control device controls the difference between d (n) and the control output y (n) at the evaluation point and the residual so as to reduce the engine vibration d (n) at the evaluation point. Although the operation is performed so as to eliminate the vibration e (n), there is a time delay or a phase delay due to the actuator 3 and the system delay 4, for example, from the output point of the IDFT calculation means 2 to the evaluation point. (For example, an actuator composed of a spring and a damper can be approximated to a second-order low-pass filter).
Therefore, the coefficient correction means 8 corrects each coefficient of the IDFT coefficient memory 1 so as to minimize the error detected by the residual vibration detection means 7 while considering the system transmission delay. For example, the coefficient correction means of claim 3 can be used for this correction algorithm, which will be described later.

In order to realize the above-mentioned coefficient correction, it is necessary to know beforehand the transfer characteristic from the output point of the IDFT calculation means 2 to the evaluation point by system identification. An FIR filter is generated and stored in the system transfer characteristic memory 5. Next, the system transfer characteristic (impulse response) on the time axis is subjected to DFT by the DFT calculation means 6 to obtain the system transfer characteristic at the frequency of ωi, and the coefficient correction means 8 uses the system transfer characteristic to correct the coefficient. Will be done.

The concrete construction method will be described in detail in the embodiment of the invention of claim 2. In the measurement window calculation means 11, the lowest order of the vibration to be eliminated, for example, the first order vibration, is set to the window (time). Width). The DFT calculation means 6 can obtain the system transfer characteristics corresponding to the 1st, 2nd, ... kth order by performing the DFT of 2k + 1 taps or more with k as the maximum order to be erased in the time window. It is possible to cause a control error by cutting off the impulse response on the time axis in the time window and omitting the remaining characteristics. However, in order to obtain the transfer characteristics of the system, one cycle or two cycles of data are realistically used. The above method can be used because it is often sufficient. The number of taps of the DFT depends on the number of vibrations to be erased and the system transfer characteristic, but if the third order is generally sufficient, 8 to 16 taps are often sufficient. Therefore, FFT (Fast Fourier Tra
If the above method is processed using, for example, a DSP (Digital Signal Processor), the above calculation can be performed almost in real time.

By the above method, the system transmission delay of the target vibration component can be easily obtained according to the engine speed, and a control system having a relatively small number of control taps (the number of vibrations to be controlled * 2) can be obtained. Since it can be realized, it is possible to configure an active vibration control device capable of reducing engine vibration at an evaluation point to a high rotation speed.

Next, the measurement window calculating means of the embodiment of the invention of claim 2 will be described with reference to the drawings.

FIG. 2 is an explanatory view of the measuring window calculating means in the embodiment of the invention of claim 2 of the present invention. In FIG. 2, (a) is a diagram showing a state of sampling for FFT in the measurement window for obtaining the system transfer characteristic,
(B) is the DFT result of (a), (c) is the system transfer characteristic stored in advance in the memory, (d) is the primary vibration, (e) is the secondary vibration, and (f) is the k-order vibration. An example is shown in the figure.

The measurement window calculating means of the present invention uses the vibration period of the vibration to be eliminated, or the integral multiple of the vibration period of the lowest order in the case of eliminating a plurality of order components, as the measurement window of the DFT. Now, the vibration to be eliminated is a periodic vibration having a correlation with the engine vibration, and is represented by (d) to (f) in FIG.
Let k vibrations from the next to the k-th order. At this time, the lowest order, that is, the longest cycle is the (d) first order vibration. Therefore, the system transfer characteristic (impulse response) stored in advance in the memory of (c) is cut out using the cycle of the primary vibration as a window as shown in (a) and divided (sampled) by the number of taps of the DFT. The result of DFT of this is (b). The component of the second tap is the system transfer characteristic of the primary vibration, the component of the third tap is the system transfer characteristic of the secondary vibration, and the k + 1 tap component. Of the k-th order vibrations show the system transfer characteristics of the k-th order vibration, and are used for the coefficient correction of the active vibration control device according to the first aspect of the invention described above. Thus, the measurement window calculation means of the present invention outputs an integral multiple of the vibration cycle of the lowest dimension as the measurement window of the DFT,
Target vibration after DFT, first to k
All the system transfer characteristics up to the following are efficiently obtained. In general, the higher the frequency of the impulse response is, the more it is included in the previous stage of the response. Therefore, it can be considered that the information about the transfer characteristic is included within one to two cycles of the target vibration. . In other words, the output has a time width that is an integral multiple of the vibration cycle of the lowest order, but the integer of 1 to 2 is usually sufficient.

Next, the coefficient correcting means of the third embodiment of the present invention will be described. The frequency of the vibration to be erased is fi (where fi = i / Ns / Ts, i = 1, ...,
Let k and Ns be the number of IDFT samples, Ts be the IDFT sampling period), the number of angular vibrations corresponding to fi be ωi, and the coefficient of the COS component at time t of fi is Wi1.
(T), the coefficient of the SIN component is Wi2 (t). When the COS component of the system transfer characteristic at the frequency fj is Ci1, the SIN component is Ci2, and the residual error (vibration or noise) at the evaluation point is e (t), I
The output z (t) after DFT is

[0035]

[Equation 3]

The above z (t) passes through the characteristics of the system C and becomes the output y (t) at the evaluation point.

[0037]

[Equation 4]

The residual error e (t) is the difference between the disturbance d (t) and the above y (t).

[0039]

[Equation 5]

Let ε be the square of the error

[0041]

[Equation 6]

From Equation 6, since ε (t) is in the quadratic form of each tap, there is a certain minimum and minimum point regarding the filter coefficient. Ε for each tap by the steepest descent method
Consider converging to a solution that minimizes (t). The gradient vector is

[0043]

[Equation 7]

From equation 7, an equation for updating each tap with a gradient vector is obtained. Expressed by discrete time n,

[0045]

[Equation 8]

Therefore, according to the equation (8), by updating each tap together with the update of the discrete time, it is possible to converge the tap value to reduce the vibration component of each fi.
This can be used as the coefficient updating means of claim 1 by updating up to i = 1, ..., K, that is, updating 2 * K taps.

[0047]

As described above, according to the present invention, the system transfer characteristic used for updating the coefficient of the control system for the vibration to be eliminated is D
It can be easily calculated using FT, and IDFT can be calculated from the above coefficients.
Since the control output is obtained by the method, the calculation time is shortened and the amount of memory is reduced by reducing the taps as compared with the conventional control using the adaptive filter.

[Brief description of drawings]

FIG. 1 is a schematic diagram of an active vibration control device according to an embodiment of the invention of claim 1 of the present invention.

FIG. 2 is an explanatory view of a measurement window calculating means in the embodiment of the invention of claim 2 of the present invention.

FIG. 3 is a schematic diagram of a conventional active vibration control device.

[Explanation of symbols]

 0 car engine 1 IDFT coefficient memory 2 IDFT calculation means 3 actuator 4 system delay 5 system transfer characteristic memory 6 DFT calculation means 7 residual vibration detection means 8 coefficient correction means 9 reference signal generation means 10 reference sine wave generation means 11 measurement window calculation Means 30 Automotive engine 31 FIR filter coefficient memory 32 FIR filter product sum means 33 Actuator 34 System delay 35 System transfer characteristic memory 36 System transfer characteristic product sum means 37 Residual vibration detection means 38 Coefficient correction means 39 Reference signal generation means 40 Reference sine Wave generation means 41 Measurement window calculation means

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Internal reference number FI technical display location G10K 11/178

Claims (3)

[Claims] 1. An actuator (control vibration source or control sound source) for eliminating vibration or noise at an evaluation point, a means for detecting residual vibration at the evaluation point, and a drive output of the actuator are calculated. IDFT
(Inverse Discrete Fourier Transform) calculating means, a coefficient memory consisting of the input coefficients of the IDFT, and a residual vibration at the evaluation point while considering the system transfer characteristic from the actuator drive output to the evaluation point from the DFT result described later. Coefficient correction means for correcting the coefficient memory, and a system transfer characteristic memory for storing the system transfer characteristic in advance as a response (impulse response) on the time axis,
DF for calculating the response on the frequency axis by DFT (Discrete Fourier Transform) of the data of the system transfer characteristic memory
An active vibration control device comprising: T calculation means; and measurement window determination means for determining the measurement window of the DFT from the frequency of the main vibration to be erased. 2. A measurement window determining means for setting a vibration window of a vibration to be erased or a measurement window of DFT to be an integral multiple of the vibration cycle of the lowest order when eliminating a plurality of order components. The active vibration control device according to claim 1. 3. The vibration of the erasing target is from the 1st to the kth, and the frequency to be input to the IDFT calculation means is fi (where fi = i / Ns / Ts, i = 1, ..., K, Ns is I
The number of DFT samples, Ts is the sampling period of the input for IDFT), and the coefficient of the COS component at time n of fi is Wi.
1 (n), the coefficient of the SIN component is Wi2 (n), and the COS component of the system transfer characteristic at the frequency fi is Cil, S
When the IN component is Ci2 and the residual error (vibration or noise) at the evaluation point is e (n), the coefficients Wi1 (n), Wi
2. The active vibration control device according to claim 1, further comprising coefficient correction means for setting 2 (n) to (Equation 8).