JP3391195B2 - Adaptive control method for periodic signals - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention belongs to the technical field of active suppression technology for periodic signals. For example, if the periodic signal is vibration, it belongs to the technical field of active vibration control, and if the periodic signal is noise, it belongs to the technical field of active noise suppression. ing.

[0002]

2. Description of the Related Art As a conventional technique for the present invention, Japanese Patent Application Laid-Open No. 8-44377 (Japanese Patent Application No. 6-201384).
Discloses an adaptive control method for a periodic signal named DXHS algorithm. The DXHS algorithm realizes an adaptive control method that controls the fundamental frequency component of a periodic signal and its harmonic component and suppresses the influence on the observation point. In the DXHS algorithm, a control target signal (periodic signal) whose effect should be suppressed and a control signal (also called an adaptive signal) generated to cancel the influence are represented by sine waves, respectively. Angular frequency, phase, and gain were defined as the main variables. Then, based on the least-squares method whose evaluation function is the square of the error signal observed at the observation point affected by the periodic signal, the gain and phase of the control signal are adaptively adjusted to reduce the influence of the periodic signal. Had been minimized. Also, in consideration of the fluctuation of the phase delay characteristic of the controlled system, the adaptability to the fluctuation of the controlled system characteristic was also improved by introducing the table data.

Such a DXHS algorithm has the advantages that it is less susceptible to the influence of disturbance and the amount of calculation is smaller than that of the conventional algorithm in the above publication.

[0004]

However, in the above-mentioned prior art, the adaptability to the time fluctuation (the transfer characteristic changes with time at a predetermined angular frequency) of the controlled system which transfers the control signal to the observation point is required. It wasn't enough. Therefore, as a prior art, the inventors have developed an improved algorithm (DXHS modified algorithm), and according to this algorithm, it is possible to quickly adapt to a large change in the transfer characteristic of the controlled system. Got This prior art is filed as Japanese Patent Application No. 7-129868.

The DXHS modified algorithm is similar to the prior art in that the periodic signal and the control signal are each defined by a harmonic function. However, the adaptive coefficient vector W
In addition to the amplitude and phase of the control signal,
The difference is that an adaptation coefficient for the phase delay of the controlled system is introduced. With the introduction of the adaptation coefficient,
The adaptive coefficient vector updating algorithm for updating the adaptive coefficient vector W (n) also includes a component for adjusting the adaptive coefficient. In addition, the convergence is improved by adding a phase adjustment parameter to the component that adjusts the adaptive coefficient of the adaptive coefficient vector updating algorithm.

As a result, according to the adaptive control method of the periodic signal using the DXHS modified algorithm, the adaptive ability with respect to the temporal change of the transfer characteristic of the controlled system is dramatically improved. By the way, in both the above-mentioned conventional technology and the above-mentioned prior art, the angular frequency ω _k of the periodic signal is measured with sufficient precision in terms of engineering, and the measured value of the angular frequency ω _k is equivalent to the true value. The theory has been developed as something that can be handled. Therefore, the measurement error of the angular frequency ω _k becomes large to some extent, and if it reaches a level that cannot be ignored, the adaptive control may be hindered.

Therefore, the present invention can be applied even when the level of the measurement error of the angular frequency ω _k becomes large to some extent, in other words, the periodic signal that is robust against the measurement error of the angular frequency ω _k . It is a problem to be solved to provide an adaptive control method.

[0008]

Means for Solving the Problems and Their Actions / Effects In order to solve the above problems, the inventors have invented the following means. Here, the angular frequency ω _k , its compensation value (estimated value of measurement error) q _k , and the transfer characteristic G (estimated gain A hat and estimated phase delay Φ hat) change with each update of time n. Therefore, it should be indicated by adding (n), but in order to avoid complication, it may be described without (n). However, the measurement error q _k and the compensation value q _k (n) have different definitions, and the compensation value q _k (n) converges on the measurement error q _k . It should be noted that the symbols in the mathematical formulas, which are commonly referred to as hats or roofs, cannot be written as they are in the text of the specification due to electronic application restrictions, and are therefore replaced with the suffix "hat".

(First Means) A first means of the present invention is a periodicity which includes a signal component of at least one angular frequency ω _k (1 ≦ k ≦ K ′, K ′ is a natural number) and affects an observation point. with respect to the signal f (n), consisting of angular sinusoidal signal of K is an estimated value estimated angular frequency omega _k hat of angular frequency _{ω k (1 ≦ k ≦ K} ≦ K ', K also a natural number) By directly or indirectly applying the adaptive signal y (n) in antiphase, the influence of the specific component of the periodic signal f (n) on the observation point is actively removed, and the detection is performed at the observation point. In the adaptive control method of the periodic signal for reducing the error signal e (n)
_An angular frequency measuring means for measuring _k and supplying a measurement value (ω _k + q _k ) including a measurement error q _k , and at time n, the estimated angular frequency ω _k hat is an angular frequency and an amplitude a _k and a phase phi _k adaptive signal generation algorithm least one generates the adaptive signal formed by combining y (n) of the sinusoidal signal component amplitude a _k and the phase phi _k of the adaptive signal y (n) , The phase delay Φ of the transfer characteristic G until the adaptive signal y (n) is transmitted to the observation point, the error signal e (n) and the measured value (ω _k + q). _k )), the angular frequency ω _k at each angular frequency component of the periodic signal f (n), the fluctuation of the amplitude and the phase, and the fluctuation of the transfer characteristic G are updated by updating each time n based on _k ). Adaptively adjust each component of the adaptive coefficient vector W (n) with respect to The adaptive coefficient vector update algorithm that, the compensation value q of the estimated angular frequency omega _k hat above
_k (n) is the error signal e (n) and the measured value (ω _k +
q _k ), the phase φ _k, and the phase delay Φ based on the time n
Of the estimated angular frequency ω _k hat [that is, the measured value (ω _k + q _k ) to the compensation value q _k
[Value obtained by subtracting (n)] and an angular frequency estimation algorithm for adaptively increasing the estimation accuracy, and the amplitude a _k and the phase φ _k that are components of the updated adaptive coefficient vector W (n). characterized in that the with the said estimated angular frequency omega _k hat, the adaptive signal y (n) the estimated angular frequency omega _k hat and shake each sinusoidal signal width a _k and the phase phi _k is updated Is an adaptive control method for periodic signals.

In this means, the measurement value (ω _k + q _{k of} the angular frequency ω _k to which a significant measurement error q _k is added from the vibration source that generates the periodic signal f (n) by the angular frequency measurement means. ) Is measured and provided to the adaptive coefficient vector update algorithm and the angular frequency estimation algorithm. The adaptive coefficient vector update algorithm uses the adaptive coefficient vector W (n) including the amplitude a _k and the phase φ _k as components as the error signal e (n).
And the like, and provides the adaptive signal generation algorithm with the amplitude a _k and the phase φ _k that have been adaptively updated and converged to appropriate values.

On the other hand, the angular frequency estimation algorithm sequentially updates the compensation value q _k (n) based on the error signal e (n), the measured value of the angular frequency (ω _k + q _k ), etc. Estimate _k . If the measurement error q _k is estimated with sufficient accuracy,
The estimated frequency error ω _k is precisely determined by subtracting the estimated measurement error q _k from the measured value (ω _k + q _k ), and the estimated value ω _k hat of the angular frequency ω _k that does not include a significant error is used as the adaptive signal generation algorithm. Can be provided to.

The value of the phase delay Φ of the transfer characteristic G may be replaced with an appropriate estimated value (constant value), or table data may be prepared to prepare an appropriate value for the angular frequency ω _k . However, some errors are absorbed by adjusting the phase φ (n). The adaptive signal generation algorithm provides the adaptive signal y (n) by providing the angular frequency ω _k of each sine wave signal from the angular frequency estimation algorithm and the amplitude a _k and the phase φ _k from the adaptive coefficient vector updating algorithm. Occur.
The adaptive signal y (n) reaches the observation point by adding the gain A and the phase Φ via the transfer characteristic G. At the observation point, the influence of the periodic signal f (n) and the influence of the adaptive signal y (n) cancel each other, and the periodic signal f (n) selected as the adaptive signal y (n).
The influence of the specific component of (n) is removed, and the error signal e (n)
Is suppressed to a low level.

Therefore, according to this means, the periodic signal f
Even if the measured value (ω _k + q _k ) of the angular frequency ω _{k in} (n) includes a significant measurement error q _k , the error signal e (n) can be converged to a low level with high adaptability. The effect is that you can do it. (Second Means) A second means of the present invention is the first means, wherein the angular frequency measuring means is the periodic signal f (n).
Is an adaptive control method for a periodic signal, which is a periodic sensor that measures a time interval of pulse signals emitted from a source at a predetermined phase, or a pulse counter that measures the number of the pulse signals within a predetermined time.

In this means, the angular frequency measuring means is an inexpensive and highly reliable sensor, which has the effect of reducing the cost of the entire control system. If it is necessary to measure the angular frequency ω _k from the source (eg, engine) of the periodic signal f (n) more precisely, the angular frequency measuring means may be a photoelectric rotary encoder or a rotary type encoder. It is also possible to use a more precise sensor such as Duct Thin.

(Third Means) A third means of the present invention is the one-input one-output type according to the first means, which controls the most main vibration (angular frequency ω) of the periodic signal f (n). Of the periodic signal, the adaptive coefficient vector update algorithm and the angular frequency estimation algorithm are collectively expressed by Equation 7, and the adaptive signal generation algorithm is expressed by Equation 8. This is an adaptive control method.

[0016]

[Equation 7]

[0017]

[Equation 8]

According to the present means, the adaptive control of the first means is possible by extremely simple calculation when the frequency component to be suppressed in the one-input one-output control system is single. Therefore, according to the present means, when the present means is implemented by a microcomputer or the like, there is an effect that the cost can be significantly reduced in terms of the amount of calculation, the calculation speed, and the storage capacity. (Fourth Means) According to a fourth means of the present invention, in the first means, L error signals e _l (n) (from the at least one observation point affected by the periodic signal f (n) are observed. 1
≦ l ≦ L, L is a natural number), and is a multi-input multi-output periodic signal that outputs M adaptive signals y _m (n) (1 ≦ m ≦ M, M is a natural number). In the adaptive control method, the adaptive coefficient vector updating algorithm is expressed by Equation 9, the angular frequency estimation algorithm is expressed by either Equation 10 or Equation 11, and the adaptive signal generation algorithm is expressed by Equation 12. , An adaptive control method for periodic signals.

[0019]

[Equation 9]

[0020]

[Equation 10]

[0021]

[Equation 11]

[0022]

[Equation 12]

In the present means, the adaptive control of the first means is possible even in the case of a multi-input multi-output control system (one input or one output is also included as a special case) when there are a plurality of frequency components to be suppressed. is there. As the angular frequency estimation algorithm, when the equation (10) is adopted, the calculation amount can be significantly reduced, and when the equation (11) is adopted, all inputs, that is, all error signals e _l (n) are used. It has the effect of being able to adapt more quickly and expanding the range of application.

(Fifth Means) According to a fifth means of the present invention, in any one of the first to fourth means, the amplitude a _k (or a _km ) and the phase φ _k (or φ _km ). and at least one value among the estimated gain a hat and the estimated phase delay Φ hat said an estimate of the transfer characteristic G is, for each compartment of the angular frequency [omega _k within the range of variation of the angular frequency omega _k It is an adaptive control method for a periodic signal, which has stored table data.

According to the present means, since one of various numerical values greatly influenced by frequency is stored in the table data divided by frequency, even when the frequency fluctuates and the characteristic of the system fluctuates greatly, You can refer to the value of table data. Therefore, according to this means, there is an effect that the error signal e (n) can be suppressed more promptly by adaptive control even when the frequency changes.

Further, the control system is also provided with an adaptive judgment function,
If there is a table data update function that updates the value of the contents of the table data when fully adapted, there is an effect that adaptive control can be performed in response to changes over time in the system to be controlled.

[0027]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the adaptive control method for a periodic signal according to the present invention will be described clearly and sufficiently in the following embodiments so that those skilled in the art can understand the embodiments. [Embodiment 1] (System configuration and theoretical development of Embodiment 1) A system for implementing an adaptive control method of a periodic signal as Embodiment 1 of the present invention is
As shown in FIG. 1, it is roughly divided into two blocks, a control system 1 and a physical system 2.

The control system 1 includes an adaptive signal generation algorithm 11 for generating an adaptive signal y (n), and the same algorithm 1
1 is composed of an adaptive coefficient vector updating algorithm 12 and an angular frequency estimating algorithm 13 which provide an amplitude a (n), a phase φ (n) and an angular frequency estimate ω hat. The control system 1 is a one-input one-output type, and the error signal e (n) corresponds to an input and the adaptive signal y (n) corresponds to an output. Further, the control system 1 uses the periodic signal f (n)
Of these, only a single frequency component (the true value of the angular frequency is ω) is the specific component to be suppressed.

On the other hand, the physical system 2 includes a periodic signal source 21 for applying the periodic signal f (n) to the observation point 24, and an angular frequency measuring means 22 for measuring the angular frequency ω of the periodic signal source 21. ,
Transfer characteristic G for transferring the adaptive signal y (n) to the observation point 24
(Reference numeral 23). Periodic signal source 2
Specifically, 1 includes a vibration source such as an engine and another transmission characteristic G ′ (not shown) that transmits the vibration to the observation point 24. The angular frequency measuring means 22 measures the angular frequency ω (true value) of the periodic signal f (n), and the measured value (ω + q) including the observation error q is updated by the adaptive coefficient vector updating algorithm 12 described above. And the angular frequency estimation algorithm 13. The transfer characteristic G exhibits a predetermined gain A and a phase delay Φ at a predetermined angular frequency ω.

Here, each algorithm 11 of the control system 1,
12, 13 can be derived as follows. First, suppose that the periodic signal f (n) = Fcos (ωT
n). Here, F is a predetermined amplitude, T is an update cycle (or sampling cycle), and n is each time (step) in T increments. On the other hand, if the adaptive signal y (n) is y
(N) = a (n) cos {(ω + Q) Tn + φ (n)}
Then, the cancellation signal z (n) applied to the observation point 24 via the transfer characteristic G [A, Φ] (reference numeral 23) is z (n).
= Aa (n) cos {(ω + Q) Tn + φ (n) −Φ}
Is.

Then, the error signal e (n) is e (n) =
F (n) + z (n) = Fcos (ωTn) + Aa (n)
cos {(ω + Q) Tn + φ (n) −Φ}. Rewriting this into an exponential function expression, the error signal e (n) = F
exp [jωTn] + Aa (n) exp [j {(ω +
Q) Tn + φ (n) −φ}] = Fexp [jωTn] ×
[1- (1 / F) Aa (n) expj {QTn + φ
(N) -Φ}].

Here, if the error signal e (n) = 0,
The relationship represented by the following mathematical formula is established between the periodic signal f (n) which is an external force, the transfer function and the filter coefficient. _{f = Fexp (jω f t)} = AaXexp {j (ω f t + φ-Φ)} On the other hand, for some reason, the angular frequency used for calculating an adaptive signal y (n) ω and periodic signals f (n) Angular frequency ω
_If an angular frequency difference Q occurs between _f and _f , the error signal e
(N) is not zero, and its value is expressed by the following equation.

[0033] e (n) = - [- AaXexpj {ω f t + φ-Φ} + AaXexpj {(ω f + Q) t + φ-Φ}] = AaX {1-exp (jQt)} · exp {j (ω f t + φ- Φ)} = B (t) {1-exp (jQt)} = B (t) {1- (cosQt + jsinQt)} However, a B (t) = AaXexp {j (ω f t + φ-Φ)}.

Here, the periodic signal f (n) and the cancellation signal z
The synchronization with (n) is a necessary condition that the control system 1 is completely adapted and the error signal e (n) is constantly zero [e (n) ≡0]. That is, the error signal e (n) ≡0 can be obtained only when the angular frequency error Q = 0. In particular, when Q is in the approximate region of zero, the error signal e (n) is expanded as follows.

[0035] Therefore, the expected square value E [e of the error signal e (n) is
² (n)] is the right side [jB] of the above equation over a predetermined time T.
It is obtained by time-integrating the square of the absolute value of (t) Q], dividing by the predetermined time T, and averaging. If the predetermined time T is a sufficiently long time, the squared expected value of the error signal e (n) is expressed by the following equation, considering that B (t) is a periodic function.

E [e ² (n)] = αQ ² (α is a positive proportional constant) Therefore, the expected value E [e of the square of the error signal e (n) E [e]
² (n)] is approximately proportional to the square value Q ² of the error Q of the angular frequency. That is, reducing the error Q of the angular frequency ωk is a necessary condition for reducing the expected value of the error signal e (n).

[0037] ^{Therefore, E [e 2 (n)} ] and approximate proportional relationship between Q ^2, or a Q ² is sufficiently small range relates the increase in ^{Q 2 E [e 2 (n} )] is a simple increase Taking advantage of the tendency, the iterative approximation method is introduced. That is, by introducing a correction term (corresponding to the compensation value q (n)) into the measurement value (ω + q) of the angular frequency ω, the compensation value q (n) is sequentially calculated by the gradient method (least squares approach). I will ask. Once the compensation value q (n) is obtained, the estimated value ω hat of the angular frequency ω is obtained by subtracting the compensation value q (n) from the measured value (ω + q), and thus the true value of the angular frequency ω is estimated. can do. In the gradient method, the adaptive speed can be adjusted by appropriately setting the step size parameter, and the convergence can be speeded up.

If the compensation value of the angular frequency is q and the adaptive coefficient vector W (n) is defined to have only the amplitude a (n) and the phase φ (n) as components, the adaptive coefficient vector according to the DXHS algorithm described above is defined. The update algorithm 12 is
It is expressed by the number 13.

[0039]

[Equation 13]

On the other hand, the angular frequency estimation algorithm 13 is
It is derived as follows. That is, the evaluation function is set to the instantaneous squared value e ^{2 of the} error signal e (n) (at the time n).
If defined as (n), the compensation value q of the evaluation function e ² (n)
The gradient D _{q with} respect to (n) is defined by the following mathematical formula. When D _q ≡ ∂e ² (n) / ∂q (n), the update formula obtained by sequentially updating the compensation value q (n) is expressed by the following mathematical formula.

[0041] q (n + 1) = q (n) + μ q '(-D q) By the way, the second component of the equation 13, the following approximate expression is derived. ∂φ (n) / ∂Tn ≈ {φ (n + 1) −φ (n)}
/ T = e (n) μφ cos {(ω + q) Tn + φ
(N) −Φ} / T Therefore, the squared error e ² (n) and (∂φ (n) / ∂T
It can be considered that there is an almost proportional relationship with n) in an instant.

It has been empirically known that the phase φ (n) and the compensation value q (n) have a linear relationship represented by the following equation. ∂φ (n) / ∂Tn = −αq (n) + β (α,
β: constant) Based on the above relationship, the gradient D _q is developed as shown in Expression 14.

[0043]

[Equation 14]

Since the above-mentioned gradient D _q can be simply expressed as in equation (14), the angular frequency estimation algorithm 13, which is an updating formula of the compensation value q (n), uses the above-mentioned q (n + 1).
_{= Q (n) + μ q} ' are Kakiarawasa as number 15 on the basis of (-D _q).

[0045]

[Equation 15]

Therefore, the adaptive coefficient vector update algorithm 12 of the equation 13 and the angular frequency estimation algorithm 13 of the equation 15 are put together into an adaptive coefficient vector update algorithm 12 and an angular frequency estimation algorithm 13.
Can be expressed by Equation 16.

[0047]

[Equation 16]

Further, the adaptive signal generation algorithm 11 is
According to the definition, it is represented by the number 17.

[0049]

[Equation 17]

As described above, the system shown in FIG.
The configurations of both the physical system 2 and the control system 1 can be formulated. (Effects of Embodiment 1) Since the adaptive control method of the periodic signal of this embodiment is configured as described above, it has the following two effects.

First, in this embodiment, the angular frequency measuring means 2
2, the measurement value (ω + q) of the angular frequency ω to which the significant measurement error q is added is measured from the periodic signal source 21 that generates the periodic signal f (n), and the adaptive coefficient vector update algorithm 12 and the angle are calculated. It is provided to the frequency estimation algorithm 13. Both algorithms 12 and 13 are
, The amplitude a (n) and the phase φ
Since the first and second components of (n) and the third component, which is the compensation value q (n), have different characteristics, both of them are referred to as algorithm 1
The operation will be described separately for 2 and 13.

Adaptive coefficient vector updating algorithm 12
Updates the adaptive coefficient vector W (n) having the amplitude a (n) and the phase φ (n) as components, based on the error signal e (n) and the like. The amplitude a (n) and the phase φ (n) converged to appropriate values by the action of the adaptive coefficient vector updating algorithm 12 are provided to the adaptive signal generation algorithm 11.

On the other hand, the angular frequency estimation algorithm 13
Based on the error signal e (n), the measured value (ω + q) of the angular frequency, etc., the compensation value q (n) is sequentially updated to estimate the measurement error q. If the measurement error q is estimated with sufficient accuracy, the estimated measurement error q (or the compensation value q is calculated from the measurement value (ω + q).
By subtracting (n)), the angular frequency ω can be accurately estimated. As a result, the estimated value of the angular frequency ω that does not include a significant error ω hat = (ω + q) -q (n)
Are provided from the angular frequency estimation algorithm 13 to the adaptive signal generation algorithm 11.

Here, the value of the phase delay Φ of the transfer characteristic G in the equation 16 is preset by the frequency sweep test of the input / output characteristic. Even if the phase delay Φ has a slight difference from the actual transfer characteristic G (reference numeral 23), it is absorbed by the adjustment of the phase φ (n), so that there is no problem in adaptability. Now, the adaptive signal generation algorithm 11 calculates the estimated value ω hat, which is the angular frequency of each sine wave signal, from the angular frequency estimation algorithm 13 and the amplitude a (n) and the phase φ (n) to the adaptive coefficient vector update algorithm 12 From each to generate an adaptive signal y (n). Adaptive signal y (n)
Shows the gain A and the phase Φ through the transfer characteristic G (reference numeral 23).
Are added, and the canceling signal z (n) arrives at the observation point 24. At the observation point 24, the periodic signal f (n) and the canceling signal z (n) cancel each other, and the main component (angular frequency ω) of the periodic signal f (n) selected as the adaptive signal y (n). ) Is removed and the error signal e (n) is suppressed to a low level.

Therefore, according to the adaptive control method of the periodic signal of the present embodiment, a significant measurement error q _k is included in the measurement value (ω _k + q _k ) of the angular frequency ω _k of the periodic signal f (n). Even if it is, there is an effect that the error signal e (n) can be converged to a low level with high adaptability. Secondly, in the present embodiment, since the control system 1 is of one-input one-output type and the frequency component to be suppressed is single in the periodic signal f (n), the control system 1 has an extremely simple algorithm 11. , 1
2, 13 (16 and 17).
Therefore, the adaptive control method of the periodic signal according to the present embodiment can be implemented by extremely simple calculation. Therefore, according to the present embodiment, when the control system 1 is realized by a microcomputer or the like, there is an effect that a great saving can be achieved in terms of the amount of calculation, the calculation speed, and the storage capacity.

(Evaluation by Numerical Simulation of Embodiment 1) In an actual system such as a vehicle-mounted vibration damping system for suppressing the vibration of an automobile engine, the periodic signal source 2 is used.
The true value of the angular frequency ω cannot be observed from 1 (engine) without a measurement error q. Here, assume that the angular frequency measuring means 22 uses the ignition pulse of the engine as a signal source of the measured value (ω _k + q _k ). In this case, if the engine is an in-line 4-cylinder 4-stroke cycle type, the ignition pulse is obtained at a rate of two per one revolution of the crankshaft.

The sampling frequency of the control system 1 is 2000H
If the engine speed is 600 rpm, the ignition pulse interval is 100 samplings, and the counting error of one sampling pulse is
Only about 1%. At this time, the angular frequency measuring means 22
Is a pulse counter for sampling pulses. However, when the engine speed reaches 6000 rpm, which is ten times higher, only 10 sampling pulses are counted in the ignition pulse time interval, so that one count error reaches 10%. That is, the measurement error q of the angular frequency ω of the angular frequency measurement means 22 reaches a level of about several% of the angular frequency ω on average, and the cycle does not consider the conventional measurement error of the angular frequency ω. In the adaptive control method of the sex signal, it is expected that the measurement error q will adversely affect the control performance.

Therefore, the inventors evaluated the performance of the adaptive control method of the periodic signal of this embodiment by performing a numerical simulation and comparing it with the above-mentioned conventional technique (DXHS algorithm). At that time, the assumed frequency corresponding to the frequency of the ignition pulse was 25 Hz (corresponding to the rotation speed of 750 rpm), and the assumed sampling frequency was 1000 Hz. Also. The measurement error q from the true value of the angular frequency ω is 0% to −
Up to 5%, 6 cases were numerically simulated in 1% steps.

As a result, as shown in Table 1 below, the variance E [e of the error signal e (n) in the steady state is shown.
² (n)] has been improved by orders of magnitude in this example. That is, the variance in the prior art increases exponentially as the measurement error q increases, whereas the variance in the present embodiment is from the case where the measurement error q is 0%. It is at a consistently low level until it reaches -5%. That is, according to the adaptive control method of the periodic signal of the present embodiment, it can be read from Table 1 that the influence does not reach the control result as long as the level of the measurement error q of the angular frequency ω is at least about several%. be able to.

[0060]

[Table 1] ───────────────────────────── q / ω Dispersion of the present embodiment Dispersion of the prior art ───── ──────────────────────── 0% 1.0 × 10 ^-8 or less 1.0 × 10 ^-8 or less -1% Same as above 7.7 × 10 ^-6 -2% Same as above 2.9 × 10 ^-5 -3% Same as above 6.0 × 10 ^-5 -4% Same as above 9.4 × 10 ^-5 -5% Same as above 1.2 × 10 ^-4 ────────── ──────────────────── Error signal e in each case of the numerical simulation
2 (a), (b) to FIG. 7 (a), comparing the graph of (n) with that of the present example and that of the prior art.
It shows in (b).

As a result, as can be seen by comparing FIG. 2A and FIG. 7A, as a result of the adaptive control of the first embodiment, the measurement error q reaches -5% of the angular frequency ω. , Error signal e
Not only the level of (n) is low, but also its convergence speed is not deteriorated at all. On the other hand, when comparing FIG. 2 (b) to FIG. 7 (b) in order, the adaptive control according to the prior art does not fully converge, and even in the quasi-steady state, a high error signal e (n) due to the magnitude of the measurement error q. You can see that you are at the level. However, the speed (elapsed time) of converging to the quasi-steady state does not change visibly even if the measurement error q expands to -5%.

From the above numerical simulation results, according to the adaptive control method of the periodic signal of the present embodiment, even under the bad condition that the measurement error q of the angular frequency ω reaches several%,
It was proved that the adaptive control can be performed with almost no change from the case where there is no measurement error q. as a result,
It was found that the error signal e (n) can be converged promptly and the level thereof is extremely low as compared with the case where there is no measurement error q.

Therefore, according to this embodiment, the angular frequency ω
The control performance does not deteriorate even if the level of the measurement error q is increased to some extent, and it is possible to provide an adaptive control method for the periodic signal that is robust against the measurement error q. Also, to a large extent, the measurement error q of the angular frequency ω
Therefore, even if an expensive sensor is not separately provided in the angular frequency measuring means 22, the angular frequency measuring means 2 is simple enough to be calculated from the ignition pulse by a pulse counter.
2 is enough. Therefore, the angular frequency measuring means 22 can be inexpensive, which also contributes to cost reduction of the entire product.

(Modification of First Embodiment) As shown in FIG. 8, in addition to the control system 1 of the first embodiment, the phase delay Φ of the transfer characteristic G (reference numeral 23 in FIG. 1) and the amplitude a and the phase φ.
It is also possible to configure a modification having a control system 1 ′ provided with table data 14 having the contents of. That is, in this modification, the values of the measured values of the proper amplitude a and phase φ of the adaptive signal y (n) and the phase delay Φ of the transfer characteristic G are the periodic signal f (n) and the angular frequency ω. The table data 14 is stored for each section of the angular frequency ω within the fluctuation range of. Here, since the gain characteristic A of the transfer characteristic G can be absorbed in the value of the amplitude a, the table data 14 does not need to include the gain characteristic A.

In this modification, numerical values (amplitude a, phase φ, phase delay Φ) that are greatly influenced by frequency are stored in the table data 14 divided for each angular frequency ω.
In the table data 14, when the estimated value ω hat of the angular frequency ω fluctuates beyond the frequency section of the table data, the switch 15 is closed and the adaptive coefficient vector update algorithm 12 and the angular frequency estimation algorithm 13 use the above numerical values. To supply. At that time, angular frequency ω (true value is unknown)
Instead of, the estimated value ω hat is referred to and the above numerical value of a predetermined section in the table data 14 is read.

Then, even when the characteristics of the system (that is, the amplitude a, the phase φ, and the phase delay Φ) greatly vary due to the variation of the angular frequency ω, the values of the table data 14 are referred to so that both algorithms 12 , 13 can be set appropriately without waiting for adaptation. Therefore, according to this modification, the angular frequency ω of the periodic signal f (n) is
Even when f is changed, there is an effect that the error signal e (n) can be suppressed more quickly by adaptive control.

Further, it is possible to adopt a modified mode in which the adaptive judgment function is also provided, and the table data update function for updating the value of the content of the table data at a time when the table is fully adapted is also possible. According to this modification, there is an effect that adaptive control can be performed in response to a change with time of the system to be controlled. The table data 14 is realized by a high-speed memory such as a ROM or a RAM as a device.

[Embodiment 2] (System Configuration of Embodiment 2) As shown in FIG. 9, an adaptive control method for a periodic signal according to Embodiment 2 of the present invention uses Embodiment 1 as a multi-input multi-output system. An expanded version of the periodic signal f
(N) is also generalized to have a plurality of frequency components (angular frequency ω _k ). That is, in this embodiment, L error signals e _l (n) (1 ≦ l ≦ L, L is a natural number) are input as inputs from at least one observation point 24 ″ affected by the periodic signal f (n). Obtained, M said adaptive signals y
_This is an adaptive control method for a multi-input multi-output periodic signal that outputs _m (n) (1 ≦ m ≦ M, M is a natural number). However,
As a special case, a one-input system, one-output system, and a periodic signal f (n) having only a single angular frequency ω component are also included in the category of this embodiment.

Similar to the first embodiment, the system configuration of this embodiment is roughly divided into a multi-input multi-output control system 1 "and a periodic signal f (n) consisting of at least one sine wave component. Physical system 2 ". The physical system 2 ″ includes a periodic signal source 21 ″, an angular frequency measuring means 22 ″, and a transfer characteristic G _m.
(Reference numeral 23 "). On the other hand, the control system 1" is composed of an adaptive signal generation algorithm 11 ", an adaptive coefficient vector update algorithm 12", and an angular frequency estimation algorithm 13 ".

Adaptive Coefficient Vector Update Algorithm 12 "
Is expressed by Equation 18, and the angular frequency estimation algorithm 13 "
Is expressed by Equation 19, and the adaptive signal generation algorithm 11 ″
Is expressed by Equation 20. Here, since the adaptive coefficient vector update algorithm 12 ″ (Equation 18) and the angular frequency estimation algorithm 13 ″ (Equation 19) have the same method of calculating each element, they are summarized as in the first embodiment. May be written as

[0071]

[Equation 18]

[0072]

[Formula 19]

[0073]

[Equation 20]

(Effect of Embodiment 2) In this embodiment, the control system 1 "has multiple inputs and multiple outputs (error signal e which is an input).
_l (n) is L, and adaptive signals y _m (n) that are output are M)
The present invention can also be applied to the control system of No. 1 and the number of frequency components to be suppressed (K). As a result, the measurement error q _k when the angular frequency ω _k of the periodic signal f (n) is measured, even though the multi-input multi-output system suppresses a plurality of frequency components.
, The error signals e _l (n) can be quickly converged and the influence of the specific component of the periodic signal f (n) can be suppressed.

Also, the angular frequency estimation algorithm 13 "
, As shown in the above equation 19, observes all the error signals e _l (n) and updates the compensation value q _k (n).
The convergence of the compensation value q _k (n) takes place quickly and accurately. For the same reason, the compensation value q _k (n) can be converged in a wider range even when the measurement error q _k , noise, and the like are large.

Therefore, according to the present embodiment, the influence of a plurality of vibration components is suppressed in the multi-input multi-output system regardless of the measurement error q _k of the plurality of angular frequencies ω _k of the periodic signal f (n). It is possible to do so, and there is an effect that the convergence range is wide. (Modification 1 of Embodiment 2) As the angular frequency estimation algorithm 13 ″ of the control system 1 ″ of Embodiment 2, a modification can be used in which Expression 21 is used instead of Expression 19 above.

[0077]

[Equation 21]

In this modification, when updating each compensation value q _k (n) by the angular frequency estimation algorithm 13 ″, the updating is performed based on only one error signal e _l (n). , Not all the error signals e _l (n) are referred to.Therefore, in this modification, the calculation amount in the angular frequency estimation algorithm 13 ″ is drastically reduced, and the load on the processor that realizes the control system 1 is reduced. There is an effect that it is possible to reduce the cost or reduce the cost by using an inexpensive processor.

The error signal e supplied to the above equation 21
_l (n) has the largest influence of the frequency component of the angular frequency ω _k to which the compensation value q _k (n) corresponds among the L error signals e _l (n) (or Ratio is large)
It is preferable that one is selected. (Modification 2 of Embodiment 2) Also in this embodiment, the amplitude a _{km, the} phase φ _km, and the phase delay Φ _km are stored for each section of each angular frequency ω _k , as in the modification of Embodiment 1. Modifications are possible with the table data being processed. According to this modification, similarly to the modification of the first embodiment, even when the angular frequency ω _k fluctuates and the characteristics of the system fluctuate significantly, there is an effect that the adaptation is performed more quickly.

[Brief description of drawings]

FIG. 1 is a block diagram showing a system configuration of a first embodiment.

FIG. 2 is an assembly diagram showing an effect of Example 1 without a measurement error. (A) Control result of Example 1 (b) Control result of conventional technology

FIG. 3 is an assembly diagram showing the effect of Example 1 with a measurement error of 1%. (A) Control result of Example 1 (b) Control result of conventional technology

FIG. 4 is an assembly diagram showing the effect of Example 1 at a measurement error of 2%. (A) Control result of Example 1 (b) Control result of conventional technology

FIG. 5 is an assembly diagram showing the effect of Example 1 with a measurement error of 3%. (A) Control result of Example 1 (b) Control result of conventional technology

FIG. 6 is an assembly diagram showing the effect of Example 1 at a measurement error of 4%. (A) Control result of Example 1 (b) Control result of conventional technology

FIG. 7 is an assembly diagram showing the effect of Example 1 at a measurement error of 5%. (A) Control result of Example 1 (b) Control result of conventional technology

FIG. 8 is a block diagram showing a configuration of a control system according to a first modification of the first embodiment.

FIG. 9 is a block diagram showing the system configuration of the second embodiment.

[Description of Codes] 1,1 ′, 1 ″: Control system 11, 11 ″: Adaptive signal generation algorithm 12, 12 ″: Adaptive coefficient vector update algorithm 13, 13 ″: Angular frequency estimation algorithm 14: Table data 15: switch 2,2 ": the physical system 21, 21 ': periodic signal source 22, 22': angular frequency measuring means 23, 23": transfer characteristic G [A, Φ], G m [A m,
Φ _m] 24,24 ": observation point f (n): periodic signal _{e (n), e l (} n): the error signal (1 ≦ l ≦ L) y (n), y m (n): Adaptation Signal (1 ≦ m ≦ M) z (n), z _m (n): Cancellation signal ω, ω _k : Angular frequency (1 ≦ k ≦ K) (ω + q),
(Ω _k + q _k ): Measurement value q, q _k : Measurement error q (n), q _k (n): Compensation value ω hat, ω _k hat: Estimated value of angular frequency a (n), a _km ( n): Amplitude of adaptive signal φ, φ _km :
Phase of adaptive signal A, _Am : Transfer characteristic gain Φ, Φ _m : Phase delay of transfer characteristic

─────────────────────────────────────────────────── --- Continuation of the front page (56) Reference JP-A-7-334165 (JP, A) JP-A-7-133842 (JP, A) JP-A-6-51787 (JP, A) JP-A-7- 64572 (JP, A) JP-A-7-84585 (JP, A) (58) Fields investigated (Int.Cl. ⁷ , DB name) G05B 11/00-13/04 G10K 11/00-13/00

Claims (5)

(57) [Claims] 1. At least one angular frequency ω _k (1 ≦ k ≦
K ', K' whereas affect periodic signals f (n) at the observation point includes a signal component of a natural number), angular the K estimated angular frequency which is an estimate omega _k of the frequency [omega _k Hat (1 ≤ k
≦ K ≦ K ′, where K is also a natural number) by directly or indirectly adding an adaptive signal y (n) consisting of a sine wave signal in antiphase, thereby observing the observation point of a specific component of the periodic signal f (n). In the adaptive control method of the periodic signal that actively removes the influence on the error signal e (n) detected at the observation point, the angular frequency ω _k is measured, and the measurement error q _k is calculated. Angular frequency measuring means for supplying a measured value (ω _k + q _k ) including: at time n, at least the sine wave signal of amplitude a _k and phase φ _{k with} the estimated angular frequency ω _k hat being the angular frequency An adaptive signal generation algorithm for generating an adaptive signal y (n) formed by combining one or more, and an adaptive coefficient vector W (n) including the amplitude a _k and the phase φ _k of the adaptive signal y (n) as components. , The adaptive signal y
The phase delay Φ of the transfer characteristic G until (n) is transferred to the observation point, the error signal e (n), and the measured value (ω _k
+ Q _k ), the angular frequency ω _k at each angular frequency component of the periodic signal f (n) and variations in amplitude and phase and the transfer characteristic G
Of the adaptive coefficient vector W (n), the adaptive coefficient vector update algorithm for adaptively adjusting each component of the adaptive coefficient vector W (n), and the compensation value q _k (n) of the estimated angular frequency ω _k hat The error signal e (n), the measured value (ω _k + q _k ), the phase φ _k
And adaptively updating each time n based on the phase delay Φ, and subtracting the compensation value q _k (n) from each estimated angular frequency ω _k hat [that is, the measured value (ω _k + q _k ). And an estimated value of the amplitude a _k and the phase φ _k , which are components of the updated adaptive coefficient vector W (n), and the estimated angular vibration. With the number ω _k hat, the estimated angular frequency ω _k hat of each sine wave signal of the adaptive signal y (n), the amplitude a _k, and the phase φ _k
A method for adaptively controlling a periodic signal, characterized in that is updated. 2. The angular frequency measuring means measures a time interval of pulse signals emitted from a source of the periodic signal f (n) at a predetermined phase, or the pulse signal within a predetermined time. The adaptive control method for a periodic signal according to claim 1, which is a pulse counter for measuring the number of 3. An adaptive control method of a one-input one-output type periodic signal for controlling the most main vibration (angular frequency ω) of the periodic signal f (n), the adaptive coefficient vector updating algorithm The adaptive control method for a periodic signal according to claim 1, wherein the angular frequency estimation algorithm and the adaptive signal generation algorithm are collectively represented by Formula 1. [Equation 1] [Equation 2] 4. L error signals e from at least one of the observation points affected by the periodic signal f (n).
_l (n) (1 ≦ l ≦ L, L is a natural number) as an input, and multi-input multi-output that outputs M adaptive signals y _m (n) (1 ≦ m ≦ M, M is a natural number) Is an adaptive control method of a periodic signal of a type, the adaptive coefficient vector update algorithm is expressed by Equation 3, the angular frequency estimation algorithm is expressed by either Equation 4 or Equation 5, and the adaptive signal generation algorithm is The adaptive control method for a periodic signal according to claim 1, wherein the adaptive control method is represented by Formula 6. [Equation 3] [Equation 4] [Equation 5] [Equation 6] 5. An at least one of an estimated gain A hat and an estimated phase delay Φ hat, which are estimated values of the amplitude a _k (or a _km ) and the phase φ _k (or φ _km ) and the transfer characteristic G. value has a table data stored in each section of the angular frequency [omega _k within the range of variation of the angular frequency omega _k, adaptation of periodic signals according to any one of claims 1 to 4 Control method.