JP4412048B2 - Magnetic bearing control device - Google Patents

Description

  The present invention relates to a magnetic bearing control device, and more particularly to a magnetic bearing control device suitable for supporting a rotating body.

  An example of a magnetic bearing that supports a rotating body is described in Patent Document 1. In the magnetic bearing described in this publication, an attempt is made to prevent the rotation system from becoming unstable due to the occurrence of a higher-order vibration mode when the low rigidity of the mechanical structure system is compensated by the control. Therefore, a notch filter is provided that stabilizes only the higher order mode without adversely affecting the stability of the lower order mode. This filter removes only higher-order mode vibration components from the control signal of the magnetic bearing and avoids instability of the higher-order mode.

  Another example of a magnetic bearing is described in Patent Document 2. The magnetic bearing described in this publication includes a phase shift circuit in which a secondary filter is deformed in a control circuit in order to stabilize a higher order mode. In this phase shift circuit, in a filter circuit having a second-order pole and a second-order zero, a low-order is obtained by combining a pole placed on the left half plane of the s plane and a zero placed on the right half plane of the s plane. The phase is shifted so that the higher-order vibration mode becomes stable without affecting the mode.

JP 2001-41240 A

JP-A-10-26137

  In the magnetic bearings described in Patent Document 1 and Patent Document 2, it is assumed that the target natural frequency does not basically change. For this reason, when trying to apply it to those whose natural frequency changes due to changes in the boundary conditions of the mechanical system, use a wideband filter that guarantees stability under all conditions, or conditions such as gain schedule control. Accordingly, it is necessary to change the constant of the filter.

  In the magnetic bearing that supports the rotating body, not only the natural frequency changes, but also the natural mode is divided into forward and backward due to the gyro effect. Furthermore, the natural frequency has only one value for each order when it is not rotating, but once the rotating body rotates, the natural frequency of each order becomes the natural frequency of the forward mode and the natural frequency of the backward mode. Divided. The magnetic bearings described in Patent Document 1 and Patent Document 2 are suitable for manipulating phases and gains in a narrow frequency range, and are basically applied only to modes in which the natural frequency does not change. Can not.

  For a vibration mode of a rotating body whose resonance frequency is divided into two, it is difficult to stabilize the two modes at the same time even if the filter constant is changed in a gain schedule. The magnetic bearing described in Patent Document 1 adjusts the sampling frequency of digital processing to correspond to the change of the natural frequency. However, when the width of the change of the natural frequency is large, both the forward mode and the backward mode are set. Attempting to stabilize results in instability of the lower mode.

  The present invention has been made in view of the above problems of the prior art, and an object of the present invention is to stably support a rotating body whose natural frequency can be changed by a magnetic bearing.

A feature of the present invention that achieves the above object is that in a control device for a magnetic bearing that rotates and supports a rotating body by two radial forces acting in the orthogonal x-axis and y-axis directions, the displacement signal in the x-axis direction of the rotating body And a compensation circuit for generating a compensation signal in the x-axis direction and a compensation signal in the y-axis direction from the displacement signal in the y-axis direction, and the compensation circuit separately generates the forward swing vibration and the backward swing vibration. When stopping, the poles and zeros of the transfer function of the compensation circuit are conjugated complex pairs, and when rotating, the poles and zeros of the transfer function of the compensation circuit are used. In addition, the frequency of the zero point is increased as the natural frequency of the rotating body is increased, and the pole frequency and the frequency of the zero point are decreased as the natural frequency of the rotating body is decreased for backward swinging vibration. Low and unconjugated complex It is an.

In this feature, it is preferable that the non-conjugate pole and zero complex pair of the transfer function can be set independently, and the pole and zero of the transfer function should be variable continuously or stepwise. Alternatively, the pole and zero may be obtained in advance by using the transfer function of the compensation circuit as a transfer function of a continuous time system, and the pole and zero of the continuous time system may be converted into the pole and zero of a discrete time system to enable digital processing.

In the above feature, the frequency of the pole in the compensation circuit is ωp when the rotating body is stopped, is ωpf for the forward whirling vibration when the rotating body is rotated, and the backward swinging when the rotating body is rotated. When ωpb with respect to vibration, ωpf>ωp>ωpb> 0 should be satisfied, and the frequency of the zero point in the compensation circuit is ωz when the rotating body is stopped, and the forward vibration during rotation of the rotating body It is preferable that ωzf>ωz>ωzb> 0 when ωzf with respect to rotational vibration and ωzb with respect to backward whirling vibration during rotation of the rotating body.

  According to the present invention, when the rotating body rotates to generate forward swing vibration and backward swing vibration, the magnetic bearing control circuit has independent vibration suppression means corresponding to each swing. Therefore, even if the eigenvalue of the rotating body changes due to the gyro effect, the vibration of the rotating body can be effectively suppressed by the magnetic bearing.

  Hereinafter, an embodiment of a magnetic bearing according to the present invention will be described with reference to the drawings. FIG. 2 schematically shows a control system for the magnetic bearing that supports the rotating body. In FIG. 2, a rotating body 1 is a rotor of a centrifugal compressor that rotates at a high speed, for example. Electromagnets 3 a to 3 d of a radial magnetic bearing that attracts the rotating body 1 are arranged to face each other with the rotating body 1 interposed therebetween. The attractive forces of the electromagnets 3a to 3d are controlled by feedback control, and the rotating body 1 is held in the air in a stable state. The position of the rotating body 1 is detected by the displacement sensors 2a to 2d. The displacement sensors 2a to 2d are arranged corresponding to the electromagnets 3a to 3d. The displacement signals detected by the displacement sensors 2a to 2d are input to the control circuit 5 having the compensation circuit 8 via the sensor converters 4a and 4b.

  The control circuit 5 is an arithmetic circuit that performs feedback control, and is a PID control circuit in this embodiment. The control circuit 5 calculates the current or magnetic flux supplied to the electromagnets 3a to 3d according to the magnitude and change of the displacement signal. And based on this calculation result, the command value to power amplifier 7a-7d is determined, and command value is passed to power amplifier 7a-7d via positive / negative selector 6a, 6b. The power amplifiers 7a to 7d supply power to the electromagnets 3a to 3d according to the command value from the control circuit 5, and control the current or magnetic flux of the electromagnets 3a to 3d.

  Here, the compensation circuit 8 included in the control circuit 5 is connected in series to the front stage or the rear stage of the PID control circuit. When the compensation circuit 8 is connected to the previous stage of the PID control circuit, the displacement signal detected by the displacement sensor becomes an input signal, and the output of the compensation circuit 8 becomes the input signal of the PID control circuit. When the compensation circuit 8 is connected to the subsequent stage of the PID control circuit, the output signal of the PID control circuit is input to the compensation circuit 8, and the output from the compensation circuit 8 is input to the positive / negative selectors 6a and 6b.

  The compensation circuit 8 used in the present embodiment expands the function of a compensation circuit represented by a second-order transfer function such as a conventional notch filter or phase circuit. That is, the control circuit of the bearing that supports the rotating body 1 so that the rotating body 1 can be stably operated even if the eigenvalue of the rotating body 1 is divided into two or the eigenvalue changes due to the gyro effect of the rotating body 1. 5 corresponds to the change of the eigenvalue of the rotating body.

  FIG. 1 shows a calculation block diagram of the compensation circuit 8 included in the magnetic bearing. As described above, the compensation circuit 8 is used to control the rotating body 1. The compensation circuit 8 is a signal processor that receives an x-direction signal and a y-direction signal that are orthogonal to each other as input signals. The high-order bending mode of the rotating body 1 is prevented from oscillating.

  The compensation circuit 8 is composed of a discrete-time arithmetic circuit. When a signal in the x direction is input, the input signal x is multiplied by coefficients 111 to 113, and the result is input to adders 101 to 103. On the other hand, the output signal u is multiplied by coefficients 114 and 115, and the result is input to the adders 102 and 103. The addition result of the adder 103 is delayed by one sample time by the delay unit 105 and input to the adder 102. The addition result of the adder 102 is delayed by one sample time by the delay unit 104 and input to the adder 101. The output result of the adder 101 is the output signal u.

  Signals in the y direction are also processed in the same manner as in the x direction. That is, the input signal y in the y direction is multiplied by the coefficients 211 to 213, and the respective calculation results are input to the adders 201 to 203. The output signal v is multiplied by coefficients 214 and 215 and input to the adders 202 and 203. The addition result of the adder 203 is delayed by one sample time by the delay unit 205 and input to the adder 202. The calculation result of the adder 202 is delayed by one sample time by the delay unit 204 and input to the adder 201. The output result of the adder 201 is an output signal v.

  The configuration of the compensation circuit 8 is referred to as a biquad direct type II transposed configuration, and is a secondary digital filter (IIR filter). The compensation circuit 8 can cope with the eigenvalue of the rotating body 1 at the time of stop, but cannot cope with the eigenvalue of the rotating body 1 changing or being divided into two due to the gyro effect. The reason is that a simple compensation circuit fixes the poles and zeros of the transfer function to conjugate complex pairs.

  Therefore, in this embodiment, in order to make the transfer function of the compensation circuit 8 correspond to a non-conjugated complex pair, a signal crossing path is provided between the signals in the x direction and the y direction. Thereby, the poles and zeros of the transfer function of the compensation circuit 8 become a non-conjugated complex pair, and the frequency characteristics can be set independently for the forward swing signal and the backward swing signal. Since the frequency characteristics can be set independently, it is possible to follow the change in the eigenvalue of the rotating body 1 due to the gyro effect.

  The intersection path will be described with reference to FIG. The input signal x in the x direction is also input to adders 201 to 203 in the y direction by multiplying coefficients 121 to 123. Similarly, the input signal y in the y direction is also input to adders 101 to 103 in the x direction by multiplying coefficients 221 to 223. The output signal u in the x direction is input to the adders 202 and 203 in the y direction by multiplying the coefficients 124 and 125. The output signal v in the y direction is multiplied by coefficients 224 and 225 and input to the adders 102 and 103 in the x direction.

  A method for determining each coefficient included in the compensation circuit 8 is as follows. First, the case of a continuous time system will be described. In a control circuit such as a magnetic bearing, a transfer function of a second-order filter used for stabilizing a higher-order bending mode is expressed by the following equation.

ここで、ｓは、ラプラスの演算子である。式（１）において、ωz＝ωpで、かつζz＝０のときは、いわゆるノッチフィルタであり、ωzを中心周波数とする狭い範囲の信号を遮断する。不安定化が懸念される高次曲げモードの固有振動数をωnとする。ωzをωnかあるいはその近傍に設定して、高次曲げモードの発振を抑える。図３に、補償回路８の周波数特性の例を、位相とゲインについて示す。前向きの振れ回り周波数（ω）および後ろ向きの振れ回り周波数（−ω）に対し、ωzと−ωzのところで信号を遮断している。

Here, s is a Laplace operator. In equation (1), when ωz = ωp and ζz = 0, it is a so-called notch filter, which cuts off a narrow range of signals having ωz as the center frequency. Let ωn be the natural frequency of the higher-order bending mode that may cause instability. Set ωz at or near ωn to suppress high-order bending mode oscillation. FIG. 3 shows an example of frequency characteristics of the compensation circuit 8 with respect to phase and gain. The signal is cut off at ωz and −ωz with respect to the forward swing frequency (ω) and the backward swing frequency (−ω).

  Another example of the compensation circuit 8 is a filter in which ζz <0 and ζp> 0. If this compensation circuit 8 is ωz ≧ ωp, the phase shift circuit described in Patent Document 2 is obtained. Since the phase can be continuously delayed by 2π in the frequency range of ωp to ωz, the phase characteristic for the higher-order mode can be improved. FIG. 4 shows the frequency characteristics of the phase shift circuit in terms of phase and gain.

  Such a compensation circuit 8 is used on the assumption that the natural frequency ωn of the higher-order bending mode in which destabilization is a concern does not change in principle, or even if it changes, the width is very small. . However, in the case where the gyro effect acts like a high-speed rotating body, the natural frequency of the higher-order bending mode that was ωn at the time of stop as shown in FIG. If it is divided into the natural frequency ωb of the mode, it deviates from the stabilization region achieved by the notch filter and the phase shift circuit.

  In order to avoid this problem, conventionally, the notch filter is adjusted so that the cutoff frequency is widened, and the phase shift circuit is adjusted so that the phase shift is obtained in a wide frequency range. However, in such an adjustment method, since the control performance of other modes is impaired, it is difficult to stabilize the higher-order bending mode of a rotating body having a large gyro effect. Therefore, in this embodiment, the compensation circuit 8 has a transfer function rooted at a non-conjugated complex number expressed by the following equation.

式（２）に示すように、回転体１の固有振動数ωnが前向きの固有振動数ωf（＞ωn）と後ろ向きの固有振動数ωb（＜ωn）に分かれるのに対応して、制御回路の極と零点も前向き用と後ろ向き用に分ける。回転体１が停止中の制御回路の極の周波数がωpであれば、回転中は極の周波数を前向きの振れ回り振動に対して、ωｐｆ（＞ωp）、後ろ向きの振れ回り信号に対して、ωｐｂ（＜ωp）としている。回転体１が停止中の制御回路の零点の周波数がωzであれば、回転中は零点の周波数を前向きの振れ回り振動に対して、ωｚｆ（＞ωz）、後ろ向きの振れ回り信号に対して、ωｚｂ（＜ωz）としている。

As shown in Equation (2), the natural frequency ωn of the rotating body 1 is divided into a forward natural frequency ωf (> ωn) and a backward natural frequency ωb (<ωn). Separate poles and zeros for forward and backward use. If the frequency of the pole of the control circuit in which the rotating body 1 is stopped is ωp, the frequency of the pole during rotation is ωpf (> ωp) for the forward whirling vibration, and for the backward whirling signal, ωpb (<ωp). If the frequency of the zero point of the control circuit when the rotating body 1 is stopped is ωz, the frequency of the zero point during rotation is ωzf (> ωz) for the forward whirling vibration, and for the backward whirling signal, ωzb (<ωz).

  In this transfer function, if ωzf = ωpf, ωzb = ωpb, and ζzf = ζzb = 0, a notch filter is obtained. The cutoff frequency of the filter is ωzf for the forward swing signal and ωzb for the backward swing signal. The cut-off frequency is set individually for the natural frequencies ωf and ωb of the forward and backward swings. Bandwidths ζpf and ζpb are also set individually for the forward and backward bandwidths of the cutoff frequency. FIG. 6 shows an example of frequency characteristics of the compensation circuit 8 set to the notch filter. The signal is cut off at the forward swing frequency ωzf and the backward swing frequency −ωzb. Here, ωzf ≠ ωzb.

  In the transfer function of this embodiment, if ζzf <0, ζzb <0, ζpf> 0, ζpb> 0, and ωzf ≧ ωpf, ωzb ≧ ωpb, a phase shift circuit is obtained. When this phase shift circuit is used, the parameters ωzf, ωpf, ζzf, ζpf for the forward swing signals and the parameters ωzb, ωpb, ζzb for the backward swing signals are used even if the natural frequency of the rotating body 1 is divided into two. , Ζpb can be set independently, so that an optimum frequency characteristic can be set for each of the natural frequency ωf of the forward swing and the natural frequency ωb of the backward swing.

  Specifically, the pair of ωzf and ωpf is set so as to increase as the natural frequency of the forward swing increases, and the pair of ωzb and ωpb is the natural vibration of the backward swing. Set to lower as the number drops. For example, when ωf is 10% higher and ωb is 10% lower than the natural frequency when the rotating body 1 is stopped, the pair of ωzf and ωpf is increased by 10%, and the pair of ωzb and ωpb is decreased by 10%. To do. Actually, since other vibration modes and the characteristics of other control circuits are affected, the change is not strictly 10%.

  ζzf, ζpf, ζzb, and ζpb are based on set values when the rotating body is stopped. At this time, the adjustment is performed in consideration of changes in ωzf, ωpf, ωzb, and ωpb, other vibration modes, and influences of other control circuits. FIG. 7 shows the frequency characteristics of the compensation circuit 8 set in the phase shift circuit. The compensation characteristic is shifted to the high frequency side for the forward (+) swing frequency, and to the low frequency side for the backward (−) swing frequency.

  FIGS. 6 and 7 are examples in which the compensation circuit 8 is used as a notch filter and a phase shift circuit. FIG. 8 shows frequency characteristics when the compensation circuit 8 is used as a low-pass filter. In the above equation (2), if ωzf → ∞ and ωzb → ∞, the numerator is 1, and the compensation circuit 8 is a low-pass filter. Although the eigenvalue of the rotating body 1 changes due to the gyro effect, the vibration is suppressed by ωpf> ωpb. The parameters ζpf and ζpb are also changed as appropriate.

  As described above, the characteristics of the compensation circuit 8 are determined by changing each parameter of the transfer function shown in the equation (2). That is, it depends on the setting of the coefficient 111 in FIG. The correspondence relationship between the coefficient 111 and the like in FIG. 1 and the parameter of Expression (2) will be described below.

  From equation (2), the coefficient of each term is expressed as a complex number as in equation (5).

伝達関数の各係数が複素数で表されるので、入出力信号も複素数として取り扱う。回転体では、互いに直交する２つの方向(ｘ、ｙ)の変位を利用できるので、この方向を複素平面の各座標方向とすれば、入出力信号が複素表示に対応する。つまり、伝達関数は複素数（ｘ＋ｉｙ）を入力信号にもつ複素数係数の系で表される。このとき、回転体はｘ軸の正方向からｙ軸の正方向に回転している。複素数を係数に含む補償回路８をアナログ演算装置で構成することは困難であるから、デジタル演算処理する。アナログ演算をデジタル演算に変換するために、次式で示される双一次ｓ−ｚ変換を用いる。

Since each coefficient of the transfer function is represented by a complex number, the input / output signal is also handled as a complex number. In the rotating body, displacements in two directions (x, y) orthogonal to each other can be used. Therefore, if these directions are the coordinate directions of the complex plane, the input / output signal corresponds to the complex display. That is, the transfer function is represented by a complex coefficient system having a complex number (x + iy) as an input signal. At this time, the rotating body rotates from the positive direction of the x-axis to the positive direction of the y-axis. Since it is difficult to configure the compensation circuit 8 including a complex number as a coefficient with an analog arithmetic device, digital arithmetic processing is performed. In order to convert an analog operation into a digital operation, a bilinear sz conversion represented by the following equation is used.

ここで、Ｔはディジタル処理のサンプリング時間で、（１／ｚ）は１サンプルだけ遅らせる演算子である。この双一次ｓ−ｚ変換を式（５）に適用すると、次式が得られる。

Here, T is a sampling time for digital processing, and (1 / z) is an operator for delaying by one sample. When this bilinear sz transformation is applied to Equation (5), the following equation is obtained.

式（７）は、デジタル演算用の伝達関数である。（１／ｚ）の係数ａ１〜ｅ２が、図１の各係数１１１等に対応する。この対応関係の詳細は、以下のとおりである。

Expression (7) is a transfer function for digital calculation. The (1 / z) coefficients a1 to e2 correspond to the coefficients 111 and the like in FIG. The details of this correspondence are as follows.

  The result obtained by multiplying the transfer function shown in Expression (7) by (x + ii) is defined as (u + iv). The relationship between the input signal (u + iv) and the output signal (x + iy) of the compensation circuit 8 is expressed by Expression (8).

式（８）の実数項は回転体のｘ軸方向についての演算であり、虚数項はｙ軸方向についての演算である。実数項と虚数項の各々について、右辺と左辺の対応関係を調べる。

In equation (8), the real term is an operation in the x-axis direction of the rotating body, and the imaginary term is an operation in the y-axis direction. For each of the real and imaginary terms, the correspondence between the right side and the left side is examined.

式（９）から、図１に示す演算回路が式（７）の伝達関数を満足していることがわかる。

From equation (9), it can be seen that the arithmetic circuit shown in FIG. 1 satisfies the transfer function of equation (7).

  By the way, when applying the bilinear sz transformation shown in Expression (6), the frequency axis may be distorted. In the bilinear sz conversion, the entire frequency axis of the continuous time system is pushed into half of the sampling frequency, so that when the cut-off frequency is high, the frequency is shifted in digitization. In order to prevent this problem, a frequency characteristic shift operation called prewarping is performed before bilinear sz conversion. Also in the present embodiment, the frequency characteristic is shifted by pre-warping processing.

  In the processing circuit of FIG. 1, the coefficients a1 to e2 are converted from the equations (5) and (7) based on the parameters ωzf, ωpf, ζzf, ζpf, ωzb, ωpb, ζzb, ζpb in the equation (2). Led through. Since the parameter in Expression (2) is changed according to the rotation speed of the rotating body, the coefficients a1 to e2 derived from Expression (2) are also changed according to the rotation speed. In the digital processing, the coefficient for each rotation speed is calculated off-line, and the calculation result is stored in the memory. Then, when the rotating body is actually rotated, a coefficient for each rotational speed of the rotating body is called from the memory. It is also possible to create an approximate expression for each coefficient and obtain the coefficient online while rotating the rotating body.

  Another example of the compensation circuit 8 is shown in FIG. The transfer function of the circuit shown in FIG. 9 is expressed by Expression (7). This circuit is based on a structure called Biquad direct form II. By transforming equation (9), the following equation is obtained.

ここで、

here,

である。
図９に示した回路は、上記式（１０）、式（１１）で表される演算を実行する。本実施例は、図１に示した実施例と加算器３０１、３０２、４０１、４０２の位置が相違している。

It is.
The circuit shown in FIG. 9 executes the operations represented by the above formulas (10) and (11). This embodiment is different from the embodiment shown in FIG. 1 in the positions of adders 301, 302, 401, and 402.

  That is, a signal p is output from an adder 301 to which an input signal x in the x direction and various signals described later are added. The output signal p is multiplied by coefficients 313 and 423 and input to adders 302 and 402. The signal p is also input to the delay unit 303 to generate a signal delayed by one sample time. The signal is multiplied by coefficients 311, 421, 314, and 424 and input to the adders 301, 401, 302, and 402. . The output of the delay unit 303 to which the signal p is input is also input to the delay unit 304, and as a result, a signal delayed by two sample times is generated. The signals delayed by two sample times are multiplied by coefficients 312, 422, 315, and 425 and input to adders 301, 401, 302, and 402.

  Similarly, the input signal y in the y direction and the output signal q output from the adder 401 to which various signals are added are multiplied by coefficients 413 and 323 and input to the adders 402 and 302. The signal q is also input to the delay unit 403, and a signal delayed by one sample time is generated. This delayed signal is multiplied by coefficients 411, 321, 414, 324 and input to adders 401, 301, 402, 302. The output signal of the delay unit 403 is also input to the delay unit 404, and as a result, a signal delayed by 2 sample times is generated. The signals delayed by the two sample times are multiplied by coefficients 412, 322, 415, and 325, and input to adders 401, 301, 402, and 302. The output of the adder 302 is obtained as an output u in the x direction, and the output of the adder 402 is obtained as an output v in the y direction.

  Another example of the compensation circuit 8 is shown in FIG. The transfer function of this circuit is expressed by equation (12) instead of equation (7). This circuit executes the operations represented by the equations (13) and (14).

この図１０で示した回路は、図１で示した回路を２段に構成したものである。ただし、図１では遅延器を２段に形成していたのに対し、本回路では遅延器を各構成段に１個としている。ｘ方向の入力信号ｘに係数５１１、５１２、５２１、５２２を乗じて、加算器５０１、５０２、６０１、６０２に入力する。加算器５０１の出力信号ｐに係数５１３、５２３、５１４、５１５、５２４、５２５を乗算し、加算器５０２、６０２、５０３、５０４、６０３、６０４に入力する。加算器５０２では、出力信号を遅延器５０５に入力する。そして、遅延器５０５で１サンプル時間だけ遅らせた信号を生成し、加算器５０１に入力する。

The circuit shown in FIG. 10 is obtained by configuring the circuit shown in FIG. 1 in two stages. However, in FIG. 1, the delay device is formed in two stages, whereas in this circuit, one delay device is provided in each component stage. The input signal x in the x direction is multiplied by coefficients 511, 512, 521, and 522, and input to adders 501, 502, 601, and 602. The output signal p of the adder 501 is multiplied by coefficients 513, 523, 514, 515, 524, and 525 and input to the adders 502, 602, 503, 504, 603, and 604. The adder 502 inputs the output signal to the delay unit 505. Then, a signal delayed by one sample time is generated by the delay unit 505 and input to the adder 501.

  On the other hand, the output signal u in the x direction output from the adder 503 is multiplied by coefficients 516 and 526 and input to the adders 504 and 604. The adder 504 inputs the output signal to the delay unit 506. The delay unit 506 generates a signal delayed by one sample time and inputs the signal to the adder 503. The adder 503 generates an output signal u in the x direction.

  Similarly, in the y direction, the input signal y is multiplied by coefficients 611, 612, 621, and 622 and input to the adders 601, 602, 501, and 502. The adder 601 multiplies the output signal q by the coefficients 613, 623, 614, 615, 624, and 625 and inputs the result to the adders 602, 502, 603, 604, 503, and 504. The adder 602 inputs the output to the delay unit 605. Then, a signal delayed by one sample time is generated and input to the adder 601.

  The output signal v in the y direction output from the adder 603 is multiplied by coefficients 616 and 626 and input to the adders 604 and 504. The adder 604 also receives an input processing signal in the x direction, an output processing signal, and an input processing signal in the y direction, and an addition signal of these signals is input to the delay device 606. The delay unit 606 generates a signal delayed by one sample time and inputs the signal to the adder 603. The adder 603 also receives an x-direction input processing signal and a y-direction input processing signal, and generates an output signal v in the y direction.

  Another example of the compensation circuit 8 is shown in FIG. In this circuit, the input / output relationship of the transfer function represented by Expression (12) is represented by Expression (15) and Expression (16) instead of Expression (13) and Expression (14).

この図１１で示した回路は、図９で示した回路を２段に構成したものである。ただし、この図１１でも遅延器を各構成段に１個だけ有する構成になっている。

The circuit shown in FIG. 11 is obtained by configuring the circuit shown in FIG. 9 in two stages. However, this FIG. 11 also has a configuration in which only one delay device is provided in each component stage.

  That is, the output signal p1 of the adder 701 is multiplied by the coefficients 712 and 822 and input to the adders 702 and 802. The output signal p1 of the adder 701 is also input to the delay unit 704, and a signal delayed by one sample time is generated by the delay unit 704. This signal is multiplied by coefficients 711, 821, 713, and 823 and input to adders 701, 801, 702, and 802.

  On the other hand, the output signal q 1 of the adder 801 is multiplied by coefficients 812 and 722 and input to the adders 802 and 702. The output signal q1 of the adder 801 is also input to the delay unit 804, and a signal delayed by one sample time is generated. The signals delayed by one sample time are multiplied by coefficients 811, 721, 813, and 723 and input to adders 801, 701, 802, and 702.

  The output signal p2 of the adder 702 is multiplied by coefficients 715 and 825 and input to the adders 703 and 803. The output signal p2 of the adder 702 is also input to the delay unit 705, and a signal delayed by one sample time is generated by the delay unit 705. This signal is multiplied by coefficients 714, 824, 716, and 826 and input to adders 702, 802, 703, and 803.

The output q2 of the adder 802 is multiplied by coefficients 815 and 725 and input to the adders 803 and 703. The output signal q2 of the adder 802 is also input to the delay unit 805, and a signal delayed by one sample time is generated. This signal is multiplied by coefficients 814, 724, 816, and 726 and input to adders 802, 702, 803, and 703. The adder 703 outputs an output signal u in the x direction, and the adder 803 outputs an output signal v in the y direction.

The control system block diagram of one Example of the magnetic bearing which concerns on this invention. FIG. 2 is a calculation block diagram of a filter used in the control system of FIG. The figure explaining the frequency characteristic of a notch filter. The figure explaining the frequency characteristic of a phase shift circuit. The figure explaining the change of the natural frequency by a gyro effect. The figure explaining the frequency characteristic of a notch filter. The figure explaining the frequency characteristic of a phase shift circuit. The figure explaining the frequency characteristic of a low-pass filter. FIG. 3 is a calculation block diagram of an example of a filter. The operation | movement block diagram of the other example of a filter. The operation block diagram of the further another example of a filter.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Rotor, 2a-2d ... Displacement sensor, 3a-3d ... Electromagnet, 4 ... Sensor converter, 5 ... Compensation circuit, 6a, 6b ... Positive / negative sorter, 7a-7d ... Power amplifier, 8 ... High order mode filter .

Claims (7)

In a magnetic bearing control device for rotating and supporting a rotating body by two radial forces acting in the x-axis direction and the y-axis direction orthogonal to each other, the x-axis is obtained from the displacement signal in the x-axis direction and the displacement signal in the y-axis direction of the rotating body. Compensation circuit for generating a compensation signal in the direction and a compensation signal in the y-axis direction is provided, and this compensation circuit controls the forward swing vibration and the backward swing vibration individually. The poles and zeros of the transfer function are conjugate complex pairs.When rotating, the poles and zeros of the transfer function of the compensation circuit are used.For forward swing vibration, the pole frequency and zero frequency are the natural frequencies of the rotating body. The frequency of the poles and the frequency of the zero point are reduced as the natural frequency of the rotating body decreases to make a non-conjugated complex pair. Of magnetic bearings featuring Control device. The magnetic bearing control device according to claim 1, wherein a complex pair of a non-conjugated pole and a zero point of the transfer function can be set independently.   2. The magnetic bearing control device according to claim 1, wherein the pole and zero of the transfer function are continuously variable.   The magnetic bearing control device according to claim 1, wherein the pole and zero of the transfer function are variable stepwise.   The transfer function of the compensation circuit is obtained in advance as a transfer function of a continuous time system, poles and zeros are obtained, and the poles and zeros of the continuous time system are converted into poles and zeros of a discrete time system to enable digital processing. The magnetic bearing control device according to claim 1. Frequency pole in the compensation circuit, the rotating body at the time of stopping a .omega.p, for the forward whirling vibration during rotation of the rotating body is Omegapf, to backward whirling vibration during rotation of the rotating body 2. The magnetic bearing control device according to claim 1 , wherein ωpf>ωp>ωpb> 0 when ωpb. Frequency of zero in the compensation circuit, the rotating body at the time of stopping a .omega.z, for the forward whirling vibration during rotation of the rotating body is Omegazf, to backward whirling vibration during rotation of the rotating body 2. The magnetic bearing control device according to claim 1 , wherein ωzf>ωz>ωzb> 0 when ωzb.

Images