JP6787236B2 - Power converter - Google Patents

Description

The present invention relates to a power converter.

In an inverter that supplies power by operating a switching element based on a voltage command value, the switching elements are simultaneously controlled to the off operating state for the purpose of preventing short-circuit breakage due to simultaneous conduction of the switching elements of the two arms that make up the power conversion means. There is a period to do it. This period is called dead time, but this dead time causes distortion in the output voltage of the inverter that is not in the voltage command value, and the control accuracy may deteriorate due to the error of the output voltage with respect to the voltage command value. It was.

As a method for compensating for the output voltage error of the inverter due to the dead time, there is a method of obtaining the influence of the short circuit prevention time and adding a voltage having the opposite polarity to the output current to the voltage command value. (For example, Non-Patent Document 1)

Further, there is a method in which current control is performed so that a constant DC current flows at two switching frequencies, and a compensation value for compensating for an output voltage error is obtained based on the difference between voltage commands at the two switching frequencies. (For example, Patent Document 1)

Japanese Patent Application No. 10-540322 (pages 10-16, Fig. 1)

Practice of theoretical design of AC servo system General Electronic Publishing (1997), P54-P58

However, in the conventional method shown in Non-Patent Document 1, the voltage of the opposite polarity to be added to the voltage command value is obtained on the assumption that the terminal voltage waveform changes instantaneously in a rectangular wave shape. Therefore, the voltage of the opposite polarity that is actually added due to the voltage switching characteristics of the switching element and its variation is different from the assumed rectangular wavy shape. The difference between this assumed voltage and the voltage actually added results in an error in dead time compensation, which may reduce the compensation accuracy.

Further, in the conventional method shown in Patent Document 1, by obtaining the compensation value based on the difference between the voltage command outputs at the two switching frequencies, it is possible to eliminate the variation due to the individual characteristics of the hardware. It is said. However, the two switching frequencies are limited in the range that can be set by the performance of the switching element, the current ripple, and the like, and the difference between the voltage commands required by the two switching frequencies cannot be set large. Therefore, the S / N ratio cannot be increased, and the compensation accuracy may decrease.

The present invention has been made to solve the above-mentioned problems, and is capable of suppressing the influence of current ripple and noise to obtain the compensation amount required for compensation of output voltage error due to dead time. The purpose is to provide the device.

The power conversion device according to the present invention has a switching element, a power conversion unit that supplies a current to a load, and a dead time length that determines the length of a dead time, which is a protection period for turning off the switching element. A control unit having a sequence management unit that outputs a command and a dead time addition unit that outputs a switching command that turns on and off the switching element to which the dead time is added based on the dead time length command. The sequence management unit is provided with a current value supplied from the power conversion unit to the load when the switching element is turned on and off by the switching command based on the dead time length command. The voltage command value for generating the switching command is used to obtain the compensation amount for compensating the voltage command value.

According to the power conversion device according to the present invention, it is possible to suppress the influence of current ripple and noise and obtain the compensation amount required for compensation of the output voltage error due to the dead time.

It is a block diagram which shows the outline of the power conversion system which includes the power conversion apparatus which concerns on Embodiment 1 and Embodiment 2 of this invention. It is a figure which shows a part of the hardware composition of the power conversion apparatus which concerns on Embodiment 1 to Embodiment 5 of this invention. It is a flowchart which shows an example of the execution procedure of the power conversion apparatus which concerns on Embodiment 1 of this invention. It is a flowchart which shows an example of the execution procedure of the power conversion apparatus which concerns on Embodiment 1 of this invention. It is a figure which shows the relationship between the voltage command value and the current command value when the length of the dead time which concerns on Embodiment 1 of this invention is changed. It is a figure which shows an example of the relationship between the current sampling time and the switch timing for demonstrating Embodiment 1 of this invention. It is a figure which shows an example of the relationship between the current waveform and sampling timing for demonstrating Embodiment 1 of this invention. It is a figure which shows an example of the time of the switching state and the switching-on state of the switching element for explaining Embodiment 1 of this invention. 8 is a flowchart showing an example of an execution procedure of the power conversion device according to the second embodiment of the present invention. It is a flowchart which shows an example of the execution procedure of the power conversion apparatus which concerns on Embodiment 2 of this invention. It is a block diagram which shows the outline of the control system of the system including the power conversion apparatus which concerns on Embodiment 3 and Embodiment 5 of this invention. It is a block diagram which shows the outline of the control system of the system including the power conversion apparatus which concerns on Embodiment 4 and Embodiment 5 of this invention.

Embodiment 1.
FIG. 1 is a configuration diagram of a power conversion system to which the power conversion device according to the first embodiment for carrying out the present invention is applied.

The entire power conversion system shown in FIG. 1 includes a power conversion device 100, a DC power supply 1, and a current sensor 3 according to the present embodiment. Further, for convenience of explanation, the load 4 is shown in FIG. The power conversion device 100 has a function of an inverter that converts DC power into AC power, converts DC power supplied from DC power source 1 which is an input power source, and supplies it to load 4. The current sensor 3 outputs a current detection signal Iout based on the value of the current supplied from the power converter 100 to the load 4. In the present embodiment, the current sensor 3 is provided separately from the power conversion device 100, but it may be built in the power conversion device 100.

The power conversion device 100 in the present embodiment is composed of a power conversion unit 2 and a control unit 6.

The power conversion unit 2 is composed of two switching elements 20 and 21 on the P side and the N side, and they are connected in series. The P-side switching element 20 and the N-side switching element 21 are composed of IGBTs (Insulated Gate Bipolar Transistors) 201 and 211 and FWD (Freewheeling Diodes) 202 and 212, respectively. Further, the terminals at both ends of the switching circuit are connected to an external DC power supply 1 to form a two-level single-phase half-bridge circuit. The P-side switching element 20 and the N-side switching element 21 are also simply referred to as switching elements.

In the present embodiment, the IGBTs 201 and 211 are used as the switching elements, but other switching elements such as MOSFETs (Metal-oxide-smiconductor field-effect transistors) can also be used. Further, although a two-level single-phase half-bridge circuit will be described as an example, the method of the present invention can be applied regardless of the number of levels and the number of phases. For convenience of explanation, it is assumed that the DC power supply 1 outputs −Vdc / 2 [V] and Vdc / 2 [V] with the position of X shown in FIG. 1 as the reference potential.

The switching of the switching elements 20 and 21 and the dead time to be added will be described below. In the circuit of this embodiment, the P-side switching element 20 and the N-side switching element 21 are complementarily turned on and off to change the terminal voltage of the output terminal for supplying power from the switching circuit to the load 4. Let me. In the present embodiment, the average terminal voltage is changed and power is supplied to the load 4 by changing the ratio of the on and off states based on the switching command Sw2 * from the control unit 6 described later.

The P-side and N-side switching elements 20 and 21 switch between the on and off states at high speed, but in reality, it takes a certain amount of time for the P-side and N-side switching elements 20 and 21 to switch between the on and off states. It takes. Therefore, if the P-side and N-side switching elements 20 and 21 are controlled without considering the time required for switching, a state in which both the P-side switching element 20 and the N-side switching element 21 are turned on appears. As a result, an electrical short circuit occurs, an excessive current flows through the circuit, and the circuit is damaged.

Therefore, the switching command Sw2 * is provided with a protection period called a dead time to prevent a short circuit. During this dead time, the P-side and N-side switching elements 20 and 21 are both turned off at the timing of complementary on and off switching. Further, since a signal such as the switching command Sw2 * passes through an insulating element such as a photocoupler, a transmission delay often occurs. Further, the transmission delay may vary due to individual differences in the circuit elements of the signal transmission path including the insulating element. These transmission delays are also factors that cause short circuits of the P-side and N-side switching elements 20 and 21. Therefore, the switching command Sw2 * is provided with a dead time in consideration of the above-mentioned transmission delay.

The control unit 6 inputs the current detection signal Iout from the current sensor 3 and outputs the switching command Sw2 *. FIG. 2 is a diagram showing a hardware configuration of the control unit 6. The control unit 6 is composed of a processor 601 and a storage device 602, and the functions of each unit described later can be realized by these. For example, in the storage device 602, in addition to the control program, data obtained by calculations such as a voltage command value Vi *, a current command value I *, and a dead time length command Td *, which will be described later, are stored and stored. .. In addition, data such as information to be temporarily stored in the process of the processing is stored and stored.

The processor 601 reads and executes a program stored in the storage device 602. In addition, information to be temporarily stored is written or read in the process of executing the operation of the present embodiment.

The storage device 602 includes a volatile storage device such as a random access memory and a non-volatile auxiliary storage device such as a flash memory. Further, as the non-volatile auxiliary storage device, an auxiliary storage device such as a hard disk may be provided instead of the flash memory.

The control unit 6 will be described with reference to the system configuration diagram of FIG. The control unit 6 includes a sequence management unit 61, a voltage command generation unit 62, a PWM processing unit 63, and a dead time addition unit (hereinafter referred to as a Td addition unit) 64.

The sequence management unit 61 outputs the current command value I * to the voltage command generation unit 62. Further, the dead time length command Td * is output to the Td addition unit 64.

The voltage command generation unit 62 performs current control processing using the current command value I * from the sequence management unit 61 and the current detection signal Iout from the current sensor 3, and outputs the voltage command value Vi * to the PWM processing unit 63 described later. Output. In the current control process, for example, the voltage command value Vi * is generated so that the deviation between the current command value I * and the current detection signal Iout becomes zero.

The PWM processing unit 63 divides, for example, the voltage command value Vi * by the DC power supply voltage value Vdc / 2 [V] with respect to the voltage command value Vi * from the voltage command generation unit 62, and converts it into a modulation factor command. Then, PWM processing is performed by comparing the magnitude with the triangular wave and the sawtooth wave, and the switching command Sw1 * is output.

The Td addition unit 64 adds a dead time to the switching command Sw1 * according to the dead time length command Td * from the sequence management unit 61. Then, the switching command Sw2 * with the dead time added is output to the power conversion unit 2.

Here, the error caused by the dead time is analyzed. During the added dead time, the IGBTs 201 and 211 of the IGBTs 201 and 211 and the FWD 202 and 212 constituting the switching elements 20 and 21 are turned off. Therefore, at least a part of the output current of the power conversion unit 2 passes through the FWD 202 and 212. Whether the output current passes through the FWD202 on the P side or the FWD212 on the N side is determined based on the polarity of the power conversion unit 2.

For example, the case where the direction of the polarity of the output current of the power conversion unit 2 coincides with the direction d from the output terminal to the load 4 in FIG. 1 is positive. At this time, if the direction of the polarity of the output current is positive, the output current passes through the FWD 212 on the N side, and the voltage of the output terminal of the power conversion unit 2 becomes −Vdc / 2 [V] through the FWD 212 on the N side. Become. When the polar direction of the output current is negative, the current passes through the FWD202 on the P side, and the output voltage of the power conversion unit 2 becomes Vdc / 2 [V] through the FWD202 on the P side.

The output voltage error of the power conversion unit 2 during the dead time occurs in the vicinity of the switching operation of the switching elements 20 and 21, for example, several [μsec] in the form of a pulse for an extremely short time. As described above, the switching command Sw2 * is obtained by adding a dead time to the switching command Sw1 * obtained through the PWM processing in the PWM processing unit 63. Therefore, when the pulsed output voltage error is converted into the average value within one cycle section of the carrier with the carrier as a reference, it can be expressed by the equation (1). The output voltage error represented by the equation (1) is hereinafter referred to as a dead time voltage error.

In the equation (1), Vtd is the dead time voltage error, fc is the carrier frequency [Hz], Td is the length of the dead time to be added [sec], Vdc is the DC link voltage that is the input of the power conversion unit 2, and sign ( i) indicates the polarity of the output current. The length of the dead time to be added is assumed to be ideally added as a square wave. Further, the polarity of the output current is positive when it coincides with the direction d from the power conversion unit 2 toward the load 4, 1 when it is positive, and -1 when it is negative.

As a method of compensating for the dead time voltage error, it is generally practiced to add a voltage having the opposite polarity to the dead time voltage error to the voltage command value. This voltage compensation control is called dead time compensation.

Dead time compensation works generally well, but compensation accuracy may be reduced for several reasons:

The first reason for the decrease in compensation accuracy is that the terminal voltage waveform of the output terminal of the power conversion unit 2 does not become an ideal step waveform. As described above, it takes a certain amount of time for the switching elements 20 and 21 to switch between the on and off states. The dead time voltage error of Eq. (1) is obtained by assuming an ideal waveform in which the terminal voltage instantly changes to a step waveform. Therefore, the difference from the ideal waveform results in an error in dead time compensation. Further, the shape of the terminal voltage waveform changes according to the output current of the power conversion unit 2. This change in the terminal voltage waveform also causes an error in dead time compensation.

The second reason why the compensation accuracy is lowered is that the transmission delay of the switching command Sw2 * generated from the control unit 6 to the switching elements 20 and 21 varies. The switching command Sw2 * is generated on the assumption that the length of the dead time is a square wave. Therefore, the variation in the transmission delay of the switching command Sw2 * causes a difference between the assumed dead time length and the actual dead time length. Therefore, an error of dead time compensation will be introduced.

The third reason for the decrease in compensation accuracy is the charging and discharging of capacitors in the snubber circuit (not shown) parallel to the FWD 201 and 211 that configure the switching elements 20 and 21 installed in the overvoltage protection of the switching elements 20 and 21. is there. The shape of the terminal voltage waveform during the dead time changes due to the charging and discharging of the capacitor in the snubber circuit. This is also an error in dead time compensation. Since the charging / discharging operation is performed by a current, the shape of the terminal voltage waveform changes depending on not only the current of the load 4 but also the length of the dead time actually applied.

The fourth reason for the decrease in compensation accuracy is that charging and discharging of stray capacitances in the vicinity of the switching elements 20 and 21 have an effect as in the case of the capacitor of the snubber circuit.

As described above, the dead time voltage error generated by a plurality of factors can be collectively expressed as a coefficient α in the equation (2). Highly accurate dead time compensation can be realized by adding a voltage value of opposite polarity to the voltage command value based on the dead time voltage error represented by the equation (2). In other words, in order to realize highly accurate dead time compensation, it is necessary to obtain the coefficient α with high accuracy. Therefore, in the present embodiment, the acquisition of the coefficient α with high accuracy is realized by performing the operation described later.

The operation related to the present embodiment will be described. FIG. 3 is a flowchart showing an example of the execution operation of the control unit 6 according to the present embodiment. In the present invention, power conversion control is performed by setting a plurality of dead time lengths. At this time, the current of the load 4 is controlled, and the voltage command Vi * corresponding to the dead time length is recorded. Here, if the general dead time compensation for adding the voltage of the opposite polarity of the formula (1) to the voltage command Vi * is turned off, the voltage command Vi * includes the dead time voltage error shown by the formula (2). The result is. It is assumed that the dead time length Td and the carrier frequency fc are known values, and the DC link voltage Vdc can be obtained by measurement. In this case, the coefficient α of the equation (2) can be obtained by obtaining the slope of each voltage command corresponding to the length of each dead time based on the equation (2). In the present embodiment, the configuration when two types of dead time lengths are set is described, and the flowchart of FIG. 3 corresponds to this. The inclination can be calculated based on the difference between the two types of voltage commands corresponding to the two types of dead time lengths.

In the current command step S101, the current command value I * is output from the sequence management unit 61 to the voltage command generation unit 62. The current command value I * at this time is an arbitrarily set value. The voltage command generation unit 62 performs the above-mentioned current control process based on the current command value I * and the current detection signal Iout from the current sensor 3. Then, the voltage command value Vi * is output to the PWM processing unit 63. Further, the PWM processing unit 63 generates a switching command Sw1 * in the Td addition unit 64 based on the voltage command value Vi * from the voltage command generation unit 62.

In the first dead time length command step S102, the dead time length command Td * is output from the sequence management unit 61 to the Td addition unit 64. In the present embodiment, Td1 * is output as a certain value of the dead time length command Td *. The Td addition unit 64 adds a dead time to the switching command Sw1 * based on the dead time length command Td1 *. Then, the switching command Sw2 * after the dead time is added is output to the power conversion unit 2.

By the operation of the steps up to this point, the power conversion unit 2 operates the switching elements 20 and 21 on and off based on the switching command Sw2 *. The power conversion unit 2 outputs the power supplied from the DC power supply 1 to the load 4 by the on and off operations of the switching elements 20 and 21.

The current supplied from the power conversion unit 2 to the load 4 at this time is detected by the current sensor 3 installed on the lead wire connecting the power conversion unit 2 and the load 4. The current sensor 3 outputs the current detection signal Iout to the voltage command generation unit 62.

In the first voltage command acquisition step S103, the voltage command value Vi * is stored in the storage device 602 as V1 *. The voltage command value V1 * is a voltage command value newly output by the above-mentioned current control process based on the dead time length command Td1 *.

The voltage command value V1 * is generated by the voltage command generation unit 62 based on the current command value I * and the current detection signal Iout as described above. At this time, since dead time compensation is not performed, as a result of compensation by the current control process, the voltage drop due to the impedance of the load 4 and the dead time voltage error component are input to the sequence management unit 61. include.

In the second dead time length command step S104, the dead time length command Td2 * is output from the sequence management unit 61 to the Td addition unit 64. The dead time length command Td2 * is a different value from the dead time length command Td1 *. The Td addition unit 64 adds a dead time to the switching command Sw1 * based on the dead time length command Td2 *. Then, the switching command Sw2 * after the dead time is added is output to the power conversion unit 2.

The switching elements 20 and 21 of the power conversion unit 2 are turned on and off based on the switching command Sw2 *. The power conversion unit 2 converts the DC power supplied from the DC power supply 1 into AC power by the on and off operations of the switching elements 20 and 21, and outputs the DC power to the load 4.

At this time, the current supplied from the power conversion unit 2 to the load 4 is detected by the current sensor 3 installed on the lead wire connecting the power conversion unit 2 and the load 4. The current sensor 3 outputs the current detection signal Iout to the voltage command generation unit 62.

In the second voltage command acquisition step S105, the newly acquired voltage command value Vi * is stored in the storage device 602 as V2 *. The voltage command value V2 * is a voltage command value newly output by the above-mentioned current control process based on the dead time length command Td2 *.

The voltage command value V2 * is generated by the voltage command generation unit 62 based on the current command value I * and the current detection signal Iout as described above. At this time, since the dead time compensation is not performed, as a result of the compensation by the current control process, the voltage drop due to the impedance of the load 4 and the dead time voltage error component are input to the sequence management unit 61. include.

In the coefficient α acquisition step S106, the coefficient α is acquired based on the voltage command values V1 * and V2 * stored in the storage device 602. This operation is performed by processor 601.

The method of obtaining the coefficient α will be described. The voltage command values V1 * and V2 * obtained in the steps S101 to S105 described above are values generated by the current control process using the detected current signal Iout and the current command value I *. The voltage command values V1 * and V2 * are the carrier frequency fc [Hz], the dead time length command [sec], and the DC link that is input to the power conversion unit 2 based on the above equation (2). It can be expressed using the voltage Vdc and the polarity sign (i) of the output current.

Equation (3) is an equation expressing the voltage command value V1 * obtained when the dead time length command is Td1 * based on equation (2). Further, the equation (4) is an equation expressing the voltage command value V2 * obtained when the dead time length command is Td2 * based on the equation (2). Each Z · i shows a voltage drop component due to the impedance of the load 4.

In the coefficient α acquisition step S106, first, the difference between the equation (3) of the voltage command value V1 * and the equation (4) of the voltage command value V2 * is calculated. That is, by finding the difference between the two voltage command values V1 * and V2 *, the voltage drop components Z · i due to the impedance of the load 4 on the right side of each of the equations (3) and (4) are canceled out, and the voltage drop The influence of the components Z and i can be eliminated. Further, although the description in the equations (3) and (4) is omitted, the on-voltage components of the switching elements 20 and 21 can be offset by similarly calculating the difference. Therefore, only the dead time voltage error can be extracted with high accuracy. At this time, since the values other than the coefficient α are known values, the coefficient α can be obtained as shown in the equation (5).

In this way, when the coefficient α is acquired in the coefficient α acquisition step S106, the operation is terminated. The above is the basic execution operation for obtaining the coefficient α in the present embodiment. As a result, an accurate coefficient α can be obtained. The values of the dead time length commands Td1 * and Td2 * may be set arbitrarily, but it is better to set the difference between the values of Td1 * and Td2 * to be larger within the settable range. Better.

Further, in order to realize more accurate dead time compensation, the current command change step S107 may be provided as shown in FIG. In this case, after the coefficient α is acquired in the coefficient α acquisition step S106, the current command is changed to an arbitrary current command I * in the current command change step S107. Then, the changed current command I * is output in the current command step S101, and the operations of S101 to S106 described above are repeated. FIG. 4 shows a flow in which measurement is performed with two types of dead time length commands for an arbitrary current command I *. The dead time length command is Td1 *, and a plurality of current commands I * are used. Then, the dead time length command may be Td2 *, and the measurement may be performed on a plurality of I * in the same manner. Depending on the type of processor 601, there is a model in which it is necessary to stop / restart the processor 601 itself when changing the dead time length command. In this case, it may take time to change the measurement conditions, but if the above order is used, the time required to change the measurement conditions can be reduced. The voltage commands V1 * and V2 * used for the calculation are stored in the storage device 602, and the calculation of the equation (5) is performed after the voltage command values are prepared, so that the essential effect of the present invention is not affected. ..

As a result, the coefficient α for each current command value I * can be obtained. That is, it can be held as a table for each output current. By having a coefficient α for each output current, it is possible to perform dead time compensation in consideration of a change in the dead time voltage according to the output current. Further, the dead time compensation can improve the output voltage accuracy of the power conversion unit 2. Further, since the coefficient α can be automatically measured by using only the current sensor 3 attached to the power conversion unit 2, it is possible to significantly reduce the labor as compared with the manual measurement.

In the current command change step S107, the current command value I * is changed so as to be uniquely increased or decreased, and by repeating steps S101 to S106, a coefficient α for each current command value I * that changes continuously is acquired. May be good. Further, it may be changed at random to acquire each coefficient α and acquire a table. The number of repetitions may be arbitrarily determined according to the table to be acquired. The execution operation when acquiring the table ends after acquiring the table.

FIG. 5 shows the relationship between the voltage command value Vi * and the current command value I *, and is also a diagram illustrating equations (3) to (5). The horizontal axis is the current command value I *, the vertical axis is the voltage command value Vi *, and the voltage command value V1 * obtained when the dead time length command is Td1 * is indicated by ◆ for each current command value I *. I'm plotting. Further, when the dead time length command is Td2 *, the voltage command value V2 * is similarly plotted with Δ for each current command value I *. Note that, for convenience of explanation, FIG. 5 shows only an example showing the case where direct current is energized, but a similar schematic diagram can be shown by expressing the effective value even if the method is changed to another energization method such as alternating current energization. It is possible. The method for obtaining the basic coefficient α is also the same.

Similar voltage drop components Z and i exist in the voltage command values V1 * and V2 * obtained by the same current command value I *, respectively. This voltage drop component Z · i can be removed by taking the difference between the voltage command values V1 * and V2 * obtained for each current command value I *. As a result, only the difference in voltage between Δ and ◆ in FIG. 5 can be obtained. Then, it can be seen from the equation (5) that the coefficient α can be obtained from the difference between the voltage command values V1 * and V2 *.

Further, depending on the type of the DC power supply 1, when a current is passed through the load 4, the DC link voltage Vdc may fluctuate due to the power consumption of the load 4. For example, a diode converter that rectifies and supplies AC power from a power supply system is equivalent. At this time, even if each DC link voltage value Vdc when the dead time length command Td * is changed is recorded, and the coefficient α is calculated using the equation (6) instead of the equation (5). Good. Note that the description of the Vdc detection location is omitted in FIG.

As described above, in the present embodiment, the coefficient α can be acquired with high accuracy by the two types of voltage command values V1 * and V2 * having different dead times. Further, by performing dead time compensation using the obtained coefficient α, it is possible to suppress a decrease in compensation accuracy. Further, by acquiring a table of the coefficient α corresponding to the output current, the coefficient α selected according to the output current can be used for dead time compensation, and more accurate compensation becomes possible. The dead time compensation can be realized, for example, by providing the control unit 6 with a compensation unit (not shown) and performing the dead time compensation as described above in the compensation unit.

The advantage of measuring the coefficient α by changing the length of the dead time will be described in more detail.

When a current is passed through the load 4 by using the power conversion unit 2, a voltage having a shape similar to a square wave is applied to the load 4 by PWM control. Therefore, the output current from the power conversion unit 2 includes a current ripple. This current ripple may reduce the measurement accuracy of α.

The first reason for the decrease in measurement accuracy of α is that the polarities of the current sampled by the current ripple and the current near the switching are different. As a method of avoiding this current ripple, for example, when the carrier wave used for comparison when performing PWM control is a triangular wave carrier, there is a method of performing current sampling at a specific timing such as the peak / valley timing.

However, depending on the electrical time constant of the load 4, the current ripple becomes remarkable. As a result, the magnitude and polarity of the current value at the switching moment and the sample current value may differ, especially near zero current. This is particularly remarkable when the carrier frequency of the triangular wave carrier is lowered.

FIG. 6 shows the relationship between the current sampling time and the switch timing. An example in which the dead time compensation accuracy is lowered will be described using this.

FIG. 6A shows a current waveform when energized at a low carrier frequency with time on the horizontal axis. FIG. 6B shows a current waveform when energized at a high carrier frequency with time on the horizontal axis. In both current waveforms, the timing marked with a circle indicates the sampling timing of the current synchronized with the apex of the triangular wave carrier, and the timing marked with ● indicates the switching timing.

The vertical fluctuation of the value of the current waveform in FIG. 6 indicates that the current ripple is occurring. Although current ripples occur in the current waveforms of both FIGS. 6A and 6B, it can be seen that the current ripples are larger at the low carrier frequency (A) than at the high carrier frequency (B). Due to this large current ripple, the polarities of the sampled current value and the current value in the vicinity of switching may differ, for example, as shown by the dotted line in FIG. 6 (A).

In the measurement of the coefficient α, the current control system is mostly composed of a PI controller. That is, in the measurement of the coefficient α, the integrator of the PI controller is in charge of the dead time voltage error. Therefore, when a phenomenon in which the current polarities are different occurs, the current control is greatly disturbed. Then, the voltage command value Vi * including this disturbance is output.

As a result, an error occurs in the calculation of the coefficient α using the above-mentioned voltage command value Vi *. The phenomenon that an error occurs in this coefficient α is remarkable in the vicinity of the current zero cross. Especially in the vicinity of the current zero cross, the current ripple becomes larger than the main current, which is the output current output to the load 4 along the current command value Vi *.

The second reason for the decrease in the measurement accuracy of α is that the current ripple becomes remarkable and a discrepancy occurs between the sampled current value and the average current value flowing through the load 4. In this case, in the conventional method using the voltage command at the two switching frequencies, the voltage drop at the load 4 is different. This difference may flow into the calculation result of the coefficient α and cause an error.

An example thereof will be described with reference to FIGS. 7A and 7B. FIG. 7A shows an example of a current waveform at a low carrier where the carrier frequency is lower than the electric time constant of the load 4 with time on the horizontal axis. FIG. 7B shows an example of a current waveform at the time of high carrier, in which time is taken on the horizontal axis and the carrier frequency is higher than the electric time constant of the load 4. The timings indicated by ◯ on both current waveforms indicate the sampling timing of the current synchronized with the apex of the triangular wave carrier. The straight line shown by the dotted line shows the average current value which is the average of each current waveform.

In FIG. 7A, since the electric time constant of the load 4 is shorter than the carrier frequency, the current changes in a curved shape. Therefore, the average current value is different from the sampled current value. On the other hand, in FIG. 7B, since the electric time constant of the load 4 is longer than the carrier frequency, the current changes linearly, and the average current value and the sampled current value match. The difference between the average current value and the sampled current value as shown in FIG. 7A is particularly problematic in the power conversion unit 2 in which the coefficient α changes greatly according to the current.

Further, the current detection signal Iout detected when the coefficient α is measured by using the current control process often contains noise. Therefore, the voltage command value Vi * also contains a noise-causing component. This noise-causing component also affects the calculation result of the coefficient α and causes an error. In order to suppress the influence of this noise, it is necessary to set a large difference between the two types of carrier frequencies used for the measurement, considering the improvement of the S / N ratio, which is the ratio of the signal and the noise.

In order to avoid the problems caused by the current ripple and noise, for example, in the conventional method, the combination of two types of carrier frequencies at the time of α measurement is adjusted. However, considering the above-mentioned current ripple, the lower limit of the lower carrier frequency of the two types of carrier frequencies is limited.

On the other hand, if the higher carrier frequency of the two types of carrier frequencies is raised too much, the switching loss will increase. Therefore, the upper limit is limited by the cooling capacity of a cooling device (not shown) connected to the switching elements 20 and 21 of the power conversion unit 2.

In order to improve the S / N ratio, a method of averaging the measured values obtained in a plurality of measurements can be considered, but the increase in the measurement time delays the normal power supply operation of the power conversion unit 2. There is a problem that occurs. Further, although the carrier frequency determines the voltage update cycle, the voltage update cycle is also changed because the carrier frequency is changed in the measurement of the coefficient α. In order to prevent oscillation and instability of the current control loop accordingly, it takes time and effort to adjust the gain of the current control system.

On the other hand, the change of the dead time length command Td * of the present embodiment has almost no effect on the magnitude of the current ripple. This is because the length of the dead time itself is originally very small with respect to the pulse width of the output voltage by PWM control. Of the two types of dead time length conditions used for measuring the coefficient α, the shorter one is limited in order to prevent vertical short circuits of the switching elements 20 and 21. However, the length of the longer dead time is not particularly limited, and the difference between Td1 * and Td2 * is large in the acquisition of the voltage command values V1 * and V2 * used in the calculation of equations (3) to (6). can do. Therefore, the S / N ratio can be significantly improved with respect to the noise of the current control process.

A specific example will be described with reference to FIG. FIG. 8 is a diagram showing an example of the time of the switching off state and the switching on state of the switching elements 20 and 21. In FIG. 8, CW indicates the carrier frequency, SWP indicates the P-side switching command after the dead time is added, SWN indicates the N-side switching command after the dead time is added, and Vout indicates the output voltage of the power conversion unit 2. For the sake of explanation, the output voltage of the power conversion unit 2 is described as changing in steps. For example, when the switching elements 20 and 21 of the power conversion unit 2 are made of silicon IGBT elements and the load 4 is driven as a motor, the carrier frequency is about 1 to 20 [kHz] and the dead time length is 3 to 4 [μsec]. ] Is often set.

For comparison, the conventional method of changing the carrier frequency will be described. 8 (A) and 8 (B) have the same dead time length but different carrier frequencies. For example, in the conventional method, after obtaining the voltage command value Vi * at the carrier frequency 5 [kHz] shown in FIG. 8 (A), the length of the dead time is not changed, and FIG. 8 (B) shows. The voltage command value Vi * at the indicated carrier frequency of 10 [kHz] is acquired. The length of the dead time is 3 [μsec], the voltage command value is 0 [V], the output current of the power conversion unit 2 is positive, and the output voltage during the dead time is −Vdc / 2 [V]. At this time, the voltage pulse width at the carrier frequency of 5 [kHz] in FIG. 8 (A) is 97 [μsec]. When the carrier frequency is changed to 10 [kHz] in FIG. 8 (B), the voltage pulse width becomes 47 [μsec]. The voltage pulse width at a carrier frequency of 10 [kHz] is 48 [%] longer than the voltage pulse width at a carrier frequency of 5 [kHz].

On the other hand, in the present embodiment, the carrier frequency is the same and the length of the dead time is changed. For example, after obtaining the voltage command value Vi * with a carrier frequency of 5 kHz and a dead time of 3 [μsec] shown in FIG. 8 (A), the carrier frequency is not changed and the dead time length is as shown in FIG. 8 (C). The voltage command value Vi * with the length changed to 6 [μsec] is acquired. At this time, the voltage pulse width at the dead time of 3 [μsec] shown in FIG. 8 (A) is 97 [μsec]. The voltage pulse width in which only the dead time length in FIG. 8C is changed to 6 [μsec] is 94 [μsec]. The voltage pulse width with a dead time of 6 [μsec] is 97 [%] longer than the voltage pulse width with a dead time of 3 [μsec].

Even in the comparison between FIGS. 8B and 8D in which the carrier frequency is 10 [kHz] and only the dead time is different, the width of the voltage pulse changes from 47 [μsec] to 44 [μsec]. That is, at a carrier frequency of 10 [kHz], the ratio of the voltage pulse width of the dead time of 6 [μsec] to the voltage pulse width of the dead time of 3 [μsec] is 94 [%].

When the length of the dead time is changed as described above, the fluctuation of the voltage pulse width is smaller than that when the carrier frequency is changed. Further, in the measurement of the coefficient α, it is not necessary to change the carrier frequency. Therefore, the range of the cooling capacity of the cooler connected to the switching elements 20 and 21, that is, the range in which the switching loss can be tolerated becomes high. From this, the S / N ratio can be significantly improved by increasing the difference between the dead time length commands Td1 * and Td2 * in the present embodiment.

Further, by setting the carrier frequency to be sufficiently higher than the electric time constant of the load 4, it is possible to suppress the influence of the current ripple on the measurement accuracy of α in the vicinity of the current zero cross.

As described above, according to the present embodiment, by acquiring two types of voltage command values Vi * having different dead time lengths, the influence of current ripple and noise is suppressed and the coefficient α is acquired with high accuracy. , The amount of compensation can be calculated. Further, by using the obtained compensation amount, the effect of improving the compensation accuracy of the output voltage error due to the dead time is obtained.

Embodiment 2.
The second embodiment will be described. In the first embodiment, the case where there are two types of dead time length commands Td * in the measurement of the coefficient α has been described as an example, but in the second embodiment, a plurality of types of dead time length commands Td * are acquired. The configuration for obtaining the coefficient α is different from that of the first embodiment. The other configurations are the same as or corresponding to those of the first embodiment, and the description thereof will be omitted. In the second embodiment, the parts different from the first embodiment will be described below.

In the second embodiment, as shown in Table 1, the voltage command value Vi * is acquired with a plurality of dead time lengths. Item X in Table 1 shows each voltage command value (V3 *, V4 *, V5 * ...) obtained when the dead time length command Td * is changed. In addition, item Y in Table 1 shows a plurality of dead time length commands (Td3 *, Td4 *, Td5 * ...) And a DC link voltage Vdc (Vdc3, Vdc4) that is input to the power conversion unit 2. , Vdc5 ...). The slope S is obtained by linearly approximating the items X and Y by the least squares method. Then, the coefficient α can be calculated by the equation (7) using the slope S and the carrier frequency fc.

The steps executed by the sequence management unit 61 will be described with reference to FIG. FIG. 9 is a flowchart showing an example of the execution operation of the control unit 6 according to the present embodiment.

In the current command step S111, the current command value I * is output from the sequence management unit 61 to the voltage command generation unit 62. The voltage command value I * at this time is an arbitrarily set value. The voltage command generation unit 62 performs the above-mentioned current control process based on the current command value I * and the current detection signal Iout from the current sensor 3. Then, the voltage command value Vi * is output to the PWM processing unit 63. Further, the PWM processing unit 63 generates a switching command Sw1 * in the Td addition unit 64 based on the voltage command value Vi * from the voltage command generation unit 62.

In the first dead time length command step S112, the dead time length command Td * is output from the sequence management unit 61 to the Td addition unit 64. In the present embodiment, Td3 * is output as a certain value of the dead time length command Td *. The Td addition unit 64 adds a dead time to the switching command Sw1 * based on the dead time length command Td3 *. Then, the switching command Sw2 * after the dead time is added is output to the power conversion unit 2.

By the operation of the steps up to this point, the power conversion unit 2 operates the switching elements 20 and 21 on and off based on the switching command Sw2 *. The power conversion unit 2 converts the DC power supplied from the DC power supply 1 into AC power by the on and off operations of the switching elements 20 and 21, and outputs the DC power to the load 4.

The current supplied from the power conversion unit 2 to the load 4 at this time is detected by the current sensor 3 installed on the lead wire connecting the power conversion unit 2 and the load 4. The current sensor 3 outputs the current detection signal Iout to the voltage command generation unit 62.

In the first voltage command acquisition step S113, the voltage command value V3 * is stored in the storage device 602. The voltage command value V3 * is a voltage command value newly output by the above-mentioned current control process based on the dead time length command Td3 *. At this time, the product of the dead time length command Td3 * and the DC link voltage Vdc3 is obtained and stored in the storage device 602.

In the second dead time length command step S114, the dead time length command Td4 * is output from the sequence management unit 61 to the Td addition unit 64. The dead time length command Td4 * is a different value from the dead time length command Td3 *. The Td addition unit 64 adds a dead time to the switching command Sw1 * based on the dead time length command Td4 *. Then, the switching command Sw2 * after the dead time is added is output to the power conversion unit 2.

The switching elements 20 and 21 of the power conversion unit 2 are turned on and off based on the switching command Sw2 *. The power conversion unit 2 converts the DC power supplied from the DC power supply 1 into AC power by the on and off operations of the switching elements 20 and 21, and outputs the DC power to the load 4.

At this time, the current supplied from the power conversion unit 2 to the load 4 is detected by the current sensor 3 installed on the lead wire connecting the power conversion unit 2 and the load 4. The current sensor 3 outputs the current detection signal Iout to the voltage command generation unit 62.

In the second voltage command acquisition step S115, the newly acquired voltage command value V4 * is stored in the storage device 602. The voltage command value V4 * is a voltage command value newly output by the above-mentioned current control process based on the dead time length command Td4 *. At this time, the product of the dead time length command Td4 * and the DC link voltage Vdc4 is obtained and stored in the storage device 602.

In the present embodiment, this operation is repeated an arbitrary number of times by changing the dead time length command Td * to Td5 *, Td6 * .... Then, after repeating the process an arbitrary number of times, the coefficient α is acquired using the equation (7) in the coefficient α acquisition step S116.

By using the product of a plurality of voltage command values Vi *, the dead time length command Td *, and the DC link voltage Vdc, it is possible to suppress variations in measurement such as noise of the voltage command value Vi *. The type of dead time length command Td * corresponding to the number of repetitions can suppress noise as the number of times increases, and therefore, it may be arbitrarily determined according to the desired suppression effect.

Further, as shown in FIG. 10, the current command change step S117 may be provided. In this case, after acquiring the coefficient α in the coefficient α acquisition step S116, the current command is changed to an arbitrary current command I * in the current command change step S117. Then, the changed current command I * is output in the current command step S111, and the operations of S111 to S116 described above are repeated. As a result, the coefficient α can be acquired for each current command value I * and can be held as a table.

As described above, in the second embodiment, since a plurality of dead time length commands Td * are set and the coefficient α is measured, it is possible to suppress variations in measurement such as noise of the voltage command value Vi *. .. Therefore, it is possible to obtain the coefficient α with high accuracy by suppressing the influence of current ripple and noise, and to obtain the compensation amount. Further, by using the obtained compensation amount, the effect of improving the compensation accuracy of the output voltage error due to the dead time is obtained.

Embodiment 3.
In the third embodiment, a specific sequence is not provided in the acquisition of the coefficient α, the coefficient α is acquired while the load 4 is normally supplied with electric power, and dead time compensation is performed. A part of the configuration of the control unit 6 is different from that of the first and second embodiments. Other configurations are the same as or corresponding to those of the first and second embodiments, the description thereof will be omitted, and the parts different from the first and second embodiments will be described below.

FIG. 11 is a configuration diagram of a power conversion system to which the power conversion device according to the third embodiment for carrying out the present invention is applied. The entire system is composed of the power conversion device 100, the DC power supply 1, and the current sensor 3 of the present embodiment. For convenience of explanation, the load 4 is also shown in FIG. The power conversion device 100 has a function of an inverter that converts DC power into AC power, converts DC power supplied from DC power source 1 which is an input power source, and supplies it to load 4. The current sensor 3 outputs a detection current signal Iuvw based on the power supplied from the power converter 100 to the load 4. Although the current sensor 3 is provided separately from the power conversion device 100 in the present embodiment, it may be included in the power conversion device 100.

Further, in the third embodiment, for convenience of explanation, the load 4 connected to the power conversion unit 2 is a three-phase load. Examples of the three-phase load include a permanent magnet synchronous motor and an induction motor.

As shown in FIG. 11, the power conversion device 100 in the present embodiment includes a power conversion unit 2 and a control unit 6.

When the load 4 is a three-phase load, the power conversion unit 2 is composed of, for example, a three-phase bridge circuit. Although the details of the three-phase bridge circuit are not shown in FIG. 11, the circuit of the three-phase bridge circuit corresponding to each of the u-phase, v-phase, and w-phase is a diagram showing the circuit configurations of the first and second embodiments. Similar to 1, it is composed of IGBT 201, 211 and FWD202, 212. The P-side switching element 20 and the N-side switching element 21 are connected in series. Further, in the present embodiment, the switching elements 20 and 21 of FIG. 1 are set as one set, and three of them are connected in parallel to form a bridge using a total of six switching elements. The switching elements 20 and 21 of the respective phases in the three-phase bridge circuit are driven based on the switching command Swuvw2 * from the control unit 6.

The current sensor 3 is, for example, a shunt resistor or a current transformer installed in a lead wire connecting the power conversion unit 2 and the load 4. The current sensor 3 is installed in each of the above phases and detects the three-phase current flowing from the power conversion unit 2 to the load 4. The three-phase current flows from between the P-side switching element 20 and the N-side switching element 21 to each phase terminal of the load 4 through a bus bar or the like and is supplied.

An example in which the current sensors 3 are provided in each of the above phases will be illustrated and described with reference to FIG. 11, but the number of current sensors 3 may be two. In this case, it is possible to grasp the current value of the three phases from the current values of the two phases of the above-mentioned phases detected by a known method.

The control unit 6 outputs a switching command Swuvw 2 * to the power conversion unit 2 based on the detected current signal Iuvw from the current sensor 3. The switching command Swuvw2 * is a command for driving the switching elements 20 and 21 of the above-mentioned phases of the power conversion unit 2.

The hardware configuration of the control unit 6 is composed of the processor 601 and the storage device 602 as in the first and second embodiments, and realizes the functions of the respective units described below. The control unit 6 will be further described.

As shown in FIG. 11, the control unit 6 includes a voltage command generation unit 62, a dead time compensation unit (hereinafter referred to as Td compensation unit) 66, a PWM processing unit 63, a Td addition unit 64, and a sequence management unit 61. To.

Unlike the operations of the first and second embodiments, the sequence management unit 61 sends a time-varying dead time length command Td * (t) to the Td addition unit 64 based on the detected current signal Iuvw of the current sensor 3. Output. Further, the coefficient α is output to the Td compensation unit 66.

The voltage command generation unit 62 generates the voltage command value Vuvw1 * and outputs it to the Td compensation unit 66. Each of the voltage command values Vuvw1 * of each phase at this time is described in the equation (8). Vamp in Eq. (8) indicates the voltage command amplitude, and θ indicates the voltage command phase. The method of generating the voltage command value Vuvw1 * in the present embodiment can be applied without any particular limitation. Therefore, the three-phase voltage command value Vuvw1 * described in the equation (8) is a general expression describing the amplitude and the phase.

The Td compensation unit 66 adds the compensation voltage having the opposite polarity to the dead time voltage error to the voltage command value Vuvw1 * to perform dead time compensation. The Td compensation unit 66 receives the polarity of the detected current signal Iuvw, the coefficient α from the sequence management unit 61, and the time-varying dead time length command Td * (t), and performs dead time compensation. As a result, the voltage command value Vuvw2 * after the dead time compensation becomes the equation (9). In equation (9), fc is the carrier frequency [Hz], Vdc is the DC link voltage input to the power conversion unit 2, and sign (iu), sign (iv), and sign (iwa) are the output currents of the above phases. The polarities are shown respectively. Further, the polarity of the output current is positive when it coincides with the direction d from the power conversion unit 2 toward the load 4, 1 when it is positive, and -1 when it is negative.

The PWM processing unit 63 performs PWM processing based on the voltage command value Vuvw2 * after the dead time compensation from the Td compensation unit 66, and outputs the switching command Swuvw1 * to the Td addition unit 64. The PWM process is performed by comparing the magnitude of the triangular wave and the sawtooth wave.

The Td addition unit 64 adds a dead time to the switching command Swuvw1 * processed by the PWM processing unit 63 according to the time-changing dead time length command Td * (t) from the sequence management unit 61. Then, the switching command Swuvw2 * with the dead time added is output to the power conversion unit 2.

The sequence management unit 61 will be described in more detail. The time-changing dead time length command Td * (t) output from the sequence management unit 61 in the present embodiment is given by the equation (10). That is, the dead time length command Td * (t) vibrates at a predetermined frequency.

In equation (10), TdAmp indicates the amplitude of the vibration component of the Td length, and θtd indicates the phase of vibration. The frequency of vibration is set avoiding the frequency components of other disturbance voltages. Further, the offset value Td_ofs is provided in order to prevent the switching elements of the above phases from being vertically short-circuited while oscillating the length of the dead time. Equation (10) oscillates the length of the dead time based on the trigonometric function. In the present embodiment, the vibration having a dead time length may vibrate with a predetermined frequency component. Therefore, for example, it may be vibrated in a rectangular wave shape. Here, assuming that the frequency is ftd, the relationship with θtd is given by Eq. (11).

The reason for setting the frequency of θtd while avoiding the frequency components of other disturbance voltages will be described. For example, when a three-phase AC motor is connected as the load 4, the motor itself may output a disturbance voltage. In a permanent magnet synchronous motor, it is an induced voltage harmonic generated by the magnetic flux and rotation of the rotor. In an induction motor, this is voltage distortion due to saturation of mutual inductance.

These include harmonic voltage components of a plurality of orders on the three-phase rest coordinates, and the three-phase current of the motor, which is the load 4, also contains the same components. In particular, the 5th and 7th order components are remarkable, and when the three-phase current and the three-phase voltage are converted on the two-axis orthogonal rotation coordinates, the sixth-order frequency component with respect to the frequency of the AC output voltage of the power conversion unit 2 is obtained. It is known to include.

On the other hand, the dead time voltage error also has a shape close to a square wave on the three-phase rest coordinates and includes odd-order harmonic voltage components. Similarly, the 5th and 7th order components are remarkable. The dead time voltage error becomes a saw-shaped voltage on the two-axis orthogonal rotating coordinates, and becomes a sixth-order frequency component. It is a well-known phenomenon that the current on the two-axis orthogonal rotating coordinates contains a sixth-order frequency component due to this dead time voltage error.

Therefore, a method for adjusting the Td compensation voltage using the sixth-order frequency component of the current on the two-axis orthogonal rotating coordinates has also been proposed. However, in the conventional method, the adjustment may be hindered by the sixth-order periodic disturbance voltage caused by the motor, and the dead time compensation accuracy may decrease.

Therefore, in the present embodiment, when the length of the dead time is changed with time, the frequency is set to a value different from the frequency component of the disturbance voltage to suppress the influence of the disturbance voltage. Specifically, in the case of a three-phase AC motor load, the time is changed in a cycle other than the sixth-order cycle. As a result, the frequency of the current pulsation component caused by the error of the coefficient α and the frequency of the disturbance component caused by the motor can be made different. As a result, the coefficient α can be adjusted without being affected by the disturbance voltage caused by the motor. Further, by compensating using the adjusted coefficient α, it is possible to improve the dead time compensation accuracy.

Next, a method of adjusting the coefficient α will be described.
When the dead time length command is changed over time and the dead time compensation is calculated and implemented accordingly, it is sufficient if the accurate value of the coefficient α can be grasped. By grasping the accurate value, the dead time voltage error is suppressed. That is, the output current to the load 4 does not include the corresponding vibration component.

However, if an error in the value of the coefficient α is included, the output current to the load 4 includes the pulsation caused by the above-mentioned oscillating dead time voltage component. Therefore, basically, the pulsation of the output current due to the vibration component of the dead time voltage error is detected. Then, the pulsation may be adjusted to be zero. Therefore, various methods can be used.

For example, the coefficient α is gradually changed within a predetermined range. At that time, the presence / absence of a pulsating component having a frequency due to the vibration of the dead time voltage error included in the detected current signal Iuvw is recorded. Then, a method is used in which the coefficient α that minimizes the current pulsation is finally output to the Td compensation unit 66 as the adjustment result. By performing this operation at appropriate time intervals, highly accurate dead time compensation can be realized while the power is supplied to the load 4.

The following method can be used as another method. The dead time voltage error assumes a state in which a three-phase sinusoidal alternating current is flowing through the load 4. In this case, the dead time voltage error has a shape close to a rectangular wave according to the equation (2). Assuming that the frequency of the output voltage of the power conversion unit 2 is f, the three-phase sinusoidal alternating current also has the same frequency f.

At this time, the frequency of the voltage command phase θ described in the equation (8) is also f. When the length of the dead time is changed with time, the amplitude of the dead time voltage error which becomes a rectangular wave is changed based on the equation (2). Since it is a kind of amplitude modulation, the vibration component included in the dead time voltage error has frequency components of f + ftd and f−ftd.

Further, the phase of the fundamental wave component of the dead time voltage error of the square wave shape is in phase with the three-phase sinusoidal alternating current. Therefore, the vibration component of the dead time voltage error has a main component in the amplitude direction of the phase current. Similarly, the current pulsation caused by the vibration component also has a main component in the amplitude direction of the phase current.

Therefore, if the amplitude is extracted in the detected current signal Iuvw corresponding to the phase current signal, the current pulsation caused by the vibration component of the dead time voltage error can be extracted. However, since there are three phases, it is necessary to handle all phases together. In this case, the current pulsation is included in the magnitude of the phase current vector converted on the two-axis orthogonal rotating coordinates.

This is shown in Eqs. (12) and (13). In the equation (12), the coordinate conversion phase is the voltage command phase θ, but any phase having the same frequency as the output voltage of the power conversion unit 2 or the three-phase alternating current may be used. In in Eq. (13) indicates the amplitude of the detected current vector.

In equation (12), a shift of the frequency component of the signal converted by the rotating coordinate conversion occurs. The frequency of the coordinate transformation phase θ is f. The magnitude in of the phase current vector described in the equation (13) includes a pulsation having the original ftd frequency due to the frequency shift.

As described above, this current pulsation occurs with an error of the coefficient α. Therefore, α is adjusted by Eq. (14) based on the magnitude in of the phase current vector.

In equation (10), vibration is generated by the cos component with respect to the phase θtd of the vibration component of the dead time voltage error. At this time, ftd is set to a value sufficiently larger than the reciprocal of the electric time constant of the load 4. As a result, the vibration component of the dead time voltage error is mainly handled by the inductance component of the load 4.

Therefore, the current pulsation caused by the vibration component of the dead time voltage error is mainly composed of a component having a phase delay of 90 degrees with respect to the vibration component of the voltage. Eq. (14) reflects this and has a simple configuration for detecting with a sin signal having a phase θtd. When the coefficient α is adjusted, the magnitude in of the phase current vector described in Eq. (13) does not include the current pulsation caused by the vibration component of the dead time voltage error. Therefore, the integration of Eq. (14) is automatically stopped. K in Eq. (14) indicates the gain that determines the response speed of α adjustment.

When such a method is used, the sequence management unit 61 executes the equations (10) to (14) for each control cycle to adjust α. This method is always operational. Therefore, even if the coefficient α fluctuates while the power is being supplied to the load 4, it can be followed.

For example, if a conventional method of vibrating the carrier frequency is applied to this method, the carrier frequency is locally lowered. As a result, the voltage command value is affected by the PWM current ripple, and the estimation accuracy of the coefficient α is lowered. Further, since the current sampling timing and the update cycle of the voltage command value also fluctuate according to the fluctuating carrier frequency, there arises a problem that the configuration of the control unit 6 becomes complicated.

On the other hand, in the present embodiment, the length of the dead time is changed with a predetermined cycle to vibrate the dead time voltage error. At this time, the dead time compensation is also performed by calculating the compensation voltage according to the length Td * (t) of the dead time that changes with time. Therefore, if the accurate value of the coefficient α can be grasped, the dead time voltage error can be suppressed. As a result, the current of the load 4 does not include the corresponding vibration component.

Further, for the current pulsation caused by the oscillating dead time voltage error component included in the current of the load 4, the coefficient α is adjusted so that the current pulsation becomes zero. As a result, an accurate value of α can be obtained. Then, the acquired coefficient α can suppress the influence of the disturbance voltage other than the dead time voltage error.

As described above, in the present embodiment, the influence of the disturbance voltage can be suppressed by vibrating the dead time voltage error with a value different from the frequency component of the disturbance voltage. Further, the coefficient α can be adjusted and acquired so that the current pulsation becomes zero with respect to the current pulsation caused by the dead time voltage error component. Therefore, it is possible to obtain the coefficient α with high accuracy by suppressing the influence of current ripple and noise, and to obtain the compensation amount to compensate for the dead time voltage error. Further, in addition to the effect that the coefficient α can be adjusted without being affected by the disturbance voltage caused by the motor, the effect is that it is possible to perform highly accurate dead time compensation while supplying normal power to the load 4.

Embodiment 4.
In the third embodiment, the dead time length command is changed with time, the dead time voltage error is vibrated, and the coefficient α is adjusted from the current pulsation caused by the vibration of the dead time voltage error included in the output current. Explained. The fourth embodiment is different from the third embodiment in that the coefficient α is adjusted based on the voltage command value Vuvw *. Other configurations are the same as or corresponding to those of the third embodiment, and the description thereof will be omitted.

FIG. 12 shows a configuration diagram of a power conversion system to which the power conversion device according to the present embodiment is applied. The current detection signal Iuvw from the current sensor 3 is input to the voltage command generation unit 62. The voltage command generation unit 62 performs current control processing in response to the current detection signal Iuvw and a voltage command from, for example, a higher control output stage (not shown), and outputs the voltage command value Vuvw * to the Td compensation unit 66. The voltage command value Vuvw * is also output to the sequence management unit 61.

The sequence management unit 61 receives the voltage command Vuvw * after the current control processing, and adjusts the coefficient α based on the voltage command Vuvw *. The Td compensation unit 66 adds a compensation voltage having the opposite polarity to the voltage command Vuvw * after the current control process to perform dead time compensation, and outputs the compensated voltage command value Vuvw2 *. Other operations are the same as those in the third embodiment, and the description thereof will be omitted.

The current pulsation caused by the dead time voltage error is a disturbance, and the operation in which the signal of the same frequency and amplitude is included in the current command value is generally very small. Therefore, when the operation of the control unit 6 includes the current control process, the current control process operates so as to suppress the component of the current pulsation. Therefore, the voltage command value Vuvw *, which is the output of the current control process, contains the same frequency component as the current pulsation with opposite polarity. Therefore, it is possible to adjust the coefficient α based on the voltage command value Vuvw *.

As described above, in the fourth embodiment in which the coefficient α can be adjusted based on the voltage command value Vuvw *, the coefficient α is acquired with high accuracy and the dead time voltage error is compensated as in the first and second embodiments. The amount of compensation to be paid can be obtained. Further, as in the third embodiment, the coefficient α can be adjusted without being affected by the disturbance voltage caused by the motor, and it is possible to perform highly accurate dead time compensation while supplying normal power to the load 4. It works.

Embodiment 5
In the third and fourth embodiments, the dead time voltage error is oscillated using the time-varying dead time length, and the coefficient α is adjusted from the current pulsation caused by the vibration of the dead time voltage error included in the output current or the voltage command. The configuration to be implemented was explained. At this time, the frequency of the dead time that changes with time was set to the same value in the three phases. In the fifth embodiment, the configuration in which the frequency for changing the length of the dead time with time is different in the three phases is different from the third and fourth embodiments. Other configurations are the same as or corresponding to those of the third and fourth embodiments, the description thereof will be omitted, and the differences will be described below.

When the frequency for changing the length of the dead time is different between the three phases, the equation (9) showing the voltage command value Vuvw2 * after the dead time compensation in the third and fourth embodiments becomes the equation (15). As a result, the excess or deficiency of the coefficient α can be grasped for each phase. By utilizing this, the coefficient α can be adjusted for each of the three phases, and the dead time compensation can be made more accurate.

Tdu ^* (t) is the Td length that changes according to the time of the u phase, Tdv ^* (t) is the Td length that changes according to the time of the v phase, and Tdw ^* (t) changes according to the time of the w phase. The Td length and αu indicate the coefficient α of the u phase, αv indicates the coefficient α of the v phase, and αw indicates the coefficient α of the w phase.

The length of the dead time is calculated for each phase using the equations (16) and (17).

ftdu indicates the vibration component frequency of the Td length of the u phase, ftdv indicates the vibration component frequency of the Td length of the v phase, and ftdw indicates the vibration component frequency of the Td length of the w phase.

The appearance of the current pulsation caused by the vibration of the dead time voltage error in the magnitude of the phase current vector is the same as the case where the three phases described in the third embodiment are changed by the same dead time length. Therefore, equations (12) and (13) can be used without change. The adjustment formula for the three-phase coefficient α is changed from the formula (14) to the formula (18).

For example, it is assumed that the conventional method of vibrating the carrier frequency is applied to this method. In this case, it is necessary to use carriers having individual frequencies in the three phases to adjust α in the three phases individually by changing the carrier frequency. Therefore, there arises a problem that the PWM space voltage vector composed of the three-phase voltages is significantly collapsed and the current is disturbed.

On the other hand, when the length of the dead time itself is changed with time as in the present embodiment, such a problem does not occur. Therefore, the coefficient α can be adjusted individually for the three phases.

As described above, in addition to the same effects as those in the first to fourth embodiments, the present embodiment has the effect of enabling more accurate dead time compensation by individually adjusting the coefficient α in the three phases.

In the present invention, each embodiment can be freely combined, and each embodiment can be appropriately modified or omitted within the scope of the invention.

1 DC power supply, 2 power conversion unit, 20 P side switching element, 21 N side switching element, 3 current sensor, 4 load, 6 control unit, 61 sequence management unit, 62 voltage command generator, 63 PWM processing unit, 64 dead Time addition unit, 65 voltage command unit, 66 dead time compensation unit, 601 processor, 602 storage device, 201 P side IGBT, 211 N side IGBT, 202 P side FWD, 212 N side FWD

Claims (7)

A power converter that has a switching element and supplies current to the load,
A sequence management unit that outputs a dead time length command that determines the length of the dead time, which is a protection period for turning off the switching element, and the dead time added based on the dead time length command. A control unit having a dead time addition unit that outputs a switching command for operating the switching element on and off is provided.
The sequence management unit determines the current value supplied from the power conversion unit to the load or the switching command when the switching element is turned on and off by the switching command based on the dead time length command. A power conversion device that obtains a compensation amount for compensating for the voltage command value based on a voltage command value to be generated. The power conversion device according to claim 1, wherein the sequence management unit outputs n types (n is a natural number of 2 or more) of the dead time length commands. The power conversion device according to claim 1 or 2, wherein the voltage command value is a voltage command value based on the current value. The sequence management unit outputs n types of the dead time length commands, and uses the n types of the voltage command values obtained in response to the n types of the dead time length commands for the n types of the dead time length commands. The power conversion according to any one of claims 1 to 3, wherein the inclination of n kinds of the voltage command values is obtained, and the compensation amount for compensating the voltage command value is obtained based on the inclination. apparatus. The sequence management unit uses the dead time length command, the voltage command value obtained based on the dead time length command, and the DC link voltage to be input to the power conversion unit. The power conversion device according to any one of claims 2 to 4, wherein a compensation amount for compensating the voltage command value is obtained. The sequence management unit outputs a time-changing dead time length command, and the load from the power conversion unit is caused by the on and off operations of the switching element based on the time-changing dead time length command. The power conversion device according to claim 1, wherein a compensation amount for compensating the voltage command value is obtained by using the current value supplied to the power conversion device. The control unit includes a dead time compensation unit that compensates by adding a compensation voltage having a polarity opposite to the voltage command value to the voltage command value based on the compensation amount, according to claims 1 to 6. The power conversion device according to any one item.

Images