POWER CONVERSION DEVICE AND ROTATING ELECTRIC MACHINE DRIVING DEVICE
TECHNICAL FIELD
[0001] The present invention relates to a power conversion
device for a rotating electric machine, and a rotating
electric machine driving device using the power conversion
device.
BACKGROUND ART
[0002] In a power conversion device that drives an electric motor by feedback control using phase current, means for detecting phase current is required. For the purpose of cost reduction, a method for calculating electric motor current from DC bus current of an inverter has been conventionally used instead of providing a current sensor between the power conversion device and the electric motor. For example, Patent Document 1 discloses that current detection means detects pulse-style current corresponding to phase current flowing through a DC bus of a power conversion unit main circuit along with switching of a switching element for each phase, and the obtained current detection value is distributed to each phase on the basis of the switching state at the time of the detection, whereby phase currents for

three phases are detected and reproduced by single current
detection means.
[0003] However, in the above method, it is difficult to
detect the center of current ripple of the phase current, and
therefore there is a problem that the current detection
accuracy decreases. Considering this, for example, Patent
Documents 2 and 3 disclose that current detection in the
first half cycle and current detection in the second half
cycle of one triangular carrier wave cycle are performed
alternately, and current detection error is compensated from
the respective obtained current detection values, whereby
appropriate current detection can be achieved even in the
case of an electric motor having small impedance and large
current ripple. :
CITATION LIST
PATENT DOCUMENT |
[0004] Patent Document 1: Japanese Laid-Open Patent
Publication No. 11-004594 !
Patent Document 2: Japanese Laid-Open Patent Publication No. 2010-11639
Patent Document 3: Japanese Laid-Open Patent Publication No. 2013-55772
SUMMARY OF THE INVENTION
I

PROBLEMS TO BE SOLVED BY THE INVENTION
[0005] In order to achieve the above conventional method,
in Patent Document 2, a current detection withholding
interval for withholding current detection during an integer
multiple of a triangular carrier wave cycle is provided
between current detection in the first half cycle of one
triangular carrier wave cycle and current detection in the
second half cycle of one triangular carrier wave cycle. In
Patent Document 3, in two carrier signal cycles, detection is
performed in the second half of the first cycle and the first
half of the second cycle. Consequently, in Patent Document 2,
a current detection cycle and the accompanying voltage !
command update cycle are 1.5 times or more of the carrier j
cycle, and in Patent Document 3, the current detection cycle
and the accompanying voltage command update cycle are twice
the carrier cycle. Therefore, in such a case of performing
high-frequency voltage output in low carrier driving, time
resolution of voltage update is reduced, and output voltage
accuracy is reduced. In addition, in Patent Document 2,
current detection in the first half of one triangular carrier
wave cycle and current detection in the second half of one
triangular carrier wave cycle are performed at positions
symmetric with respect to the control cycle start point,
whereby detection accuracy is improved. However, in such a
method, current ripples periodically occur at the same

location, and noise of the same frequency occurs intensively. [0006] The present invention has been made to solve the above problem, and an object of the present invention is to provide a power conversion device that enables accurate current detection even in the case of an electric motor having small impedance and enables to minimize reduction in output voltage accuracy even in low carrier driving by performing current detection and voltage update per one triangular carrier wave cycle.
MEANS OF SOLUTION TO THE PROBLEMS
[0007] A power conversion device according to the present invention includes a PWM conversion unit which compares three-phase voltage commands with a triangular carrier wave, and performs conversion to PWM pulses; a power conversion unit main circuit which drives switching elements on the basis of the PWM pulses, and converts DC voltage to three-phase AC voltage; a DC bus current detection unit which detects current flowing through a DC bus of the power conversion unit main circuit; a timing determination unit which sets, on the basis of switching timings of the switching elements, two detection timings for detecting currents for two phases by the DC bus current detection unit, at least once per one cycle of the triangular carrier wave; a phase current calculation unit which calculates three-phase

AC current values once per one cycle of the triangular carrier wave on the basis of current values of the DC bus for the two phases detected at the timings determined by the timing determination unit; and a voltage command generation unit which updates the three-phase voltage commands on the basis of the three-phase AC current values once per one cycle of the triangular carrier wave. When the triangular carrier wave is divided into two intervals of a monotonic increase interval and a monotonic decrease interval, in a first cycle of two consecutive triangular carrier wave cycles, the DC bus current detection unit detects current values of the DC bus in one of the monotonic increase interval and the monotonic decrease interval. And in a second cycle of the two consecutive triangular carrier wave cycles, the DC bus current detection unit detects current values of the DC bus in the other one, of the monotonic increase interval and the monotonic decrease interval, in which detection has not been performed in the first cycle. The phase current calculation unit calculates the three-phase AC current values, on the basis of the current values of the DC bus for two phases detected respectively by the DC bus current detection unit in the monotonic increase interval and the monotonic decrease interval in the last two consecutive triangular carrier wave cycles.

EFFECT OF THE INVENTION
[0008] In the power conversion device configured as
described above, voltage update is performed per one cycle of
a triangular carrier wave cycle, and current detection is ;
alternately switched between the raonotonic increase interval
and the monotonic decrease interval. Whereby the voltage
update cycle can be shortened as compared to the conventional
case. Further, also in a case of low carrier drive, it is
possible to perform accurate current detection from the DC
bus and minimize reduction in output voltage accuracy. In |
addition, the current detection positions in the monotonic '
!
increase interval and the monotonic decrease interval are set j
i
to be asymmetric with respect to the control cycle start j
point positioned therebetween, whereby the positions of
current ripples are dispersed in the control cycle and the
frequency that causes noise can also be dispersed.
i
BRIEF DESCRIPTION OF THE DRAWINGS j
[0009] [FIG. 1] FIG. 1 is a block configuration diagram showing a hardware configuration of a power conversion device according to embodiment 1.
[FIG. 2] FIG. 2 is a block configuration diagram ;
showing a specific configuration of the power conversion '
device according to embodiment 1.
[FIG. 3] FIG. 3 is a flowchart showing operation

by a timing determination unit.
[FIG. 4] FIG. 4 is a diagram illustrating operation for determining timings of current detections performed by the timing determination unit.
[FIG. 5] FIG. 5 is a circuit diagram showing the state of current flow at the time of current detection.
[FIG. 6] FIG. 6 is a circuit diagram showing the state of current flow at the time of current detection.
[FIG. 7] FIG. 7 is an enlarged diagram showing a part of the DC bus current waveform diagram shown in FIG. 4.
[FIG. 8] FIG. 8 is diagrams (A) and (B) illustrating an example for ensuring a current detection possible interval.
[FIG. 9] FIG. 9 is a timing chart of a current detection process, a current calculation process, and a voltage generation process with respect to triangular carrier wave cycles.
[FIG. 10] FIG. 10 is a timing chart of a current detection process and a current calculation process with respect to triangular carrier wave cycles.
[FIG. 11] FIG. 11 is a diagram illustrating operation of a power conversion device according to embodiment 2.
EMBODIMENTS FOR CARRYING OUT THE INVENTION

[0010] Embodiment 1
Hereinafter, a power conversion device according to
embodiment 1 will be described in detail with reference to
drawings. FIG. 1 is a block configuration diagram showing a
hardware configuration of the power conversion device
according to embodiment 1. In the drawing, the power
conversion device 10 includes, as hardware, a processor 100,
a storage device 101, a power conversion unit main circuit 1, |
and a DC bus current detection unit 2 for detecting current flowing through a DC bus 102. Although not shown, the storage device 101 includes a volatile storage device such as a random access memory, and a nonvolatile auxiliary storage device such as a flash memory. Although not shown, the storage device 101 may include an auxiliary storage device such as a hard disk, instead of the nonvolatile auxiliary storage device.
[0011] The processor 100 executes a program inputted from the storage device 101. Since the storage device 101 includes the auxiliary storage device and the volatile storage device, the program is inputted to the processor 100 from the auxiliary storage device via the volatile storage device. The processor 100 may output data such as a calculation result to the volatile storage device of the storage device 101, or may cause the auxiliary storage device to store the data via the volatile storage device.

[0012] FIG. 2 is a block configuration diagram showing a specific configuration of the power conversion device 10. As shown in FIG. 2, the power conversion device 10 of the present embodiment includes a control unit 6, and the control unit 6 includes: a PWM conversion unit which compares three-phase voltage commands with a triangular carrier wave TW to perform conversion to PWM pulses; and a unit for determining gate switching timings and current detection timings
(hereinafter, abbreviated as a timing determination unit) which compares three-phase voltage commands with the triangular carrier wave TW to perform conversion to gate switching timings Sup to Swn, and which also serves as
current detection point determination means for outputting DC ;
bus current detection timings TA to TD on the basis of the |
gate switching timings Sup to Swn. The PWM conversion unit
includes a current detection interval generation unit which, i
i
when it is determined that it is difficult to perform DC bus current detection, adds a correction amount to a voltage
command on a side on which current detection is performed, of
j
a monotonic increase interval and a monotonic decrease j
interval, so as to enable the detection, and subtracts the !
correction amount from the voltage command on the other side. The power conversion device 10 includes the power conversion
unit main circuit 1, the DC bus current detection unit 2, a
detection current storage unit 3, a phase current calculation

unit 4, and a voltage command generation unit 5. Here, the
power conversion unit main circuit 1 drives switching
elements SW1 to SW6 on the basis of the gate switching
timings Sup to Swn, and converts DC voltage to three-phase AC
voltages. The DC bus current detection unit 2 detects
current flowing through the DC bus 102. The detection
current storage unit 3 acquires and stores current values
detected by the DC bus current detection unit 2 by using the
DC bus current detection timings TA to TD. The phase current
calculation unit 4 converts DC bus current stored in the
detection current storage unit 3, to three-phase AC current
values. The voltage command generation unit 5 updates three- j
phase voltage commands Vu to Vw once per one cycle of the j
triangular carrier wave TW.
[0013] A rotary machine M is connected to the power
conversion unit main circuit 1, and an object of the present
embodiment is to detect three-phase AC currents flowing
between the rotary machine M and the power conversion unit
main circuit 1, at the DC bus 102. Here, the connected
rotary machine M may be any type of rotary machine such as a
synchronous electric motor, an induction electric motor, or
an electric generator. In the rotary machine M shown in FIG.
1 and FIG. 2, stator windings are connected by Y connection.
However, delta connection may be used. In FIG. 2, the phase
current calculation unit 4, the voltage command generation

unit 5, and the timing determination unit in the control unit 6 are realized by the processor 100 for executing a program stored in the storage device 101, or a processing circuit such as a system LSI (not shown). The detection current storage unit 3 is realized by the volatile storage device of the storage device 101. The above functions may be executed by cooperation of a plurality of processors 100 and a
plurality of storage devices 101, or may be executed by cooperation of a plurality of processing circuits. Alternatively, the above functions may be executed by cooperation by a combination of a plurality of processors 100, a plurality of storage devices 101, and a plurality of processing circuits.
[0014] FIG. 3 is a flowchart showing operation by the timing determination unit. As shown in FIG. 3, a current detection interval is calculated from a difference among the magnitudes of the voltage commands Vu to Vw of three phases generated by the voltage command generation unit 5 (step S301), and whether current detection at the DC bus 102 is possible in the current detection interval or not is determined (step S302). If DC bus current detection is impossible, as described later, a correction amount is added to the voltage commands Vu to Vw so that the sum value in the voltage update cycle does not change, so as to enable the detection (step S303). On the basis of the voltage commands

Vu to Vw to which the correction amount has been added
according to necessity, the gate switching timings Sup to Swn
for the switching elements SW1 to SW6 of the power conversion
unit main circuit 1 are determined (step S304), the DC bus
current detection timings TA to TD are determined by using
the gate switching timings Sup to Swn (step S305), and then
current is detected by the DC bus current detection unit 2.
[0015] As shown in FIG. 1 and FIG. 2, the power conversion
unit main circuit 1 is composed of six sets of the switching
elements SW1 to SW6 and diodes, and has a role of converting DC power to three-phase AC power. That is, the power i
conversion unit main circuit 1 forms an inverter circuit, and
is composed of the U-phase upper-arm switching element SW1,
I
the U-phase lower-arm switching element SW2, the V-phase i
upper-arm switching element SW3, the V-phase lower-arm
switching element SW4, the W-phase upper-arm switching
element SW5, and the W-phase lower-arm switching element SW6.
Signals based on the gate switching timings Sup to Swn
inputted from the timing determination unit are inputted to
the switching elements SWl to SW6, whereby the switching
elements SWl to SW6 are driven. As the DC bus current
detection unit 2, a detection element (for example, a Hall
sensor, a resistor, or a current transformer) for detecting
current is provided on the path of the DC bus 102, and
voltage between both ends thereof or the output voltage

thereof is detected, via an amplifier, a buffer, and the like
according to necessity, and the detected current value Idc is
sent to the detection current storage unit 3. In FIGS. 1, 2,
the detection element is provided on the low voltage side of the DC power supply. However, the detection element may be
provided on the high voltage side. i
[0016] The detection current storage unit 3 has a role of
acquiring and storing detection currents detected by the DC
bus current detection unit 2, at the DC bus current detection j
timings TA to TD determined by the timing determination unit. |
i
FIG. 4 is a diagram illustrating operation for determining j
i
i
the timings TA to TD of current detections performed by the j
i
timing determination unit. In order to detect currents for
two phases from the DC bus 102, the detection needs to be performed in an interval in which the switching elements for three phases are in switching states corresponding to a so-called non-zero voltage vector in which they are not all ON or not all OFF. That is, it is necessary to select two of such current detection intervals and to perform the detection in the selected intervals. The DC bus current detection timings Ta to TD are determined depending on gate switching timings for the middle-voltage phase (V phase in FIG. 4) determined in the timing determination unit. The voltage commands Vu, Vv, Vw for three phases generated by the voltage command generation unit 5 are defined as, from the largest

one, a maximum phase voltage command, a middle phase voltage command, and a minimum phase voltage command. In FIG. 4, the maximum phase voltage command is Vu, the middle phase voltage command is Vv, and the minimum phase voltage command is Vw. In FIG. 4, Bu indicates a maximum phase (U phase) upper arm switch state, Bv indicates a middle phase (V phase) upper arm switch state, and Bw indicates a minimum phase (W phase) upper arm switch state.
[0017] In FIG. 4, in the detection currents for two phases, the DC bus current value detected in the non-zero voltage vector period that appears first is denoted by IdcA, and the DC bus current value detected in the non-zero voltage vector period that appears second is denoted by IdcB. The detection timing TA for the current value IdcA is determined to be at a predetermined time period Tl before a gate switching timing Svp (TS) for turning on the gate of the middle phase (V phase). Here, the time period Tl is set to be longer than a time period required for current detection. Subsequently, The detection timing TB for the current value IdcB is determined to be at a predetermined time period T2 after the timing TA. Here, for the determination of the time period T2, it is necessary to consider a dead time interval provided for preventing each pair of switching elements connected in series in the power conversion unit main circuit 1 from becoming conductive at the same time. Further since the DC

bus current may oscillate by sharp voltage change due to
ON/OFF operations of the switching elements, it is also
necessary to consider the current oscillation. That is, the
upper-arm switching element and the lower-arm switching
element (for example, SW1 and SW2) have such a relationship
that, when one of these elements is ON, the other is OFF, and
a dead time interval is provided so that, after the one is
turned off, the other is turned on, in order to prevent both
elements from becoming conductive at the same time. The time
period T2 is determined by considering the dead time interval.
Further, high-frequency oscillation of DC current occurs due
to surge voltage occurring when the switching elements perform switching (see current detection impossible interval
T10) . Therefore, it is also necessary to consider an
interval required until the high-frequency oscillation of DC ]
i
current attenuates and the amplitude of the oscillation !
S
becomes equal to or smaller than a predetermined value. FIG. !
4 shows an example of the above and shows a current detection
impossible interval T10 and a current detection possible
interval Til in the non-zero voltage vector interval of the j
gate switching timings Sup to Swn. That is, the timing TA is determined within the current detection possible interval Til. !
Hereinafter, the interval in which current detection is
impossible is referred to as a current detection impossible
interval.

[0018] Therefore, the time period T2 is set so that the timing TB is not within the current detection impossible interval. That is, T2 > Tl + (current detection impossible interval) is satisfied. Here, the current value IdcA indicates a current value of the maximum-voltage phase (U phase), and the current value IdcB indicates a current value of the minimum-voltage phase (W phase).
FIG. 5 is a circuit diagram showing flow of
currents Iu, Iv, Iw at the time of detection of the current
value IdcA, and FIG. 6 is a circuit diagram showing flow of !
currents Iu, Iv, Iw at the time of detection of the current
value IdcB. In FIG. 5, the switching elements SW1, SW4, SW6
are ON and the switching elements SW2, SW3, SW5 are OFF, and
therefore the detected current is Iu, i.e., current of the
maximum-voltage phase (U phase). In FIG. 6, the switching !
elements SW1, SW3, SW6 are ON and the switching elements SW2, j
SW4, SW5 are OFF, and therefore the detected current is Iw,
i.e., current of the minimum-voltage phase (W phase). In FIG.
5 and FIG. 6, dotted-line arrows indicate the directions of j
currents flowing in the respective phases. Between the gate j
switching timings Swp and Swn, the switching elements SW1,
SW3, SW5 in the upper arms are all ON and the switching j
elements SW2, SW4, SW6 in the lower arms are all OFF, and ;
therefore currents flow as freewheel current via the diodes in the upper arms, so that current is not detected at the DC

bus current detection unit 2. Similarly, also when the switching elements in the upper arms are all OFF, currents flow as freewheel current via the diodes, so that current is not detected at the DC bus current detection unit 2. The phase current calculation unit 4 calculates the current value of the middle-voltage phase (V phase) from the two current values IdcA, IdcB. The difference between the detection
timing for the current value IdcA and the detection timing i
!
for the current value IdcB is set to be as short as possible. The reason is as follows. By the non-zero voltage vector,
current flows through the rotary machine M, and the DC bus ;
current also changes if time elapses too much. Therefore, if
the timing Tfl and the timing TB are away from each other, the
current detection accuracy for the middle phase is
deteriorated. That is, as described above, in order to i
detect currents for two phases from the DC bus 102, the
j I
detection needs to be performed in an interval in which the j
switching elements for three phases are in switching states ;
corresponding to a so-called non-zero voltage vector in which !
they are not all ON or not all OFF, and it is necessary to
select two of such current detection intervals and to perform '
the detection in the selected intervals. ;
[0019] Timings Tc, TD in FIG. 4 are the detection timings I
for the current values IdcC, IdcD when current detection is '
performed in the second half of the triangular carrier wave

cycle. The timings Tc, TD are determined on the basis of the gate switching timing Svn (TS) for the middle-voltage phase
(V phase) . As in the timings TA, TB, the timing Tc is determined to be set at a predetermined time period Tl before the gate switching timing Svn for turning off the gate of the middle-voltage phase (V phase), and the timing TD is determined to be set at a predetermined time period T2 after the timing Tc. Here, as in the above case, the current value IdcC indicates the current value of the minimum-voltage phase
(W phase), and the current value IdcD indicates the current value of the maximum-voltage phase (U phase). Thus, it is possible to determine the current detection timings for two phases on the current detection possible interval sides in which detection is performed per triangular carrier wave cycle, it is possible to detect the DC bus current at these timings, and it is possible to store the detected current values IdcA to IdcD in the detection current storage unit 3.
[0020] Here, in order to perform current detection with high accuracy, it is necessary to reduce influence of current ripple. That is, the DC bus current value is not horizontal, as shown in FIG. 4, the DC bus current value increases rightward in the interval between the gate switching timings Sup and Swp and the DC bus current value decreases rightward in the interval between the gate switching timings Swn and Sun, and therefore the influence of these needs to be reduced.

FIG. 7 is an enlarged diagram showing a part of the DC bus
current waveform diagram shown in FIG. 4. In FIG. 7, an
interval T70 and an interval T71 respectively indicate non¬
zero voltage vector intervals, and a current value X ;
indicates an average current value in the non-zero voltage
vector intervals. As shown in FIG. 7, the current values
IdcB, IdcC differ from the average current value X in the -;
non-zero voltage vector intervals, and the differences arise
with opposite signs in plus and minus. The same applies to
the current values IdcA, IdcD. Therefore, in order to reduce
error due to current ripple, it is necessary to detect the
current values IdcA, IdcB and the current values IdcC, IdcD
alternately. That is, the following method is appropriate:
detection in the monotonic decrease interval and detection in i
i
the monotonic increase interval of the triangular carrier ;
wave TW are performed alternately, and an average value of the detected values is calculated. Therefore, the detection ;
current storage unit 3 needs to store a total of four current
values IdcA, IdcB, IdcC, IdcD, i.e., at two points (two j
intervals of monotonic decrease interval and monotonic
increase interval) for each of two phases. At the time of
calculation, the phase current calculation unit 4 obtains the
stored current value data to start calculation, and the phase
current calculation unit 4 calculates the average value of
currents for each phase of the currents for two phases.
90

[0021] The phase current calculation unit 4 calculates three-phase AC currents from the detected currents for two phases stored in the detection current storage unit 3. The signs of the current values for two phases are determined on the basis of the voltage commands Vu, Vv, Vw for three phases generated by the voltage command generation unit 5. The current value for the other one phase can be easily calculated from the already obtained current values for two phases, by using the fact that the sum of the current values for three phases is zero. By the above calculation, it is possible to detect three-phase AC current values flowing in the rotary machine M. On the basis of the phase current values, the voltage commands Vu, Vv, Vw are determined, and further, the phase currents are used in order to monitor the output of the rotary machine M. The phase currents are also used for control processes in the respective parts. [0022] The voltage command generation unit 5 has a role for generating the voltage commands Vu, Vv, Vw which are three-phase voltage commands outputted to the power conversion unit main circuit 1. According to the control method for the rotary machine M, various voltage command generation methods are known. However, these methods are not the essence of the present embodiment, and therefore the description thereof is omitted. The voltage command generation unit 5 generates the voltage commands Vu, Vv, Vw

which are three-phase voltage commands to be outputted to the
timing determination unit (control unit). Next, each
operation of the timing determination unit shown in the
flowchart in FIG. 3 will be described. First, the current j
detection interval shown in FIG. 4 is calculated from the
difference among the magnitudes of the voltage commands Vu,
Vv, Vw generated by the voltage command generation unit 5.
The current detection interval is determined depending on the
difference among the magnitudes of the voltage commands Vu,
Vv, Vw, and the smaller the difference is, the shorter the
current detection interval is.
[0023] Subsequently, whether or not it is possible to
determine the DC bus current detection timings TA to TD in
the current detection interval is determined. Whether or not j
it is possible to determine the DC bus current detection !
timings TA to TD is determined by comparing a current
detection interval T12 with the current detection impossible
interval T10. If the current detection interval T12 is |
smaller than the current detection impossible interval T10,
it is determined that the detection is impossible. The
details thereof will be described below with reference to FIG.
8. If it is determined that the detection is impossible, it
is necessary to add correction amount to the voltage commands
Vu, Vv, Vw so as to ensure a current detection interval.
[0024] FIG. 8(A) and FIG. 8(B) are diagrams illustrating

an example for ensuring a current detection possible interval
when it is determined that the detection is impossible in the
current detection interval calculated from the voltage
commands Vu, Vv, Vw. Here, an example in which DC bus i
current detection is performed in the monotonic decrease
interval of the triangular carrier wave TW, is shown.
However, the same process can be performed also in the case
where DC bus current detection is performed in the monotonic ;
increase interval. The voltage commands Vu, Vv, Vw inputted
from the voltage command generation unit 5 are defined as, ;
from the largest one, the maximum phase voltage command Vu,
the middle phase voltage command Vv, and the minimum phase
voltage command Vw. These voltage commands Vu, Vv, Vw are
compared with the triangular carrier wave TW in order to ;
determine the gate switching timings Sup to Swn. If the
current detection interval is equal to or smaller than the i
current detection impossible interval shown in FIG. 4, it is ']
difficult to perform the detection. In an interval T80 in
FIG. 8(A), the current detection interval is smaller than the
current detection impossible interval. Therefore it is
determined that the DC bus current detection timing TB cannot
be determined. l
[0025] Then, as shown in FIG. 8(B), in the case where it '
is determined that it is impossible to perform the detection in the current detection interval, the triangular carrier

wave cycle is divided into two parts, i.e., a detection interval ensuring side and a voltage compensation side. Further a voltage amount AV is added to the voltage command on the detection interval ensuring side so as to enable the detection interval to be ensured, and the same voltage amount AV is subtracted from the voltage command on the voltage compensation side. An interval T81 in FIG. 8(B) becomes larger than the interval T80, and the current detection interval becomes larger than the current detection impossible interval, therefore the DC bus current detection timing TB can be ensured. Here, the voltage amount AV for ensuring the detection interval is determined depending on the current detection impossible interval. Then, the gate switching timings Sup to Swn are corrected in accordance with the corrected voltage command. Thus, it is possible to detect the DC bus current without changing the length of the ON period of the switching element for each phase, i.e., without changing the output voltage of the inverter. It is noted that FIG. 8 shows an example in which the correction amount is added to the middle phase voltage command Vv, but a correction amount is added likewise also to the maximum phase voltage command Vu or the minimum phase voltage command Vw according to necessity. In FIG. 8(A), the case where the timing TB cannot be ensured because the difference between Vv and Vw is small has been described. However, it can also be

assumed that the timing TA cannot be determined because the difference between Vu and Vv is small. In this case, the voltage amount AV is subtracted from the voltage command on the detection interval ensuring side, and the same voltage amount AV is added to the voltage command on the voltage compensation side.
[0026] Owing to the above operation, it is possible to
perform current detection even if the difference between the
magnitudes of the voltage commands for the respective phases
is small. Further, it is possible to perform current
detection even if the magnitude relationship between the
voltage commands for the respective phases is switched, e.g.,
in the case where the rotary machine M is driven with the ;
voltage command Vv made larger than the voltage command Vu in j
accordance with the driving condition. In addition, it is i
possible to perform current detection under all driving conditions including a low-speed driving condition in which the currents Iu, Iv, Iw are small. When the detection interval is ensured, current detection cannot be performed on the voltage compensation side. Therefore, in one triangular carrier wave cycle, detection is performed for only any one of the pair of the current value IdcA and the current value IdcB, and the pair of the current value IdcC and the current value IdcD. As described above, the update cycle of the voltage command Vv to which the correction amount AV is added

in order to ensure the current detection interval needs to be equal to or longer than the triangular carrier wave cycle. In addition, determination as to whether or not current detection is possible basically needs to be performed every cycle. Therefore, the cycle with which the voltage commands Vu, Vv, Vw are inputted to the timing determination unit also needs to be equal to or longer than the triangular carrier wave cycle.
[0027] Finally, the voltage command Vv to which the
correction amount AV has been added according to necessity is
compared with the triangular carrier wave TW, the gate ;
switching timings Sup to Swn for three phases are determined, ;
the DC bus current detection timings TA to TD are determined by using the gate switching timings Sup to Swn, and the ;
detection currents detected by the DC bus current detection ;
unit 2 are acquired and stored. In addition, the gate switching timings Sup to Swn are outputted to the power conversion unit main circuit 1, and the DC bus current detection timings TA to TD are outputted to the detection current storage unit 3, whereby operation of the timing determination unit is finished.
[0028] As described above, in order to perform accurate j
current detection from the DC bus current without being influenced by current ripple, it is necessary to alternately perform the current detection in the monotonic decrease

interval and the monotonic increase interval of the
triangular carrier wave TW and to cause the update cycle of
the voltage commands Vu, Vv, Vw to be equal to or longer than
the triangular carrier wave cycle. Under this condition, in ;
order to minimize reduction in output voltage accuracy, it is necessary to cause the update cycle of the voltage commands Vu, Vv, Vw to be equal to the triangular carrier wave cycle. A process for achieving this will be described below. FIG. 9 shows a timing chart of a current detection process (E) , a current calculation process (F), and a voltage command
generation process (G) relative to triangular carrier wave i
cycles. Here, the current detection process (E) is a process ■
for storing the current value detected by the DC bus current I
detection unit 2 into the detection current storage unit 3, ;
the current calculation process (F) is a process for i
calculating three-phase AC currents by the phase current j
calculation unit 4, and the voltage command generation j
process (G) is a process for generating the voltage commands
Vu, Vv, Vw by the voltage command generation unit 5.
[0029] As described above, by alternately performing
current detection in the monotonic decrease interval and the
monotonic increase interval of the triangular carrier wave TW, j
it is possible to perform accurate current detection from DC
bus current. For this purpose, first, the triangular carrier
wave cycle is divided into two intervals of the monotonic ,

decrease interval and the monotonic increase interval. One
of these intervals is set as a current detection interval.
In the set current detection interval, a current detection
timing is determined, current is detected, and the detection !
value is stored. In the next cycle, an interval opposite to
the interval that has been set as the current detection
interval in the previous cycle is set as a current detection
interval, and then, in the same manner, current is detected
and the detection value is stored. That is, in FIG. 9, in
the first cycle, current detection is performed in the
monotonic decrease interval, and in the second cycle, current i
detection is performed in the monotonic increase interval. j
Here, as described above, errors due to current ripple arise !
with opposite signs in plus and minus in the current
detection value in the monotonic decrease interval and the j
current detection value in the monotonic increase interval of j
the triangular carrier wave TW. Therefore, by using the current values detected in the previous two cycles, detected |
current values in the monotonic decrease interval and the i
monotonic increase interval can be obtained, and by also I
performing an averaging process by the next phase current
i
calculation unit 4, the influence of detection error due to j
current ripple can be reduced.
[0030] Subsequently, the phase current calculation unit 4
executes the three-phase AC current calculation process (F)

once per one cycle of the triangular carrier wave TW. In FIG. 9, an example in which the process (F) is performed in the monotonic decrease interval is shown. However, the process
(F) may be performed in any part such as the monotonic
increase interval within one cycle. However, in the case
where the current calculated by the phase current calculation
unit 4 is used in the next voltage command generation unit 5, ;
the three-phase AC current calculation process (F) needs to |
be performed before the voltage command generation process |
(G) . The three-phase AC current calculation process (F) is
performed by using the currents stored in the detection
j
current storage unit 3. As shown in FIG. 9, by using the ]
average value of the current values obtained one cycle before j
and two cycles before, the influence of error due to current
ripple can be reduced. As a specific example of current I
calculation method, the average value of the current values j
IdcA and IdcD stored in the detection current storage unit 3 i
is used as the current value of the voltage command maximum
i
phase (U phase), the average value of the current values IdcB |
i
and IdcC is used as the current value of the voltage command j
minimum phase (W phase), and the current value of the voltage j
command middle phase (V phase) which is the other one phase j
is calculated on the basis of the relationship in which the [
sum of the current values of three phases is zero. [0031] Finally, in order to minimize reduction in output

voltage accuracy, the voltage command generation process (G)
is also executed once per one cycle of the triangular carrier
wave. In FIG. 9, an example in which the process (G) is
performed in the monotonic decrease interval is shown. As
shown in FIG. 9, current detection is performed in the
monotonic increase interval and the monotonic decrease
interval of the triangular carrier wave, and at the start of
each triangular carrier wave cycle, averaging calculation is
performed by using the last two current detection values.
Whereby a current value is calculated. In the voltage
command generation process (G), the above current value may
be used, or may not be used. It is noted that FIG. 9 shows
the case where the calculated current is used in the voltage -
command generation process (G). In this case, the voltage
command generation process (G) needs to be executed after the
three-phase AC current calculation process (F). It is noted
that, within a permissible range of reduction in output !
i
voltage accuracy, the cycle of the voltage command generation
process (G) may be set to be equal to or larger than the
triangular carrier wave cycle.
[0032] Here, there are differences in time lags between
the timings of the current calculation process (F) and the |
two current detection timings. For example, in FIG. 9, time
lags from the current detection timing in the monotonic
increase interval are 0.5 cycle and 1.5 cycles of the

triangular carrier wave, depending on the timing of the current calculation process (F). And time lags from the current detection timing in the monotonic decrease interval are 1 cycle and 2 cycles of the triangular carrier wave cycle, depending on the timing of the current calculation. In order to eliminate the influence of differences in time lags from the detection timings, the phase current calculation unit 4 performs correction so as to perform coordinate conversion of currents detected at each timing by using a rotational angle at that timing and so as to average the result thereof when
the current is calculated. Thus, detection error in the ;
i current detection values can be reduced, and therefore it is
possible to perform the current calculation and the voltage
command generation process per triangular carrier wave cycle.
[0033] As described above, currents detected at each timing are subjected to coordinate conversion into a rotating
coordinate system, so as to be used in control. At this time, j
i
if the coordinate conversion is performed after the currents
are averaged, the rotational angle used for the coordinate
conversion is influenced by the time lag, and error in
current values in the rotating coordinate system may occur.
Therefore, in order to eliminate the influence of differences |
in the time lags from the detection timings, in the current
calculation, correction as described below is performed, that
is, the currents detected at each timing are subjected to

coordinate conversion by using the rotational angle at that timing, and the result is averaged. FIG. 10 shows an example thereof. FIG. 10 shows aNprocedure for calculating output three-phase current values and the current values in the rotating coordinate system in the case where the time lag from the current detection timing in the monotonic increase interval is 0.5 cycle of the triangular carrier wave and the time lag from the current detection timing in the monotonic decrease interval is 2.0 cycles of the triangular carrier wave described in FIG. 9. Here, first, the current values of three phases at each timing are calculated from the current values of two phases detected at each timing. [0034] Next, the three-phase currents for the respective timings are averaged in each phase, and three-phase current values to be outputted is obtained. Subsequently, in feeding back detected currents to the voltage command generation unit 5, in many cases, current values in the rotating coordinate system are used instead of three-phase current values. Here, if the above three-phase currents to be outputted are subjected to coordinate conversion to calculate the current values in the rotating coordinate system, difference between the rotational angle used in the coordinate conversion and the rotational angle at the time of the actual current detection occurs. Therefore, initial rotational angles 91 and 92 in the monotonic decrease interval and the monotonic

increase interval in which current detection has been
performed is detected or estimated, and the three-phase
current values according to the detection at each timing are
subjected to coordinate conversion by using the rotational
angle in the corresponding interval. The average values of
current values Idl, Iql, Id2, Iq2 in the rotating coordinate
system for the respective timings calculated as described
above are used as current values in the rotating coordinate j
system. FIG. 10 shows the case where the time lags are 0.5
cycle and 2.0 cycles of the triangular carrier wave, but the
same process is performed also in the case of 1.0 cycle and
1.5 cycles. Thus, detection error in the current values in
the rotating coordinate system can be reduced, and therefore ;
it is possible to generate voltage commands with high I
accuracy even in the case where current is calculated per triangular carrier wave cycle and the voltage command generation process by using current values in the rotating coordinate system is performed.
[0035] In FIG. 10, in the monotonic decrease interval, DC bus current values IdcA, IdcB are detected (block 150). Then, three-phase current values Iul to Iwl are obtained through calculation (block 151). Further, coordinate conversion is performed by using the rotational angle 91, and rotating coordinate system current values Idl, Iql are obtained (block 152) . On the other hand, in the monotonic increase interval,

DC bus current values IdcC, IdcD are detected (block 200). Then, three-phase current values Iu2 to Iw2 are obtained through calculation (block 201). Further, coordinate
conversion is performed by using the rotational angle 02, and j
rotating coordinate system current values Id2, Iq2 are obtained (block 202). Finally, Idl, Iql, Id2, Iq2 are respectively averaged, and rotating coordinate system current values Id, Iq are obtained (block 300).
[0036] Through the procedure as described above, it is !
possible to update the voltage commands Vu, Vv, Vw per
triangular carrier wave cycle which is the minimum necessary j
cycle for determination as to whether or not current i
detection is possible, and it is possible to perform current j
detection with reduced influence of current ripple. Thus,
even in the case of driving at a small triangular carrier
wave frequency in which the triangular carrier wave cycle is j
long, it is possible to perform accurate current detection
and to minimize reduction in output voltage accuracy. That j
!
is, even if the wavelength of the triangular carrier wave TW j
is large and the cycle is long, the accuracy is enhanced :
because detection is performed per one cycle. In the
operation as described above, the triangular carrier wave j
cycle is the same as the current detection cycle and the |
voltage update cycle, current detection is alternately
performed in the monotonic decrease interval and the

monotonic increase interval of the triangular carrier wave TW, and the time difference between the detection timings for two phases for detection can be fixed to be constant and as small as possible. Therefore, also in the case of so-called low carrier drive in which the wavelength of the triangular carrier wave TW is large and the cycle is long, it is possible to drive the rotary machine while performing accurate current detection for all of the three phases and minimizing reduction in output voltage accuracy. [0037] Embodiment 2
FIG. 11 is a diagram illustrating operation of a power conversion device according to embodiment 2. FIG. 11 shows a positional relationship of current detection timings TA to TD determined by a timing determination unit in embodiment 2. In FIG. 11, a triangular carrier wave TW is shown at the upper side, and the state of an output voltage vector is shown at the middle. Here, the output voltage vector includes non-zero voltage vector intervals HI to H16 in which currents for two phases can be detected from the DC bus 102 as described above, and zero voltage vector intervals Jl to J7 in which currents cannot be detected. Further, in FIG. 11, an outline Z of a current waveform for one phase is shown at the lower side.
[0038] A time period T20 from a detection timing TA determined by the timing determination unit in the control

unit 6 to the next triangular carrier wave cycle start point
SI, and a time period T21 from the triangular carrier wave
cycle start point SI to a timing TD, are set to be different
from each other (T20 ¥= T21) . Along with this, a time period
T22 from a detection timing TB determined to be at a
predetermined time period T2 after the detection timing Tfi to
the next triangular carrier wave cycle start point SI, and a
time period T23 from the triangular carrier wave cycle start
point SI to a detection timing Tc determined to be at a
predetermined time period T2 before the detection timing TD,
are also different from each other (T22 # T23). In addition,
similarly, a time period from the detection timings Tc, TD to ■
the next triangular carrier wave cycle start point S2, and a
time period from the triangular carrier wave cycle start
i
point S2 to the detection timings TB, TA, are set to be j
different from each other (T24 + T25).
[0039] Through the procedure as described above, in
addition to the effects obtained in embodiment 1, the detection position differs every triangular carrier wave
cycle, and further, the position of the current detection
interval also differs. For example, when the triangular i
carrier wave cycle start point SI is defined as a reference, |
the position of the non-zero voltage vector interval HI is
different from the position of the non-zero voltage vector
interval H8. That is, the relative positions of the current
36

detection intervals are different from each other. Thus, the
detected current values differ, the current values calculated
by the phase current calculation unit 4 also differ, the
voltage commands generated by the voltage command generation
unit 5 differ, and the gate switching timings Sup to Swn also
differ, so that eventually, the shape of current ripple
changes every triangular carrier wave cycle. An example
thereof is shown as the current waveform Z for one phase in
FIG. 11. As shown in FIG. 11, the current waveform is
different between the first-half two cycles and the latter-
half two cycles of the triangular carrier wave cycles, and
thus frequency components of current ripple due to the :
triangular carrier wave cycle are dispersed. As a result,
the frequency that causes noise is dispersed, and harsh noise
is suppressed. [0040] As described above, the power conversion device
according to the present invention is useful for a power
conversion system widely applicable to various electric motors and systems, and in particular, suitable for a power I
conversion system that drives at a small triangular carrier wave frequency. Further, it is possible to drive a rotating electric machine by a rotating electric machine drive device using the power conversion device according to the present invention.
It is noted that, within the scope of the present

invention, the above embodiments may be freely combined with each other, or each of the above embodiments may be modified or eliminated appropriately.

We claim:
[1] A power conversion device comprising:
a PWM conversion unit which compares three-phase
voltage comands with a triangular carrier wave, and performs ;
conversion to PWM pulses;
a power conversion unit main circuit which drives switching elements on the basis of the PWM pulses, and converts DC voltage to three-phase AC voltage;
a DC bus current detection unit which detects j
current flowing through a DC bus of the power conversion unit ■
main circuit; j
a timing determination unit which sets, on the basis of switching timings of the switching elements, two |
detection timings for detecting currents for two phases by i
the DC bus current detection unit, at least once per one j
cycle of the triangular carrier wave;
a phase current calculation unit which calculates
three-phase AC current values once per one cycle of the I
triangular carrier wave on the basis of current values of the j
DC bus for the two phases detected at the timings determined j
by the timing determination unit; and i
a voltage command generation unit which updates the j
three-phase voltage commands on the basis of the three-phase
AC current values once per one cycle of the triangular j
carrier wave, wherein !

when the triangular carrier wave is divided into two intervals of a monotonic increase interval and a monotonic decrease interval, in a first cycle of two consecutive triangular carrier wave cycles, the DC bus current detection unit detects current values of the DC bus in one of the monotonic increase interval and the monotonic decrease interval. And in a second cycle of the two consecutive triangular carrier wave cycles, the DC bus current detection unit detects current values of the DC bus in the other one, of the monotonic increase interval and the monotonic decrease interval, in which detection has not been performed in the first cycle.
Further the phase current calculation unit calculates the three-phase AC current values, on the basis of the current values of the DC bus for two phases detected respectively by the DC bus current detection unit in the monotonic increase interval and the monotonic decrease interval in the last two consecutive triangular carrier wave cycles.
[2] The power conversion device according to claim 1, wherein
the DC bus current detection unit detects the currents for two phases in each of the monotonic increase interval and the monotonic decrease interval of the

triangular carrier wave cycle, and the phase current calculation unit calculates an average of the currents for each phase of the currents for the two phases.
[3] The power conversion device according to claim 1 or 2, further comprising a current detection interval generation unit which
adds a correction amount to the three-phase voltage
commands in the interval of a side on which current detection
is performed, of the monotonic increase interval and the
monotonic decrease interval of the triangular carrier wave
cycle, so as to ensure the detection timings, and subtracts the correction amount from the three-phase voltage commands in the interval of a side on which current detection is not performed, or
which subtracts a correction amount from the three-phase voltage commands in the interval of a side on which current detection is performed, and adds the correction amount to the three-phase voltage commands in the interval of a side on which current detection is not performed.
[4] The power conversion device according to any one of claims 1 to 3, wherein
on the basis of a switching timing TS of the switching element as to a middle-voltage phase corresponding

to the three-phase voltage command that has a second largest magnitude, the timing determination unit sets a detection timing for a first phase to be at a predetermined time period Tl before the switching timing TS, and sets a detection timing for a second phase to be at a predetermined time period T2 after the detection timing for the first phase,
the time period Tl is set to be longer than a time
period required for current detection, and :
the time period T2 is longer than a sum of: a dead '
time interval provided in order to prevent a pair of the
switching elements connected in series in the power
conversion unit main circuit from becoming conductive at the
same time; and an interval until high-frequency oscillation of DC current occurring at the time of switching of the ■
switching elements attenuates and an amplitude of the I
oscillation becomes equal to or smaller than a predetermined j
value. I
[5] The power conversion device according to any one of
claims 2 to 4, wherein a time period from the current detection timing for
the two phases in the monotonic increase interval to a next }
triangular carrier wave cycle start point is different from a time period from the triangular carrier wave cycle start point to the current detection timing for the two phases in

the monotonic decrease interval, and
a time period from the current detection timing for the two phases in the monotonic decrease interval to a next triangular carrier wave cycle start point is different from a time period from the triangular carrier wave cycle start point to the current detection timing for the two phases in the monotonic increase interval.
[6] The power conversion device according to any one of claims 2 to 5, wherein
in a case where difference occurs in time lags between timings of current calculation by the phase current calculation unit and timings of current detection by the DC bus current detection unit in the monotonic increase interval and the monotonic decrease interval, in current calculation by the phase current calculation unit, current detected at each timing is subjected to coordinate conversion by using a rotational angle at the timing, and an average value of current values in a rotating coordinate system at the respective timings is calculated.
[7] The power conversion device according to claim 6, wherein
the time lags in the case where current detection is performed in the monotonic increase interval are 0.5 cycle

and 1.5 cycles of the triangular carrier wave, and the time lags in the case where current detection is performed in the monotonic decrease interval are 1 cycle and 2 cycles of the triangular carrier wave.
[8] A rotating electric machine driving device that drives a rotating electric machine by using the power conversion device according to any one of claims 1 to 7.