JP2800377B2 - Control device for constant current power supply - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial Application Field] The present invention relates to a magnetic resonance apparatus (hereinafter referred to as an MRI) which has been widely used as a medical diagnostic apparatus capable of obtaining a tomographic image of a patient without affecting a human body. It is necessary to control the current with high precision like a constant current power supply of a gradient magnetic field coil that generates a gradient magnetic field that is superimposed on a highly uniform magnetic field generated by a main coil to set a tomographic plane. The present invention relates to a control device for a constant current power supply when there is a metal coil used for a main coil and its support structure.

[Conventional technology]

In order to obtain one tomographic image, the gradient magnetic field coil of the nuclear magnetic resonance apparatus is supplied with a pulse-like current several hundred times, and generates a pulse-like magnetic field having a waveform substantially similar to this current and varying in intensity. I do. Since the waveform of the magnetic field intensity affects the image quality of the tomographic image, a constant current power supply that can control the waveform of the pulsed current and the peak value with high precision is used as the power supply of the gradient coil.

FIG. 4 is a circuit diagram of a main circuit including a gradient coil and a constant current power supply and a control device thereof. In this figure, the power converter 1 converts AC power supplied from an AC power supply 200 serving as a power system, and supplies a pulsed current having a predetermined waveform and a peak value to a gradient magnetic field coil 6 serving as a load. To control the waveform and peak value of a time-varying current with high precision, a high-frequency PWM with a carrier frequency of several tens of kHz is used.
An inverter is used. Therefore, the output voltage V ₁ of the power converter 1 as the same pulsating carrier frequency is superimposed, the low-pass filter 4, as shown to cut it is inserted.

The low-pass filter 4 includes an inductor 41 inserted in series, a capacitor 42 inserted in parallel, and a resistor 43 inserted in series with the capacitor 42. The resistor 43 is inserted as a damping resistor for preventing resonance from occurring between the capacitor 42, the inductor 41, and the respective inductances of the gradient coil 6, and the resonance causes the control system to become unstable. Has been prevented.

The magnetic flux generated by the gradient magnetic field coil 6 has a pulse-like waveform substantially proportional to the load current I, and thus generates an eddy current in the conductor near the gradient magnetic field coil 6. Superconducting MR
In the case of the I device, the conductor is a cryostat container that stores the main coil and keeps it at a very low temperature, or the aluminum winding frame of the main coil corresponds to the conductor. The support structure, the cooling plate, and the like become conductors through which the eddy current flows. Thus, in the MRI apparatus, the gradient coil 6
Is inevitably affected by the eddy current. Reference numeral 61 in the figure denotes an eddy current circuit, and an infinite number of circuits are required to exactly simulate an actual eddy current circuit. self inductance L ₃ which are magnetically coupled by the inductance M _3, resistance is obtained equivalently represented by a single eddy current circuit 61 with R _3. Of course, to improve the accuracy of equivalence, it is sufficient to simulate with more eddy current circuits such as two than one and three rather than two.
There is a problem that accuracy does not increase so much as the circuit becomes complicated.

A current divider 5 is provided in series with the gradient magnetic field coil 6, and the load current I is measured by the current divider 5. The controller uses the current set value i ^* as an input signal and subtracts the measured value i of the load current I from the current set value i ^* to obtain a deviation value e.
, And a current regulator 7 that outputs a control signal u specifying the pulse width of the PWM inverter of the power converter 1 using the deviation value e as an input signal. As the current regulator 7, a proportional-integral regulator composed of a proportional element and an integral element is usually used. The current regulator 7 is adjusted so as to have optimal control characteristics as a whole by adjusting the coefficients of the respective elements.

[Problems to be solved by the invention]

As described above, an eddy current is generated in the peripheral conductor by the pulse-shaped magnetic flux generated by the gradient magnetic field coil 6, and the load current flowing through the gradient magnetic field coil 6 and the value and distribution of the magnetic flux generated by the eddy current are caused by the eddy current. There is a problem of change. Most of the conductors in which eddy currents are generated are arranged outside the gradient magnetic field coil 6, and the effects of the eddy current can be considered only after these are assembled and completed as a magnet. It is difficult to set the control characteristics by adjusting the current regulator 7 in consideration of the above. Even if it is possible, the denominator of the transfer function is a quartic expression relating to the operator S of the positive transformation due to the influence of the eddy current circuit 61 as described later. Therefore, there is a problem that good control characteristics cannot be obtained.

SUMMARY OF THE INVENTION It is an object of the present invention to solve such a problem and to provide a control circuit of a constant current power supply capable of canceling the influence of eddy current and obtaining good control characteristics.

[Means for solving the problem]

According to an embodiment of the present invention, there is provided a control device for controlling a load current supplied from a power converter to an inductive load via a low-pass filter. A constant current comprising: a subtractor for calculating a deviation value obtained by subtracting a load current actual value from a load current set value; and a current regulator for outputting a control signal of the power converter with an output signal of the subtractor as an input signal. In the power supply control device, a dynamic compensator is inserted in series between the current regulator and the power converter, and the dynamic compensator includes a delay element having an output signal of the current regulator as an input signal. And simulating a main circuit from the power converter to a load using the output signal of the delay element and the actual value of the load current as input signals to estimate the load current and its higher-order differential value, and to perform the dynamic compensation. Output signal A state observer to force, the estimated value of the load current is estimated by the state observer, Part 1
A plurality of higher-order differential estimated values of the second order or higher and an output signal of the delay element as input signals, multiplying these input signals by different coefficients, and adding them, and using the added value as an output signal. The output signal of the arithmetic unit is fed back to the input side of the delay element, and the input signal of the state observer is used as the output signal of the dynamic compensator.

[Action]

In the configuration of the present invention, a dynamic compensator for compensating a control signal is inserted in series between the current regulator and the power converter in order to eliminate the influence of the eddy current, and the dynamic compensator is connected to the current regulator. The output signal of the delay element as an input signal, and the output signal of the delay element and the actual value of the load current are used as input signals to simulate the main circuit from the current converter to the load to calculate the load current and its higher-order differential value. A state observer for estimating and outputting a compensated control signal which is an output signal of a dynamic compensator, a load current estimated value estimated by the state observer, and a plurality of first-order or higher-order differential estimated values , And an operation unit that takes the input signal of the state observer as an input signal, multiplies these input signals by different coefficients, adds and outputs the result, and outputs the output signal of this operation unit to the input side of the delay element. By giving feedback Instead of the estimated value, the actual value of the load current, a plurality of first-order or higher-order differential values, and the sum of the control signals of the rectifier multiplied by different coefficients are added and fed back to the input side of the delay element. Approximately the same control characteristics as those described above can be obtained, and by selecting appropriate values for the aforementioned coefficients included in the transfer function that represents the control characteristics and is defined by the ratio of the load current to the control signal, The term caused by the current is canceled out, and the coefficient of each term of the sum of the powers of the Laplace transform operator S of the denominator of the transfer function can be set to have the optimum control characteristic.

EXAMPLES The present invention will be described below based on examples. FIG. 1 is a circuit diagram showing a constant current power supply and a control device thereof according to an embodiment of the present invention. The same reference numerals are given to the same components as those in FIG. 3, and a detailed description thereof will be omitted. In this figure, the difference between the two figures of the main circuit between the power converter 1 and the gradient coil 6 is that the resistor 43 is inserted in FIG. 3, whereas it is not in FIG. Accordingly, the inductor 45, the capacitor 46 and the circuit constant of the low-pass filter 40 shown in FIG.
Are different from those of the inductor 41 and the capacitor 42 of FIG. This is because the inductance and capacitance of the low-pass filter may be small when there is no braking resistance to make the amount of pulsation contained in the load current I equal. As will be described later, by adopting the present invention, the control system does not become a vibration system even if no braking resistor is inserted, so that this is omitted.

As shown in FIG. 1, the control device connects the dynamic compensator 8 in series to a current regulator 70 which uses the deviation signal e as an input signal and the signal _u1 as an output signal, and outputs the output signal of the power converter 1 A control signal u having a configuration of a control signal u as an input signal
Is a control signal that compensates for the influence of the eddy current induced by the magnetic flux generated by the gradient coil 6. The above-mentioned problem is solved by using the control signal as the control signal of the gradient coil 6.

In this figure, the circuit equations of the circuits subsequent to the gradient coil 6 are as follows. Note that S is an operator of Laplace transform.

V: terminal voltage of the gradient coil 6 I: load current of the gradient coil 6 I ₃ : eddy current of the eddy current circuit 61 L; self-inductance of the gradient coil 6 R: resistance of the gradient coil 6 M ₃ ; mutual inductance L ₃ between the gradient coil 6 and the eddy current circuit 61; eddy current circuit 61 self-inductance R _3; eddy current circuit 61 of resistance erase and voltage V and current eddy current I ₃ from the equation When the relationship of I is obtained, the following expression is obtained.

Incidentally, the eddy current I ₃ If the load current I Motomare is determined from the following equation.

Now, to facilitate conversion to the control circuit, V ₀ , I ₀ , R
_{Let 0 be} the rated voltage and rated current of the gradient magnetic field coil 6 and the rated resistance defined as the ratio between these rated voltages and the rated current, respectively. Based on these, the voltage, current and circuit constant of the preceding equation are calculated based on the following equations. To make it dimensionless. Here, the constant represented by T is a time constant having a time dimension.

This is represented by the following equation when expressed by a transfer function.

Next, the output voltage v ₁ of the power converter ₁ and the load current i
When the relation is obtained, the following equation is obtained. However, the circuit constants of the low-pass filter 4 are normalized as follows.

_{T 3 = L 3 / R 0} , K1 = R 0 / R 1, T 2 = L 3 / R 0, K 1 = R 0 / R 1 Substituting equation (5) into this equation and rearranging it results in the following equation.

FIG. 2 is a block diagram of the dynamic compensator 8 shown in FIG. In this figure, a dynamic compensator 8 includes a state observer 81 simulating the transfer coefficient F (S) of the main circuit,
It comprises an arithmetic unit 82 and a delay element 83 for performing an operation of multiplying the load current estimated value estimated from 81 and its higher-order differential value by different coefficients and adding the results.

The output signal u ₁ of the current regulator 70 in FIG. ₁ is the input signal of the dynamic compensator 8. The output signal vf (= vf1
+ Vf2) is fed back at the summation point 91 to form a signal u ₁
And is input to the state observer 81 through the delay element 83. The state observing percussion 81 has a transfer coefficient F (S) of the equation (7)
The input signal u ₁ first passes through a proportional element 111 that simulates the rectifier 1 and becomes an estimated value of the output voltage v ₁ of the rectifier 1. After this, the four integral elements connected in series
Next current estimated value _e through 31, 32, 33, 34, the _e
And the input signal of the preceding integral element 34 are multiplied by the coefficients C1 and C2 of the proportional elements 15 and 16, respectively, and an addition point 96 is added. It is an estimated value. The signals on the input side of the respective integral elements 31, 32, 33, 34 are the coefficients a ₀ , a ₁ , a of the proportional elements 18, 19, 20, 21 respectively.
₂ and a ₃ are multiplied, added at addition points 98, 99, and 100, and fed back in a form of being subtracted at addition point 92. With such a configuration of the state observer 81, the relationship between the estimated voltage value and the estimated load voltage value matches the equation (7). Deviation e ₁ between load current actual value i and load voltage estimation value
Is added to the addition points 92, 93, 94, 95 via the proportional elements 11, 12, 13, 14 based on the state observer theory, and the load voltage estimated value is compared with the load current actual value i. The coefficients g _{1 to} g ₄ are determined in order to optimize the followability, and as a result, the deviation value e ₁ becomes a small value, and the analysis of the state observer 81 is made substantially as = i. You can go there. More specifically, it can be approximated to follow with a small time constant of primary delay with respect to i. However, in the present invention, such detailed analysis is only complicated, and is omitted. Treated as i. Therefore, the coefficients g1, g2, g3,
g4 can also be ignored.

The proportional element 111 simulates the rectifier 1 as described above. If the rectifier 1 controls the load current i by controlling the firing angle of the thyristor with a thyristor rectifier, for example, the rectifier 1 Becomes a firing angle command value, and the relationship between the control signal u and the output voltage V1 of the rectifier 1 at this time becomes V ₁ ∝cos (u) and becomes non-linear. Cannot be simulated with elements. In such a case, a function calculator is used instead of the proportional element 111. In the case of the PWM inverter, since the control signal u specifies the pulse width, the rectifier 1 can be simulated by the proportional element as described above. Here, the rectifier 1 is simulated by the proportional element 111 based on this. is there.

The transfer function between the input signal u ₁ and the load current estimate in this dynamic compensator 8 becomes the following equation.

Here, a0 to a3, C1 and C2 are coefficients of the proportional element simulating the respective values of the equation (7). (7) To distinguish it from the value of the equation, it is an accurate expression to attach a hat to the sign in the same manner as the voltage or the current, but the same sign is used because the equation becomes complicated.

When the denominator of equation (8) is multiplied by (S + γ) S ⁴ and the denominator and numerator are arranged with respect to S, the following equation is obtained.

However, Q (S) = b ( C 2 S + C 1) P (S) = 2S 5 + (2γ + a 3 + k 4 K 5) S 4 + (a 3 γ + a 2 + hk 4 + γk 4 k 5 + k 3 k 5) S _{^{3 + (a 2 γ + a}} 1 + hk 3 + γk 3 k 5 + k 2 k 5) S 2 + (a 1 γ + a 0 + hk 2 + γk 2 k 5 + k 1 k 5) S + (a 0 γ + hk 1 + γk 1 k 5) P Coefficients γ, k _{1 to} K ₅ , which can be set arbitrarily, are included in the coefficients of the powers of S except for the term S ⁵ of (S). It can be optimally selected.

Incidentally, equation (9) is an equation that holds in the dynamic compensator 8 of FIG. 2, but in an actual control system, the signal u obtained by the dynamic compensator 8 is input as a control signal of the rectifier 1. Thus, the load current I is controlled. In this case, the object of the present invention can be achieved only by improving the transfer function. This will be described below.

FIG. 3 is a block diagram showing the circuit of FIG. 1 in another form. In this figure, the rectifier 1 is a proportional element 110,
The relationship between the output voltage V1 of the rectifier ₁ and the load voltage I is represented by a transfer function F (S).

As described above, the state observer 81 simulates a main circuit, and can be represented by f (S) shown by a block 121 with respect to a transfer coefficient F (S) of the main circuit shown by a block 120. Equation (9) is derived assuming that the control signal u is an input signal of the state observer 81, and this control signal u
Is also input to the rectifier 1 of the main circuit, so that eventually the transfer function between the signal u ₁ and the load current I may be approximately proportional to equation (9), and (9) I instead of the load current estimate of the denominator on the left side of the equation, and the proportional element on the right side
By using the coefficient H of the proportional element 110 instead of the coefficient h of 111, the control characteristics of the main circuit can be approximated. Therefore, when the coefficient that can be arbitrarily set for improving the control characteristics in the equation (9) is optimally set, the control characteristics of the main circuit are also optimally set.

In the equation (9), there are various methods for selecting the coefficients γ and k _{1 to} k ₄ so as to obtain the optimum control characteristics, but all of them are performed based on the automatic control theory. An example is as follows.

P (S) is selected to have a root of S = -C ₁ / C _2.
As a result, (C ₂ S + C ₁ ) in Q (S) is eliminated, and the denominator becomes a quartic equation relating to S, and the equation (9) becomes the following equation.

(10) the coefficients d ₃ to d ₀ in the denominator of the equation, for example, to be between the following values that are called coefficients of the Butterworth filter. d ₃ = 2.1, d ₂ = 3.4, d ₁ = 2.7, d ₀ = 1.
0. In the equation (9), there are six coefficients γ and k _{1 to} k ₄ , which can be arbitrarily selected for five of the coefficients of the respective terms S ^{4 to} S ⁰ as described above. Therefore, it is possible to satisfy at the same time.

By setting these coefficients γ and k _{1 to} K ₄ so that the coefficients d 3 to d 1 in the equation (10) become optimal values, the resistance for vibration suppression is connected in series with the capacitor 45 of the low-pass filter 40. Resonance will not occur even if no
The influence of the eddy current flowing in the eddy current circuit 61 on the control characteristics of the load current I is also eliminated, and the control device of the constant current power supply has good control characteristics. The current adjuster 7 adjusts the coefficients of the proportional element and the integral element included in the current adjuster 7 with respect to the optimally set control characteristics as described above. Since the above-mentioned optimization includes approximation, the above-mentioned is not strictly true in practice, so that better control characteristics can be obtained by adjustment by the current regulator 7.

〔The invention's effect〕

According to the present invention, as described above, a dynamic compensator for compensating a control signal is provided in series between a current regulator and a power converter so as to obtain a control characteristic that eliminates the influence of eddy currents and suppresses resonance. insert. The dynamic compensator simulates a delay element to which the output signal of the current regulator is input, and a main circuit from the power converter to the load using the output signal of the delay element and the actual value of the load current as input signals. A state observer for estimating a load current and its higher derivative and outputting a compensated control signal; a load current estimated by the state observer; An arithmetic unit for multiplying these input signals by different coefficients, adding the differential estimation value and the input signal of the state observer as input signals, and then adding and outputting the output signals; and outputting the output signal of the arithmetic unit to the input side of the delay element. Feedback to By adopting such a configuration, a configuration in which the load current actual value, a plurality of first-order or higher differential values thereof, and an added value obtained by multiplying the control signal of the rectifier by a different coefficient and adding the same are fed back to the input side of the delay element. Approximately the same control characteristics are obtained. The transfer function defined by the ratio of the load current to the output signal of the current regulator expresses this control characteristic. The delay element and the coefficient of the arithmetic unit included in the transfer function are set to appropriate values. In this way, the term caused by the eddy current can be canceled, and the coefficient of each term of the sum of the powers of the Laplace transform operator S of the denominator of the transfer function can be selected so as to have the optimum control characteristic. As a result, it is possible to eliminate the influence of eddy current flowing in the metal conductor placed near the load, and to suppress the resonance characteristics that resonate at a frequency determined by the capacitor and the inductor constituting the low-pass filter and the inductance of the load. it can. Therefore, the effect that the adjustment of the control characteristics by the current controller has become difficult due to the influence of the eddy current is eliminated, and an effect is obtained that the adjustment of the parameters of the current controller becomes easier. In addition, since the resonance characteristics are suppressed, the resistance for vibration damping, which is inserted in series with the capacitor of the low-pass filter, can be omitted. As a result, the reduction of the attenuation factor of the low-pass filter is eliminated, and the capacitance of the capacitor and inductor of the low-pass filter can be reduced. Therefore, the effect of reducing the size and cost of the reduced-pass filter can be obtained.

[Brief description of the drawings]

FIG. 1 is a circuit diagram showing a constant current power supply and its control device according to an embodiment of the present invention, FIG. 2 is a block diagram of a dynamic compensator shown in FIG. 1, and FIG. FIG. 4 is a block diagram showing a power supply and its control device, and FIG. 4 is a circuit diagram showing a conventional constant current power supply and its control device. 1 rectifier, 4, 40 low-pass filter, 5 shunt, 6 gradient coil (load), 61 eddy current circuit, 7, 70 current regulator, 71 Subtractor 8, Dynamic compensator 81 State observer 82 Operation unit 83 Delay element

──────────────────────────────────────────────────続 き Continued on the front page (58) Field surveyed (Int.Cl. ⁶ , DB name) G05F 1/10 H02M 7 / 21,7 / 48 A61B 5/05 G01R 33/22

Claims (1)

(57) [Claims] 1. A control device for controlling a load current supplied from a power converter to an inductive load via a low-pass filter, wherein a load current actual value as a measured value of the load current is set to a load current setting value. A control device for a constant current power supply, comprising: a subtractor that calculates a deviation value subtracted from the value; and a current regulator that outputs a control signal of the power converter with an output signal of the subtractor as an input signal. A dynamic compensator inserted in series between a regulator and the power converter, the dynamic compensator comprising a delay element having an output signal of the current regulator as an input signal, and an output signal of the delay element. A state in which the main circuit from the power converter to the load is simulated with the load current actual value as an input signal and the load current and its higher-order derivative are estimated, and the output signal of the dynamic compensator is output. Observer and this state An output value of the load current estimated by the observer, a plurality of first-order or higher-order differential estimated values thereof and the output signal of the delay element are used as input signals, and these input signals are multiplied by different coefficients. An arithmetic unit that adds the added value to the output of the delay element, feeds back the output signal of the arithmetic unit to the input side of the delay element, and outputs the input signal of the state observer to the output signal of the dynamic compensator. A control device for a constant current power supply.