JP3372177B2 - Filter circuit for orthogonal 2-axis signal - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a quadrature biaxial signal filter circuit suitable for controlling an inverter that outputs a sine wave voltage having an arbitrary frequency, controlling a rotary motor, or the like.

[0002]

2. Description of the Related Art In an inverter device having a variable basic AC frequency, the output is generally a rectangular wave by frequency width modulation (PWM), but since the rectangular wave has many harmonic components, noise is likely to occur, and as a countermeasure against this, a rectangular wave is generated. A method has been proposed in which the sine wave waveform is output by rounding the corners of the to eliminate harmonic components. At that time, a method is adopted in which only the harmonic components are extracted by a normal high-coefficient high-pass filter and fed back. However, since the high-pass filter is fed back without completely removing the low-frequency fundamental wave component, the difference between the high and low frequencies narrows, especially when the frequency of the fundamental wave component rises, making it more difficult to remove it. There was a problem that the performance of the.

In addition, in a conventional system using a filter of a real coefficient transfer function, two resonance frequencies having the same sign and close to each other, as well as resonance frequencies having mutually opposite signs (+,-) whose absolute values are equal or extremely close to each other are used. Was difficult to separate. For example, if there are two resonance modes with opposite signs whose absolute values are very close to each other, one of them is sufficiently damped for some reason, so do not touch that mode. If only the other mode that was set is unstable, then only the unstable mode needs to be selected. However, in the conventional filter as described above, only the mode cannot be selected, so that both modes are controlled at the same time, which may adversely affect the mode that does not need to be touched.

As a countermeasure, a rotating coordinate system in which a stationary coordinate system is rotated at the angular velocity of the mode is assumed as a tracking filter, and when viewed from the rotating coordinate system, the angular velocity component has a frequency of zero, so that it is a direct current component. There is known a filter that becomes equivalent and separates this DC component with a high-pass or low-pass filter and then inversely transforms it again into a static coordinate system. The role of such a filter is not to separate a specific AC component from multiple AC components,
Only the DC component converted into the rotating coordinate system is separated, and separation with a high-pass or low-pass filter becomes extremely easy.

In the above-mentioned inverter control device, the fundamental wave component is treated as a direct current by the coordinate conversion from the stationary coordinate system to the direct rotation so as to be removed through the high-pass filter, and then only the harmonic component is returned to the original stationary coordinate system. A method has been proposed in which coordinates are converted back to.

That is, FIG. 16 shows an example of a controller for a sine wave output inverter. AC power of commercial voltage and frequency is input from the commercial three-phase AC power supply 11, and the inverter 12
Is equivalently output AC power of an arbitrary voltage and frequency according to a command of the pulse width modulation circuit 15 and supplies it to the load 13. The resonance circuit 14 is a pulse width modulated inverter 1
Since the output of 2 is a rectangular wave, it is for forming a voltage waveform close to a sine wave by removing this high frequency component. The output voltage waveform to the load 13 is converted into a signal of two orthogonal axes by the three-phase / two-phase conversion circuit 17, and is converted from the stationary coordinate system to the rotating coordinate system by the coordinate conversion circuit 18. This conversion is performed by matrix calculation of the matrix C. Since the fundamental wave component of the signal converted into the rotating coordinate system becomes a DC component, this is removed by the high-pass filter 19 and the gain K is multiplied by the arithmetic circuit 20. And
The coordinate conversion circuit 21 converts the matrix C ^-1 into a stationary coordinate system again, and further converts it into a three-phase signal by a two-phase / three-phase conversion circuit. Then, this signal is subtracted by the adder 23 from the signal of the fundamental wave signal generation circuit 16 to feed back to the input side of the inverter.

Further, in a speed control circuit including an inverter such as a motor or a synchronous machine, it is well known that dq coordinates or quadrature coordinates called γ-δ coordinates are converted for design / study. . However, in the present invention described later,
No consideration is given to a filter circuit utilizing a complex coefficient transfer function as a two-input / two-output system and an application of a complex gain.

[0008]

In the above-mentioned example of the control device for the inverter, since the fundamental wave is an alternating current, the rotating coordinate system has a constant rotation angle, that is, is not stationary, but continues rotating. Therefore, for the coordinate conversion, as shown in FIG. 16, a matrix operation is performed in which a sine wave and a cosine wave of the rotation angle of the rotating coordinate system are always generated as a time function, and they are multiplied by two inputs and vector-composed and combined. It requires a circuit, a so-called resolver circuit. Moreover, since the resolver circuit also needs to be converted back to the original stationary coordinate system, two sets are required, and if it is configured with analog circuits, many elements are required as hardware and a complicated circuit configuration is required. Further, the digital method has a drawback that the calculation time is long.

Further, in the above-mentioned conventional method, it has been impossible to adjust the phase of a signal on two orthogonal axes to an arbitrary value regardless of the frequency like the gain. Furthermore, it has been conventionally impossible to sharply change the phase by ± 180 ° only between certain specific angular frequencies.

The present invention has been made in view of the above-mentioned circumstances, and abrupt filter characteristics can be obtained in a narrow bandwidth without performing conversion / inverse conversion from a stationary coordinate system to a rotating coordinate system. An object of the present invention is to provide a biaxial signal filter circuit having excellent filtering characteristics. In other words, the present invention provides a filter circuit, which facilitates the separation of two modes whose frequencies are very close to each other, or two modes of opposite signs whose absolute values are equal or very close to each other, for example, in a controller of an inverter. It is provided.
Furthermore, give arbitrary gain and phase to a specific frequency range, and sharpen the phase only within a certain specific angular frequency.
It is an object of the present invention to provide a quadrature biaxial signal filter circuit that can be changed by 180 °.

[0011]

According to a first aspect of the present invention, an input signal related to orthogonal two axes (the Laplace transform amounts thereof are Ux and Uy, respectively) is converted into a complex variable ( U = Ux + j).
Uy), the transfer function of the low-pass or high-pass filter with real coefficients is F (s), the central angular frequency to be passed or blocked is ω, the imaginary unit is j, and the Laplace operator s of the transfer function F (s) is Let the complex coefficient transfer function replaced by (s-jω) be F (s-jω), and the output signal related to the orthogonal two axes when the input complex variable U passes through the complex coefficient transfer function F (s-jω). In the transfer expression U · F (s−jω) = V , which is obtained by expressing the Laplace transform amounts Vx, Vy) by one complex variable ( V = Vx + jVy), the denominator of F (s−jω) is Connected with transfer elements of real coefficients so that the real part and the imaginary part of the transfer expression after multiplying both sides of the transfer expression are equal, and when ω in them is a time function It is characterized by multiplication as an input.

According to a second aspect of the present invention, the two positive or negative central angular frequencies to be passed or blocked are defined as constant or time functions ω ₁ and ω ₂ , respectively, and ω ₁ ≠ ω ₂ , Bandwidth before and after ω ₁ (2a)
In order to pass the component including and to block the component including the bandwidth (2b) before and after ω ₂ , and /
Or to change the phase with ω ₁ and ω ₂ as boundaries, k
Is an arbitrary gain or coefficient, and z is an arbitrary dimensionless number, the complex coefficient transfer function is k (s + b−jω ₂ ) / (s + a−jω ₁ ), or k (s + b−jω ₂ ) / {(s It is characterized in that it is given in the form of −jω ₁ ) ² + 2za (s−jω ₁ ) + a ² }.

According to a third aspect of the present invention, the numerator and / or the denominator of the transfer function F (s) in the first aspect are:
k is an arbitrary real number or coefficient, and 2s includes a quadratic system represented by k (s ² + 2zas + a ² ) with a bandwidth centered on the central angular frequency ω of pass or stop, and s is (s−jω ),
It is characterized in that the dimensionless constant z is arbitrarily selected so that the vicinity of the break points at both ends of the bandwidth in the gain characteristic of the frequency characteristic of the filter is closer to the ideal filter.

Further, by combining a plurality of orthogonal biaxial signal filter circuits according to any one of claims 1 to 3, a specific band in the entire frequency band is passed or blocked, or the center of the band is blocked. The phase may be changed with the angular frequency as a boundary.

The invention according to claim 4 of the present invention is claim 1
The complex coefficient transfer function of the orthogonal biaxial signal filter circuit described in any one of 1 to 3 is multiplied by a complex gain (A + jB) in which A and B are real constants, and the circuit is configured by the same means as in claim 1. It is characterized in that an arbitrary gain and phase are given to a specific frequency region.

According to a fifth aspect of the present invention, in a control device for an inverter that outputs a sine wave voltage having an arbitrary frequency, if the inverter device is a multi-phase output, two orthogonal two-axes are used.
Coordinate conversion is performed on the phase output, and the signals of the two orthogonal axes are input to the filter circuit according to any one of claims 1 to 4,
The inverter device which passes or blocks a specific band in the entire frequency band, or if the output signal is such that the phase is changed at the center angular frequency as a boundary, if the output signal is polyphase, is subjected to coordinate conversion to polyphase. It is characterized by giving feedback to.

[0017]

1 is a diagram showing a filter circuit according to a first embodiment of the present invention. This is, for example,
17 is a configuration example of a circuit in a part of the inverter control circuit shown in FIG. 16 described above, in which a low-pass filter opposite to the high-pass filter shown in FIG. That is, an orthogonal biaxial voltage signal converted from three phases to two phases is input, only the time-varying fundamental wave angular frequency ω component contained in the input signal is extracted, and other angular frequency components are sharply removed. Signal is 2 at the right end of the figure
It is an example of a tracking filter circuit that outputs so as to be an input to a conversion circuit from three phases to three phases. In FIG. 16, θ is another path input, but in FIG. 1, the angular velocity ω, which is the change rate, is another path input. The relationship between the two is θ = ∫ωdt
Is.

First, the two outputs from the conversion circuit 17 in FIG. 16 are input to the x, y biaxial input signals U = (Ux, Uy) which are orthogonal to each other, and are input to the conversion circuit 22 in the biaxial directions of the filter. The output signal is V 2 = (Vx, Vy). The filter of the real coefficient used for this filter is created based on the transfer function F ₁ (s) of the following first-order low-pass filter. F ₁ (s) ≡k / (s + a) (1)

The Laplace operator s of this transfer function F ₁ (s)
To the complex coefficient transfer function by replacing (s-jω) with F (s-
j 1), F ₁ (s−jω) = k / (s + a−jω) (2) The Laplace transform amounts of the input signals related to the x and y axes are Ux and Uy, respectively, and one complex variable U = Ux + jUy , And the output signal (the Laplace transform amounts thereof are Vx and Vy, respectively) related to the x and y axes when U passes through the complex coefficient transfer function F (s−jω) is expressed by one complex variable V = Vx + jVy. transfer expression U · F obtained by (s-jω) = V ( 3) becomes _{k (U x + jU y)} / (s + a-jω) = V x + jV y (4). Multiplying both sides of the above equation by the denominator (s + a−jω) and expanding it to correspond the real part and the imaginary part, U _x −aV _x −ωV _y = sV _x / k (5) U _y −aV _y + ωV _x = SV _y / k (6)

When this is materialized, it becomes a filter circuit as shown in FIG. That is, an integrator 1 / s is provided in the signal path of each of the x and y axes, and negative feedback with a as a which is ½ of the passband width is applied to each integrator 1 / s
The following low pass filters are configured. Then, the output of each integrator is multiplied by an angular velocity signal ω which is not a constant from another path, and then feedback that intersects between the x and y axes is fed back. Feedback from the x axis to the y axis is positive, and feedback from the y axis is the x axis. Feedback should be negative. By this feedback, the angular frequency component is selectively passed.

In this embodiment, ω = 3750 [rad /
2] shows a Bode diagram when a = k = 7.5 [rad / s]. This shows the effect of passing the band including the central angular frequency ω with a steeply selective characteristic.

In the conventional biquad filter corresponding to the filter of the configuration of FIG. 1, V = {2zωs / (s ² + 2zωs + ω ² )} U (7) is given. z is a dimensionless number that gives the steepness of the filter and corresponds to the attenuation ratio or the reciprocal of the selectivity.
Equation (7) is a real number system and should be provided separately for the two axes x and y. The concrete block diagram is shown in FIG.
8 A total of four integrators are required with two on each axis.
Further, when z is fixed, the slope of attenuation is constant in the Bode diagram, but when the horizontal axis is a linear scale, the steepness is reduced in the high frequency region, and the bandwidth is equivalently increased. In order to eliminate these drawbacks in this configuration, it is necessary to provide z as a function of ω and to provide z and ω through different paths to provide a large number of multipliers. Therefore, cost or digital calculation increases the calculation time.

On the other hand, in the complex filter of FIG. 1, the bandwidth remains unchanged at a, and it is simple and straightforward and requires about half the hardware, and digital control has the advantage that the calculation time is short. If ω is constant, the multiplier is not necessary, but in the conventional example shown in FIG.
Changes from moment to moment, so matrix operations using sine and cosine function generators and multipliers remain essential.

FIG. 3 shows a second embodiment of the present invention, which is a so-called notch filter that blocks only a signal having a specific angular frequency ω. The real coefficient transfer function that is the basis of this notch filter is F ₂ (s) ≡k (s + b) / (s + a), a, b> 0 (8). The complex coefficient transfer function calculates s of this denominator as (s-j
F ₂ (s−jω) replaced by ω). A specific connection method for this is as follows. In F ₃ , which will be described later, ω ₁ = ω ₂ = ω
Omit it because it is a special case. Further, b may be zero.

When this is materialized, it becomes a filter circuit as shown in FIG. That is, a direct connection path having a gain k is provided in the signal path of each axis, and a first-order lag element 1 / (s + a) is provided in parallel to the signal path on the input side thereof. A value (ab) obtained by subtracting b which is 1/2 of the stop band width from the corner frequency a of the next-delay element is multiplied as a gain and is returned to the direct connection path to be subtractively coupled, and each primary order A crossing path is provided to cross the axes from the output of the delay element and feed back to the primary delay element. In this crossing path, the angular frequency ω to be blocked is inserted as a variable from another path, and the crossing feedback signal from the output of the first-order lag element is multiplied by the multiplier. At the connection point, the x axis changes to the y axis. Route is addition, y
The paths from the axes to the x-axis are connected in a subtractive way.
As a result, the signals of the angular frequency components are canceled and the passage is blocked. This filter circuit can prevent only the positive or negative angular frequency from passing, and can compensate the remaining signal so that stable operation can be maintained by giving an appropriate phase. If ω is constant, no multiplier is needed and instead a block with ω as the gain or coefficient.

When b>a> 0, a pass filter is provided, and when k = a / b, the bottom of the tail of the pass characteristic is 0 [dB].
Becomes In the vicinity of the stop or pass band, only the phase change with almost no change in gain (see FIG. 4) can be used for control.

FIG. 5 shows a third embodiment of the present invention and relates to a bilinear tracking filter circuit. This is a combination of the first and second embodiments, and functionally operates in the same manner as when the above two filter circuits are connected in series. That is, a filter circuit for passing the positive or negative angular frequency component ω 1 necessary for the signal configuration for controlling the two axes of x and y and blocking the other positive or negative angular frequency component ω _2. Is. This is based on the transfer function F ₃ ≡k (s + b) / (s + a) (9) when the denominators are both first-order (bilinear), and s of the denominator is (s−jω ₁ ), (s) -J
It is expressed by the complex coefficient transfer function F ₃ ≡k (s + b−jω ₂ ) / (s + a−jω ₁ ) (10) obtained by replacing with ω ₂ ). First, F ₃ = k {s + a−jω ₁ − (a−b) + j (ω ₁ −ω ₂ )} / (s + a−jω ₁ ) = k + k [{j (ω ₁ − [omega] ₂ )-(a-b)} / (s + a-j [omega] ₁ )] (11). Therefore, the input is multiplied by k and passed through,
It can be decomposed into a first-order lag element or a low-pass filter including the complex constant of the second term. Here, the construction method of only the second term will be described.

[{B−a + j (ω ₁
-Ω ₂ )} / (s + a−jω ₁ )] is input U = (Ux + jU
If the product of y) is output V ₂ = V ₂ x + jV ₂ y, the real and imaginary parts of the complex equation after multiplying both sides by the denominator are equal, so Ux (b−a) −Uy (ω ₁ −ω ₂ ) = V ₂ x (s + a) + V ₂ yω ₁ (12) Ux (ω ₁ −ω ₂ ) + Uy (b−a) = V ₂ y (s + a) −V ₂ xω ₁ ( 13) is obtained, and from the formula (12), V ₂ x = [Ux (b−a) −Uy (ω ₁ −ω ₂ ) −V ₂ yω ₁ ] / (s + a) (14) Similarly, (13) ), V ₂ y = [Ux (ω ₁ −ω ₂ ) + Uy (b−a) + V ₂ xω ₁ ] / (s + a) (15) is obtained. Multiplying the denominator and the conjugate of the denominator in equation (10) to make the denominator a real number is not desirable because it increases the number of components, raises the order of the system, and increases the calculation time in digital control.

The connection according to the equations (14) and (15) is shown in FIG.
It is shown in the part of the dotted line other than the passage of. That is, x,
The signal path of each y-axis is provided with a direct connection path having a gain of 1 and two parallel circuits branched from the direct connection path.
One of the parallel circuits gives the amount (b−a) obtained by subtracting the corner frequency a of the low pass filter LPF connected to the rear half of the stop band width (b−a) as a gain. It is used as input to the LPF. On the other hand, (ω ₁ −ω ₂ ) is given as a gain and connected to the input part of the low-pass filter LPF so as to cross it. The connections are such that the path from the first axis to the second axis is additive and the path from the second axis to the first axis is subtractive.

The same low-pass filter LPF and ω ₁ as in FIG.
The circuit including the cross feedback of the above constitutes a bandpass filter and is the same as that described in FIG. 1, in which ω is replaced by ω ₁ . That is, an integrator is provided in the signal path of each axis, and negative feedback with a which is ½ of the pass bandwidth is applied to each of them. Then, the feedback crossing between the x and y axes from each integrator is performed so that the feedback from the x axis to the y axis is positive and the feedback from the y axis to the x axis is negative, and the feedback paths are provided to these feedback paths. Is a gain, if the angular frequency ω ₁ to be passed is a constant, it is inserted as a gain, and if it is a time function, the output of the LPF is multiplied by a multiplier. The output of the bandpass filter as a whole is additively connected to the direct connection path.

This example is particularly effective when ω ₁ ≈ω ₂ . In other words, this is a case where only the mode of ω ₁ is selected and it is not desired to touch the ω ₂ mode in the immediate vicinity. a = b = 1
FIG. 6 is a Bode diagram when 0 [rad / s], ω ₁ = 5000 [rad / s], and ω ₂ = 5100 [rad / s]. The gain difference between passing and blocking is about 40 [dB], despite the fact that the frequencies are close to each other and the bandwidth is considerable, which shows the effectiveness of this filter circuit. Further, according to this circuit configuration, only two integrators are required.

FIG. 7 is an example for the purpose of delaying the phase angle of only a certain angular frequency band by 180 degrees and not affecting the phases of other angular frequency regions in the circuit of FIG.
ω ₁ = 10000 [rad / s], ω ₂ = 40000 [rad / s], b = 0, a
= 1 [rad / s], the phase is sharply reversed over the bandwidth between the angular frequencies by setting a so small as not to oscillate. However, the gain characteristic is not completely flat. If ω ₁ and ω ₂ are reversed, the phase between them will be advanced 180 degrees, and the slope of the gain will also be reversed.

Next, an example in which a quadratic low-pass filter having a transfer function with a real number coefficient as shown in the following equation is used as a band-pass filter having a steep gain characteristic will be described. F ₄ ≡ka ² / (s ² + 2zas + a ² ) (16) The z of this filter is usually called the damping ratio.
It is usual to set to <1 and make the bump near the break point thin and high so as to have a resonance characteristic as shown in FIG. However, in such a usage method, the high-frequency side of the bump has a steep characteristic to some extent, but the low-frequency side has a drawback that the bottom is shallow. Therefore, it is common to multiply the numerator of the equation (16) by s. The shape design including the width of the bump was complicated.

In the present invention, steep characteristics can be obtained even with a second-order filter, and not only the bandwidth but also the shape of the bump can be freely selected, and the positive / negative of the pass band can be freely selected. If s in equation (16) is replaced by (s−jc) to re-arrange it with respect to an arbitrary central angular frequency c regardless of positive or negative, the relationship with inputs U and V is obtained by multiplying both sides by a denominator containing complex numbers. , Ka ² U = {(s-jω) ² + 2za (s-jω) + a ² } V = {(s ² + 2zas + a ² -ω ² ) -j ² (ωs + zaω)} (Vx + jVy) (17) The real number of this equation Part and imaginary part are equal, {ka ² Ux−2 (ωs + zaω) Vy} / (s ² + 2zas + a ² −ω ² ) = Vx (18) {ka ² Uy + 2 (ωs + zaω) Vx} / (s ² + 2zas + a ² −ω ² ) = Vy (19) The denominator and the denominator of the equations (18) and (19) are divided by a ² , and the actualized circuit is the complex second-order filter shown in FIG. still,
In the figure, k is omitted. However, ω is a constant, that is, in the case of the above-mentioned inverter, the fundamental frequency is unchanged.
If ω is variable, all ω in the figure must be given from another path and calculated by the multiplier.

Both x and y axes are secondary circuits,
By crossing between the two, it has extremely advanced functions and performance. Numerical examples are a = 250 [rad / s], ω = 5000 [rad /
FIG. 9 shows three Bode diagrams where z in the case of [s] is set to z = 0.5, 0.707, and 1.0. However, the abscissa is a linear scale of angular frequency. The bandwidth can be given by 2a, and the proper choice of z helps to improve the corners of the bandwidth. The smaller the bandwidth, the steeper the gain characteristic.

FIG. 10 shows still another embodiment of the present invention, which is an example of a circuit for giving a gain and a phase for making signals on the x and y axes necessary for control. . From the input signals from the two orthogonal axes, the required angular frequency component or band is extracted regardless of whether it is positive or negative, the phase angle required for the component or band is given and output, and the output signal is effectively used for control, etc. It is for doing. At that time, assuming a fictitious coordinate system that rotates at the angular frequency that you want to extract or remove from the stationary coordinates, extract or remove it as a DC component, and return to the original stationary coordinate system. It can be realized only by the complex filter 5 on the left. Therefore, they are the pass filters or the stop filters described in FIGS. 1 to 3 or a combination of these filters, which pass different frequencies or remove nearby disturbing frequency components. If the frequency is variable, it is necessary to introduce it from another path and calculate it by the multiplier.

As shown in FIG. 11, the complex gain circuit 6 on the right side of the drawing branches each of the x and y2 axes and supplies a constant A as a gain to one side and a constant B to the other path as a gain. And intersects between the axes to connect at the rear portion of the A. At this connection point, the path from the x-axis to the y-axis is added, and the path from the y-axis to the x-axis is subtracted.

In such a complex gain circuit,
Arbitrary phase angle tan ^-1 by arbitrarily selecting A and B
Only [B / A] makes it possible to increase or decrease the phase of the output regardless of the input frequency. The gain at this time is (A ² + B ² ) ^1/2
Is.

FIG. 12 shows an example of application of the filter circuit having the above configuration to the control device. In the figure, (A) shows the overall configuration of the control device for the inverter, and (B) shows the orthogonal 2 in (A).
A circuit configuration in which the axial filter circuit 25 is replaced with a real number system is shown. That is, the orthogonal biaxial filter circuit 25 of the present invention is
Since it is inserted between the two conversion circuits of 3 phases / 2 phases in FIG. 16, the coordinate conversion circuit is completely unnecessary. 16 is slightly different from FIG. 16 in that it is controlled by ω (t) without using the command voltage θ (v), and that the numerator is not simply s but s + b. The reason is that some delay in signal transmission is unavoidable, especially in the case of digital circuits.
The actually generated ω (t) does not always completely match the ω (t) of the command voltage v ⁺ , and may include a slight error. Therefore, it is safer to give the notch characteristics a slight width. The circuit itself is the same as that of FIG. 3, and since ω ₁ = ω ₂ is set in FIG. 5, the multiplier is halved.

FIG. 13 shows a case where the present invention is applied to, for example, a magnetic bearing control device, the controlled object controlled from two orthogonal axes is 1 / (s ² + ω ² ), and there are two resonance modes of ± jω. It is an example of a circuit for giving attenuation in the case. That is, ± jω
An example in which a control signal is generated by multiplying modes by individual gains has been shown. For example, in the case of the simplest simple vibration, two vibration modes of forward rotation + ω and backward rotation −ω are generated on the x and y planes by the restoring force. Now suppose you want to give different damping to both modes. For example, one may not want to be touched because it is given some attenuation from the other.
The attenuation of the gain of the k + is the positive mode, including them in the reverse mode of k _- shall be given the attenuation of the gain. Since both vibration modes are forward and reverse circular motions when viewed from the stationary coordinates, the damping is to apply a force in the tangential direction of the brake circle.

When the x and y input signals are displacement sensor signals to be controlled, and in the plus mode, assuming a coordinate system that rotates at ω with respect to the stationary coordinates, the plus mode is stationary with respect to the coordinate system. Since the minus mode rotates at −2ω, the two can be easily separated. Therefore, as described in the example of the inverter, the complex filter circuit of FIG. 1 can be used because it is only necessary to transform the coordinates to select the DC component and return the component to the original static coordinates again. However, since ω is a constant, no multiplier is required, and ω enters the position of the multiplier as a gain. Since only + ω is selected as the output of the filter, the tangential direction, which is the braking direction, is the delay direction at a right angle to the displacement direction, so −j
It will be multiplied by a complex gain of k ₊ . This corresponds to a specific example of the invention of claim 4.

[0042] Similarly, in the negative mode the ω in Figure 1 the opposite sign, also the inverse of + jk brake _- thus multiplying the complex gain that. In the complex gain circuit in FIG. 11, the positive mode is set to A = 0, B = + k ₊ , and the negative mode is set to A = 0, B = −k ₋ , and therefore, the frequency ω has a phase change of −90 degrees. The gain of k ₊ is set to the frequency −ω
Receives a +90 degree phase change and a gain of k ₋ .
FIG. 14 shows the dynamic characteristics of the controlled object in the example of FIG. 13 as 1 / (s ² +
FIG. 4 is a complex representation block diagram expressed as ω ² ). The action of the complex gain circuit 6 may be either advanced or delayed. Since the gain does not require dynamic characteristics, even if the phase is advanced, the ratio B / A
The gain will not increase unless both are increased while maintaining. If necessary, a plurality of such frequency (band) selection paths may be provided to provide the gain and phase necessary for stabilization.

FIG. 15 shows a case where the control target has the same dynamic characteristic as that of FIG. 14, but the actuator at the final stage which outputs the control force has a first-order delay having a large time constant T. The transfer function of this delay is shown in FIG. 1 / (1 + Ts), and the + jω mode is sufficiently damped due to some influence, but only the −jω mode is underdamped. Therefore, in this case, it suffices to select only the -jω mode component, give a constant complex gain k, and perform feedback. The Laplace transform of the equation of motion of the entire system including this feedback is s ² + ω ² = k / [(1 + Ts) (s + a + jω)] where k is an unknown complex number. (20) The characteristic equation of this system is F (s ) = (S ² + ω ² ) (1 + Ts) (s + a + jω) − k = 0 (21) When Taylor expansion is performed in the vicinity of s = −jω, approximately F (s) | _{s = −jω} = −2jωa (1- jωT) (s + jω) −k (22), so the approximate solution in the neighborhood is s = −jω + k / {-2jωa (1-jωT)} (23) Since the second term on the right side is a damping term, it is a negative real number. Is desirable. Therefore, if k = j (1-jωT) k (24), then s = -jω-k / 2ωa (25) where k is a positive real number and corresponds to a normal feedback gain.

Eventually, the expression (20) becomes s ² + ω ² = jk (1-jωT) / {(1 + Ts) (s + a + jω)} (26). Here, (1-jωT) is newly added to the numerator in order to remove the influence of the first-order lag. Since the sensor output is x + jy and the specific connection of the numerator excluding the selection filter is (x + jy) jk (1-jωT) = k [−y + xωT + j (x + yωT)] (27), the connection as shown in FIG. Becomes With this configuration, in this control circuit, ± ω
Of these, only -ω can be passed to compensate for the delay of the electromagnet and damping can be provided.

[0045]

As described in detail above, according to the present invention, high-pass or low-pass conversion is not applied to an input signal of two orthogonal axes without using conversion / inverse conversion from a stationary coordinate system to a rotating coordinate system. Various filter characteristics such as abrupt band pass characteristics can be obtained by using the filter. Therefore, for example, by using it in a feedback circuit of an inverter control device that forms a sinusoidal voltage waveform, the circuit configuration can be greatly simplified and good filter characteristics can be provided.

Particularly, according to this two-axis signal filter circuit, two modes of extremely close frequencies existing in the two-axis signal and two modes of opposite signs whose absolute values are equal or close to each other are separated. It is possible to select only one mode and estimate its state quantity or apply a predetermined control law. Therefore, with a relatively simple device or software, it is possible to create filtering and related control signals that match the operating conditions of orthogonal two-axis signals, and control the basic AC frequency variable inverter control, rotary motor control, magnetic bearing Make functions such as control more effective.
Further, in the filter circuit or circuit system for the two-axis signals of the rotating coordinate axis of the present invention, it is possible to simultaneously adjust the gain and the phase, which has been impossible with the conventional method, regardless of the frequency. It is possible to perform control suitable for operating conditions such as variable frequency inverter control, rotary motor control, magnetic bearing control, and the like.

[Brief description of drawings]

FIG. 1 is a block diagram showing a first embodiment of the present invention.

FIG. 2 is a graph showing characteristics of the circuit of the embodiment of FIG.

FIG. 3 is a block diagram showing a second embodiment of the present invention.

FIG. 4 is a graph showing characteristics of the circuit of the embodiment of FIG.

FIG. 5 is a block diagram showing a third embodiment of the present invention.

FIG. 6 is a graph showing characteristics of the circuit of the embodiment of FIG.

FIG. 7 is a graph showing characteristics of another example in which the constant is changed in the circuit of the embodiment of FIG.

FIG. 8 is a block diagram showing a fourth embodiment of the present invention.

9 is a graph showing the characteristics of the circuit of the embodiment of FIG.

FIG. 10 is a block diagram of a control device according to an embodiment of the present invention.

FIG. 11 is a block diagram of complex gain realization according to another embodiment of the present invention.

FIG. 12A is a block diagram of an inverter control device according to an embodiment of the present invention, and FIG. 12B is a block diagram of an orthogonal biaxial filter circuit.

FIG. 13 is a block diagram of a control device using a control circuit for a rotating body according to another embodiment of the present invention.

14 is a block diagram showing a transfer function of the control circuit of FIG.

FIG. 15 is a block diagram of a control device using a control circuit for a rotating body according to another embodiment of the present invention.

FIG. 16 is a block diagram of a conventional inverter control device.

FIG. 17 is a diagram showing a conventional filter circuit.

FIG. 18 is a graph showing characteristics of a conventional second-order filter circuit.

[Explanation of symbols]

5 Complex band filter 6 complex gain 12 inverter 15 Pulse width modulation circuit 16 Fundamental signal generator 23 adder 25 Quadrature 2-axis filter circuit

─────────────────────────────────────────────────── ─── Continuation of front page (58) Fields surveyed (Int.Cl. ⁷ , DB name) H03H 11/04 G05B 11/36 501 H02M 7/48

Claims (5)

(57) [Claims] 1. An input signal related to orthogonal two axes (the Laplace transform amounts thereof are Ux and Uy, respectively) is defined as one complex variable ( U = Ux).
+ JUy), the transfer function of the low-pass or high-pass filter with a real number coefficient is F (s), the central angular frequency to be passed or blocked is ω, the imaginary unit is j, and the Laplace operator s of the transfer function F (s) is Let the complex coefficient transfer function replaced by (s-jω) be F (s-jω), and the output signal related to the orthogonal two axes when the input complex variable U passes through the complex coefficient transfer function F (s-jω). In the transfer expression U · F (s−jω) = V , which is obtained by expressing the Laplace transform amounts Vx, Vy) by one complex variable ( V = Vx + jVy), the denominator of F (s−jω) is Connected with transfer elements of real coefficients so that the real part and the imaginary part of the transfer expression after multiplying both sides of the transfer expression are equal, and when ω in them is a time function A filter circuit for orthogonal two-axis signals, which is multiplied as an input. 2. The two positive or negative central angular frequencies to be passed or blocked are defined as a constant or a time function ω, respectively.
₁ and ω ₂ , and ω ₁ ≠ ω ₂ , let a component including the bandwidth (2a) before and after ω ₁ as the center pass and include the bandwidth (2b) before and after ω ₂ as the center. K to an arbitrary gain or coefficient, and z to block the component and / or change the phase at the boundaries of ω ₁ and ω _2.
Is an arbitrary dimensionless number, the complex coefficient transfer function is k (s + b-jω ₂ ) / (s + a-jω ₁ ), or k (s + b-jω ₂ ) / {(s-jω ₁ ) ² + 2za (s The filter circuit for orthogonal two-axis signals according to claim 1, wherein the filter circuit is given in the form of -jω ₁ ) + a ² }. 3. The bandwidth centered on the central angular frequency ω of 2a, where k is an arbitrary real gain or coefficient in the numerator and / or denominator of the transfer function F (s) in claim 1. Including a quadratic system represented by k (s ² + 2zas + a ² ), replacing s with (s−jω) and arbitrarily selecting the dimensionless constant z, the bandwidth of the gain characteristic of the frequency characteristic of the filter The orthogonal biaxial signal filter circuit according to claim 1, wherein the vicinity of the break points at both ends is made closer to the ideal filter. 4. The complex coefficient transfer function of the quadrature biaxial signal filter circuit according to any one of claims 1 to 3 is multiplied by a complex gain (A + jB) in which A and B are real constants. A quadrature biaxial signal filter circuit, characterized in that a circuit is constituted by similar means and an arbitrary gain and phase are given to a specific frequency region. 5. A controller for an inverter that outputs a sine wave voltage of an arbitrary frequency, if the inverter device is a polyphase output, coordinate conversion is performed on two-phase outputs of two orthogonal axes, and the signals of the two orthogonal axes are converted. 5. An output which is input to the filter circuit according to any one of claims 1 to 4 so as to pass or block a specific band of the entire frequency band, or to change the phase at the central angular frequency as a boundary. In the case of a multi-phase signal, the multi-phase coordinate conversion is performed and the signal is fed back to the inverter device.