JP5329995B2 - Dead time compensation device and dead time compensation method for inverter system - Google Patents

Description

  The present invention relates to a dead time compensation apparatus such as an inverter control system having a nonlinear circuit and a nonlinear load in a feedback control system to be controlled having a dead time, and a method thereof.

  In recent years, research has been conducted on contactless power supply and contactless power supply instead of control system cables for mobile objects such as factory vehicles and elevators, and some of them have been put into practical use. The contact power supply method requires maintenance and replacement due to defective brushes, wear, and sparks. However, contactless power supply eliminates the need for any contactless power supply. Research on non-contact power supply type X-ray CT apparatus for medical devices is being carried out by applying this technology.

  Realization of this non-contact power feeding type X-ray CT apparatus requires non-contact of control data transmission and reception as well as non-contact of the power supply portion. There is infrared communication as non-contact data communication, but feedback of control data is delayed by about 1 ms due to input / output AD / DA conversion and error detection. Such a system that causes a delay in input / output of the system due to information transmission delay is called a dead time system.

  FIG. 1 is a block diagram showing the overall configuration of a non-contact power supply type X-ray CT apparatus for medical equipment. As shown in FIG. 1, the non-contact power feeding type X-ray CT apparatus includes a power source 1, an inverter 2, a winding transformer 3 by non-contact power feeding, a main transformer 4, a rectifier 5, a smoothing capacitor 6, And an X-ray tube 7. The power source 1 generates a DC voltage, and the inverter 2 converts the DC voltage output from the power source 1 into AC. In the X-ray CT apparatus, since the X-ray tube 7 rotates around the subject, the main transformer 4, the rectifier 5, the smoothing capacitor 6, and the X-ray tube 7 are all configured as a rotating unit 10. The power source 1 and the inverter 2 are both configured as a fixed portion 9. A winding transformer 3 for electromagnetically coupling the winding connected to the output side of the inverter 2 of the fixed portion 9 and the winding connected to the input side of the main transformer 4 of the rotating portion 10 is They are arranged opposite to each other so as to supply required power by contact, and are provided between the fixed portion 9 and the rotating portion 10. The main transformer 4 boosts the AC voltage output from the inverter 2, and the rectifier 5 and the smoothing capacitor 6 make the output voltage of the main transformer 4 a DC high voltage. Thus, a high voltage generator for the X-ray tube 7 is formed.

  On the other hand, in the X-ray CT apparatus, the output information of the voltage (tube voltage) on both sides of the X-ray tube 7 is fed back by a communication system such as an optical fiber so that the X-ray radiation is constant, and the frequency and output of the inverter 2 are output. The voltage is controlled. As shown in FIG. 1, the feedback unit converts the voltage between the X-ray tubes 7 into a digital signal by the A / D conversion circuit 11 and transmits the voltage between the fixed unit 9 and the rotating unit 10 by the non-contact data communication device 12. Thus, the inverter control circuit 8 controls the inverter 2 to stabilize the voltage of the X-ray tube 7. Since this feedback is non-contact between the fixed unit 9 and the rotating unit 10, there is a time delay. Therefore, the inverter control circuit 8 has a dead time of 1 ms to several ms for feedback of the tube voltage.

  This dead time associated with non-contact impairs the stability of the control system. When PID control is performed on a control target with a long normal dead time, the amount of overshoot increases, and the controllability deteriorates and becomes unstable. In order to compensate for the dead time, a compensation device using the Smith method has been conventionally used in the control system.

A normal feedback control system is shown in FIG. In FIG. 2, C (s) is a controller, G (s) is an object to be controlled, e- ^sL is a ^dead time element, R (s) is a control command, U (s) is an operation amount, D (s) is disturbance, Y (s) indicates an output. In this feedback control system, when the control target is only a first-order lag and there is almost no dead time (L≈0), it can be easily controlled by PI control. However, if the dead time of the control target is large, the control is greatly affected and the control becomes difficult. The transfer function of FIG. 2 at that time is as follows.

  However, no disturbance is considered (D (s) = 0). From the above equation, the characteristic equation of this control system is

It becomes. Since this characteristic equation includes dead time, control performance deteriorates and controller design becomes difficult.

Therefore, as described above, the normal Smith method is used to compensate for this dead time, and its basic configuration is shown in FIG. In FIG. 3, P (s) represents a controlled object model, e ^−sL represents a ^dead time model, and these are collectively referred to as a Smith predictor. 3, the control object G (s) ^e -sL with dead time, the feedback through a transfer function P of the controlled object model (s), dead time element e ^-sl the control object G (s) It is configured by feeding ^back the difference of e- ^sL . The Smith method has a controlled object model and a dead time model, and performs control while always predicting a phenomenon that appears after the dead time has elapsed. That is, the operation amount U (s) is calculated by the controller C (s) based on the control amount and the set value R (s), and a change in the control amount with respect to the operation amount U (s) is predicted using the Smith predictor. Further, the operation amount U (s) is calculated in the controller C (s) based on the change amount. Therefore, the transfer function from the control command value R (s) to the output Y (s) when the Smith predictor is added to the feedback control system of FIG. 3 is as follows.

  If there is no model error, that is, if P (s) = G (s), the characteristic equation is as follows.

  The denominator polynomial of the transfer function is a characteristic polynomial that determines the stability of the control system, and is stable because no dead time element remains in the denominator polynomial (Equation 4) of (Equation 3). By using such a Smith predictor, the influence of the dead time can be eliminated from the system, the output Y (s) can accurately follow the control command R (s), and the same controller as when there is no dead time. Can be used.

  The Smith method eliminates wasted time from the control loop because the wasted time can be eliminated, and eliminates wasted time from the characteristic equation of the feedback control system. In other words, there is a feature that, for example, the dead time remains as it is, so that it is left in a complicated state.

As described above, the Smith method configuration requires an accurate control target model G (s) and a ^dead time model e- ^sL . Therefore, if the control target can be defined by a transfer function, it is possible to deal with the ^dead time element. Easy.

Japanese Patent Laid-Open No. 10-336897 Japanese Patent Laid-Open No. 05-035306 Japanese Patent Laid-Open No. 08-234802

  However, for example, a high voltage generator of an X-ray CT apparatus includes a diode rectifier circuit that is a non-linear element in a control target, and thus this control target is expressed by a transfer function or state equation required by a Smith predictor. There is a problem that it is difficult to create a model, and it is difficult to apply the model to a control device.

  In order to solve the above-described problems, an object of the present invention is to provide a dead time compensation device capable of accurately compensating for a dead time of a control target regardless of the configuration of the control target.

The dead time compensator of the invention of claim 1 is a dead time compensator for maintaining stability of a feedback control system in a control object as a control object having a dead time, wherein the control object is a non-linear element. It includes a diode rectifier circuit, to have the same response characteristics as the control target, characterized by comprising a Smith predictor which implements the model circuit comprising a diode rectifier circuit.

In dead time compensator of the invention 請 Motomeko 2, the model circuit, an analog computer operational amplifier circuit, or characterized in that it comprises a hardware circuit FPGA (Field Programmable Gate array) circuit.

In the dead time compensator according to the invention of claim 3, the model circuit receives the same signal as that to be controlled to acquire an output signal, negatively feeds back the acquired output signal to the feedback control system, and The acquired output signal is input to the same dead time element as the control target, and the difference signal between the input dead time element output signal and the output signal of the control target is negatively fed back to the feedback control system. It is characterized by that.

The dead time compensating apparatus according to claim 4 is characterized in that the control target includes an A / D conversion circuit, a D / A conversion circuit, a sampling circuit, or a communication delay circuit.

The dead time compensation method of the invention of claim 5 is a dead time compensation method for maintaining the stability of the feedback control system in the control object as a control object having the dead time, wherein the control object is a non-linear element. It includes a diode rectifier circuit, to have the same response characteristics as the control target, characterized by comprising a Smith predictor which implements the model circuit comprising a diode rectifier circuit.

In dead time compensation method of the invention 請 Motomeko 6, wherein the model circuit, an analog computer operational amplifier circuit, or characterized in that it comprises a hardware circuit FPGA (Field Programmable Gate array) circuit.

In the dead time compensation method of the invention of claim 7, the model circuit inputs the same signal as the control target, acquires an output signal, negatively feeds back the acquired output signal to the feedback control system, and The acquired output signal is input to the same dead time element as the control target, and the difference signal between the input dead time element output signal and the output signal of the control target is negatively fed back to the feedback control system. It is characterized by that.

The dead time compensation method according to claim 8 is characterized in that the control target is an A / D conversion circuit, a D / A conversion circuit, a sampling circuit, or a communication delay circuit.

  According to the present invention, a nonlinear circuit and a nonlinear load, which are nonlinear elements, are configured by an analog computer (op-amp circuit) without using a transfer function or a state equation. Realizing a Smith predictor by building a Smith method that directly incorporates a diode rectifier circuit into the model to be controlled can solve dead time compensation, such as an inverter control system, and stabilize the inverter control system. it can.

The block diagram of the high voltage generator of the non-contact electric power feeding type X-ray CT apparatus for medical equipment is shown. The block diagram of the conventional normal feedback control system is shown. The basic block diagram of the Smith method for compensating the conventional dead time is shown. The equivalent circuit of the tube voltage control circuit from the output of an inverter to an X-ray tube is shown. An integration circuit of an operational amplifier is shown. The differential circuit of an operational amplifier is shown. 5 shows an equivalent circuit obtained by dividing the circuit of FIG. 4 into three parts. The 1st equivalent circuit of the equivalent circuit divided | segmented in FIG. 6 is shown. 7a shows a model circuit of the first equivalent circuit of FIG. The 2nd equivalent circuit of the equivalent circuit divided | segmented in FIG. 6 is shown. Fig. 8b shows a model circuit of the second equivalent circuit of Fig. 8a. The 3rd equivalent circuit of the equivalent circuit divided | segmented in FIG. 6 of this invention is shown. 9a shows a model circuit of the third equivalent circuit of FIG. 9a of the present invention. FIG. 5 illustrates diode forward voltage drop compensation for the model circuit of the present invention. FIG. The X-ray tube voltage control block diagram to which the Smith method of the present invention is applied is shown. The simulation result of the time response of the ideal X-ray tube voltage control system without the conventional dead time is shown. The simulation result of the time response of the X-ray tube voltage control system with the conventional dead time 1ms is shown. A simulation result of a time response when the present invention is applied to an X-ray tube voltage control system having a dead time of 1 ms is shown.

With reference to the accompanying drawings, a non-linear circuit and a non-linear load in which transfer functions cannot be defined in the present invention are directly configured by analog computer hardware to form a Smith predictor, and a ^delay due to a ^dead time element e- ^sL . An embodiment for solving dead time compensation such as an inverter control system for improving stable performance will be described.

FIG. 4 shows an equivalent circuit of a tube voltage control circuit from the output of the inverter to the X-ray tube as a load. Equivalent circuit of the tube voltage control circuit as shown in FIG. 4 shows the inverter output voltage v _s to the voltage v _dc across the X-ray tube. For simplification, the X-ray tube portion in the equivalent circuit is replaced with a resistor _Rx . In order to create a Smith predictor, an operational circuit model of the equivalent circuit of FIG. 4 is created.

  This model circuit is created by an analog circuit using an operational amplifier. Calculation is performed by replacing the capacitor in the equivalent circuit of FIG. 4 with the analog circuit of the operational amplifier of FIG. 5a and the reactor portion with the analog circuit of the operational amplifier of FIG. 5b.

The magnifications of the voltage sensor and current sensor of the equivalent circuit are 1/2000 and 1/10000, respectively. Thus, the impedance ratio K _z the model circuit and the equivalent circuit can be calculated as follows.

Therefore, since the impedance ratio is 5, the impedance in the equivalent circuit is calculated as 5 times in the model circuit. The input voltage (inverter output voltage) v _s and the input current i _s (inverter output current) of the equivalent circuit are measurable values, and the current and voltage flowing through the equivalent circuit are calculated using these values. Thus, an X-ray tube voltage predicted value v _dc is obtained. Below, the calculation method is demonstrated.

Here, the equivalent circuit of FIG. 4 is divided into three circuit parts (first equivalent circuit 100, second equivalent circuit 200, and third equivalent circuit 300) as shown in FIG. 6, and a model circuit for each is designed. FIG. 7a shows a first equivalent circuit of the equivalent circuit, and FIG. 7b shows its model circuit. First, by calculating the voltage drops v _cs , v _rs , and v _lsr of C _s , R _s , and L _sr of the equivalent circuit, respectively, and subtracting the sum v _lcr of those voltage drops from the input voltage v _s , The voltage v _x output from this circuit is obtained. For example, the calculation of the voltage drop by C _s is performed as (Equation 6) as an equivalent circuit and (Equation 7) as its model circuit.

Using the impedance ratio _Kz , the capacitor and resistance corresponding to this circuit can be calculated as follows.

  Therefore,

Here, if C ₁ = 100 pF, R ₁ in FIG.

  Hereinafter, the calculation is performed in the same manner. A second equivalent circuit of the equivalent circuit of FIG. 6 is shown in FIG. 8a, and a model circuit designed by the operational amplifier is shown in FIG. 8b. Similarly, FIG. 9a shows a third equivalent circuit of the equivalent circuit of FIG. 6, and FIG. 9b shows a model circuit designed by the operational amplifier.

The equivalent circuit portion including the diode bridge, which is the nonlinear load in FIG. 9a, is used as it is on the model circuit with the impedance increased by a factor of five. Then, the voltages v ^to _dc at the final stage are used as the output of the model circuit. Also, since it is difficult to realize the dead time element in an analog circuit, only this part is configured by a digital control system.

  On the other hand, a diode in an electronic circuit has a dead region and a forward voltage drop of about 0.3V. That is, when the forward voltage applied between the anode and the cathode exceeds 0.3V, the current is turned on and current flows, and the voltage other than 0 to 0.3V does not turn on. Even in an actual equivalent circuit, there is a diode insensitive region, but the ratio between the diode in the electronic circuit and the diode in the power module is greatly different, which may cause an error in the model circuit.

  Therefore, in order to eliminate the influence of the insensitive region of the diode in the electronic circuit, the model circuit is changed as shown in FIG. That is, by replacing the diode bridge portion of FIG. 9b with an absolute value circuit 400 that becomes an output voltage of full-wave rectification, and further adding a load circuit 500 applying a peak hold circuit using a half-wave rectification circuit and a voltage follower circuit, Reproduce the ON / OFF of the diode. Therefore, since these compensation circuits become ideal diodes in the model circuit, no dead area is generated. Note that the forward voltage drop of the power module diode in the equivalent circuit can be corrected by subtraction before the absolute value circuit of the model circuit.

  Based on the above, the time response of the X-ray tube voltage control system was simulated. The X-ray tube voltage is controlled to change to 570 V in a stepwise manner to approach the state of emitting X-rays. The X-ray tube voltage is controlled by using a phase shift inverter, and each gain such as PI control. The one that has already been adjusted was used.

FIG. 11 shows a control circuit block diagram in which a Smith predictor is added to the X-ray tube voltage control. A non-linear control target as shown in FIGS. 9a, 9b, and 10 configured by an analog computer (op-amp circuit) is mounted on the model circuit G (s) in FIG. The dotted line in FIG. 11 is a portion configured by CPU software. In this software, a ^dead time e- ^sL is incorporated to constitute a feedback control system.

FIG. 12 shows an ideal time response simulation result when there is no dead time in the feedback of the conventional X-ray tube voltage control system. 11 and 12, v _dc ^* indicates an X-ray tube voltage command value, and v _dc indicates a feedback value of the X-ray tube voltage. Since this feedback has no AD / DA conversion and no dead time, the X-ray tube voltage follows the command value well.

Next, a simulation is performed in which communication delays and dead time elements due to AD / DA conversion are added to the control system. Sampling by AD / DA conversion was set to 20 kHz, and dead time due to communication delay was set to 1 ms. FIG. 13 shows the simulation result of the time response of the conventional X-ray tube voltage control system having a dead time of 1 ms. In FIG. 13, v _dcdelay is an X-ray tube voltage feedback value after the _dead time element. As is apparent from FIG. 13, the X-ray tube voltage v _dc oscillates largely around the command value v _dc ^* and cannot be controlled. This seems to be caused by the large proportional gain of the X-ray tube voltage control. Accordingly, since the dead time is large, a stable response cannot be obtained only by the PI control.

On the other hand, FIG. 14 shows a simulation result of time response when the Smith method of the present invention is applied to an X-ray tube voltage control system having a dead time of 1 ms. The model used the output of the above-mentioned analog calculation, and the dead time was reproduced by holding a digital value for each sampling. From the simulation result of FIG. 14, the response v _dcdelay of the X-ray tube voltage in the control using the Smith predictor is almost the same as the response in the ideal state when a delay time of 1 ms occurs but there is no _dead time. It is understood that this is effective for this dead time system.

  Considering the case where a model error has occurred in the dead time model, the dead time model was set to 1.1 ms, and a simulation was performed assuming a worst-case error of ± 20%. However, the X-ray tube voltage was found to follow the command value.

As described above, in the embodiment of the present invention, digital control such as a non-contact power feeding type X-ray CT apparatus having a dead time including an A / D conversion circuit, a D / A conversion circuit, a sampling circuit, or a communication delay circuit is provided. A dead time compensator for maintaining the stability of a feedback control system at an X-ray tube voltage, which is a control target, with a system as a control target, the control target G including nonlinear elements such as a diode rectifier circuit and an X-ray tube (S) A model circuit P (s) having the same response characteristic as that of e ^−sL is provided with a Smith predictor in which an analog computer using an operational amplifier circuit or a hardware circuit of an FPGA circuit is mounted, and the model circuit P (s) The same signal as that to be controlled is input to obtain an output signal, the obtained output signal is negatively fed back to a feedback control system, and the obtained output signal is further obtained. The inputs to the same dead time element e ^-sl and the control object, a differential signal of the output signal and the control target of the output signal of the dead time element e ^-sl that the input, thereby negatively fed back to the feedback control system .

  In this way, without using a transfer function or state equation, the nonlinear circuit and the nonlinear load are configured with an analog computer (op-amp circuit) to speed up the operation, and the diode rectifier circuit, which is a nonlinear element, can be achieved. By constructing the Smith method directly incorporated in the model to be controlled and realizing the Smith predictor, it is possible to solve the dead time compensation of the inverter control system, for example, and to stabilize the inverter control system.

  In addition, this invention is not limited to the said Example, A various deformation | transformation implementation is possible within the range of the summary of this invention. For example, the non-linear element in the control target of the present embodiment is not limited to the diode rectifier circuit but may be a transistor circuit or the like.

  An inverter or converter driven by a digital control system has a non-linear element such as a diode rectifier circuit and an active circuit portion having a plurality of control modes. In a digital control system having a dead time when power is supplied from the primary side inverter circuit to the secondary side via the transformer, a transfer function is defined when such a nonlinear element and a plurality of control modes are provided. In such a difficult part, an analog computer typified by an operational amplifier circuit, an FPGA circuit, or the like is mounted to easily compensate for dead time. Thus, the dead time compensation can be applied to a digital control system often used in an industrial system having a dead time such as an A / D conversion circuit, a D / A conversion circuit, a sampling circuit, or a communication delay circuit.

1 Power supply (DC voltage)
2 Inverter 3 Winding transformer 4 by non-contact power supply 4 Main transformer 5 Rectifier 6 Smoothing capacitor 7 X-ray tube 8 Inverter control circuit 9 Fixed part 10 Rotating part 11 A / D conversion circuit 12 Non-contact data communication device 100 First equivalent Circuit 200 Second equivalent circuit 300 Third equivalent circuit 400 Absolute value circuit (full-wave rectifier circuit)
500 Load circuit

Claims (8)

As a control object having a dead time, a dead time compensation device for maintaining the stability of the feedback control system in the controlled object,
The control target includes a diode rectifier circuit as the nonlinear element, the control to have the same response characteristics as target, dead time, characterized in that it comprises a Smith Predictor which implements the model circuit comprising a diode rectifier circuit Compensation device. The model circuit, an analog computer operational amplifier circuit, or characterized in that it comprises a hardware circuit of the FPGA circuit, dead time compensator according to claim 1. The model circuit receives the same signal as that to be controlled to acquire an output signal, negatively feeds back the acquired output signal to the feedback control system,
Further, the acquired output signal is input to the same dead time element as the control target, and the difference signal between the input dead time element output signal and the control target output signal is negatively fed back to the feedback control system. The dead time compensator according to claim 1 or 2 , characterized in that: The dead time compensation apparatus according to any one of claims 1 to 3 , wherein the control target includes an A / D conversion circuit, a D / A conversion circuit, a sampling circuit, or a communication delay circuit. As a control object having a dead time, a dead time compensation method for maintaining the stability of the feedback control system in the controlled object,
The control target includes a diode rectifier circuit as the nonlinear element, the control to have the same response characteristics as target, dead time, characterized in that it comprises a Smith Predictor which implements the model circuit comprising a diode rectifier circuit Compensation method. The model circuit, an analog computer operational amplifier circuit, or characterized in that it comprises a hardware circuit of the FPGA circuit, dead time compensation method according to claim 5. The model circuit receives the same signal as that to be controlled to acquire an output signal, negatively feeds back the acquired output signal to the feedback control system,
Further, the acquired output signal is input to the same dead time element as the control target, and the difference signal between the input dead time element output signal and the control target output signal is negatively fed back to the feedback control system. The dead time compensation method according to claim 5 or 6 , wherein: The controlled object, A / D conversion circuit, D / A conversion circuit, a sampling circuit, or being a communication delay circuit, dead time compensation method according to any one of claims 5 to 7.

Images