FORM 2
THE PATENTS ACT, 1970
(39 of 1970)
&
THE PATENTS RULES, 2003
COMPLETE SPECIFICATION
[See section 10, Rule 13]
CONTROL DEVICE FOR ROTATING MACHINE
MITSUBISHI ELECTRIC CORPORATION, A CORPORATION ORGANISED
AND EXISTING UNDER THE LAWS OF JAPAN, WHOSE ADDRESS IS 7-3,
MARUNOUCHI 2-CHOME, CHIYODA-KU, TOKYO 1008310, JAPAN
THE FOLLOWING SPECIFICATION PARTICULARLY DESCRIBES THE
INVENTION AND THE MANNER IN WHICH IT IS TO BE PERFORMED.
2
DESCRIPTION
Field
[0001] The present disclosure relates to a control
device for a rotating machine that obtains rotor position
information for control without using a position sensor5
that detects a rotor position.
Background
[0002] In order for driving a rotating machine with full
use of capabilities, position information of a rotor is10
needed. Therefore, position information detected by a
position sensor attached to the rotating machine has been
used in the driving of the rotating machine. On the other
hand, position sensor–less driving techniques have been
developed for rotating machines in recent years from the15
perspectives of reducing manufacturing costs of rotating
machines further, downsizing rotating machines, and
improving reliability of rotating machines.
[0003] One position sensor–less control method for
rotating machines is to apply high-frequency signals to a20
rotating machine. In this method, stator currents are
first detected when high-frequency voltages are applied to
the rotating machine, and then high-frequency currents that
are components having the same frequencies as the high-
frequency voltages are extracted. Furthermore, the rotor25
position is estimated utilizing inductance of the rotating
machine, that is to say, the fact that amplitude of the
high-frequency current changes at a frequency twice an
electrical angular frequency of the rotor position. Such a
method of using the high-frequency signals has an advantage30
of satisfactorily estimating the rotor position even when
the rotating machine is in a zero-speed or low-speed range;
on the other hand, the method has a disadvantage of causing
3
torque pulsation and noise due to the superimposed high-
frequency voltages.
[0004] Other methods are also available. For example,
Patent Literature 1 given below discloses a method of
estimating the rotor position from stator voltages and5
stator currents of a rotating machine without applying
high-frequency signals. According to Patent Literature 1,
first, the stator voltages and the stator currents are
input into an observer. Then, the observer estimates a
component that, among flux linkage components, rotates10
synchronously with the rotor position and computes the
rotor position from a phase of the estimate to output the
rotor position.
Citation List15
Patent Literature
[0005] Patent Literature 1: Japanese Patent Application
Laid-open No. 2006-230174
Summary of Invention20
Problem to be solved by the Invention
[0006] In a conventional technique represented by the
one in Patent Literature 1, the stator voltages used for
the estimation of the rotor position are not actual
voltages, but stator voltage command values that are25
command values for the stator voltages. There is
inevitably an error between the stator voltage and the
stator voltage command value. Furthermore, a detection
error occurs in detecting the rotor current as well.
Therefore, with the conventional method, these voltage and30
current errors cause an error in an estimate of the rotor
position and in some cases, pulsating components as well.
If the rotor position estimate having such an estimation
4
error is used in the control of the rotating machine,
torque or power pulsates and may adversely affect a
connected mechanical or power system.
[0007] The present disclosure has been made in view of
the above, and an object of the present disclosure is to5
obtain a control device for a rotating machine that is
capable of reducing torque and power pulsations that result
from an estimation error that an estimate of rotor position
can include.
10
Means to Solve the Problem
[0008] In order to solve the above-stated problem and
achieve the object, a control device for a rotating machine
according to the present disclosure includes a voltage
application unit, a current detector, a control unit, a15
pulse-width modulator, and a position estimator. The
voltage application unit is connected between a direct-
current power supply and a rotating machine and applies a
rectangular stator voltage to the rotating machine through
on-off switching of a plurality of switching elements20
included for phases. The current detector detects a stator
current flowing between the voltage application unit and a
stator winding of the rotating machine. On the basis of
the stator current and a rotor position serving as position
information of a rotor of the rotating machine, the control25
unit computes a voltage command value that is a command
value for the stator voltage to be applied to the stator
winding. The pulse-width modulator performs on-off control
of the switching elements so that a smoothed value of the
stator voltage matches the voltage command value. The30
position estimator estimates, on the basis of the voltage
command value and the stator current, the rotor position
through a filter that removes a frequency component of a
5
fundamental frequency of rotational speed of the rotating
machine.
Effects of the Invention
[0009] The control device for a rotating machine5
according to the present disclosure has an effect of
reducing torque and power pulsations that result from an
estimation error that an estimate of the rotor position can
include.
10
Brief Description of Drawings
[0010] FIG. 1 is a diagram illustrating a configuration
example of a control device for a rotating machine
according to a first embodiment.
FIG. 2 is a diagram illustrating a configuration15
example of a main circuit of a three-phase inverter used as
a voltage application unit in FIG. 1.
FIG. 3 is a diagram provided for describing how a
pulse-width modulation (PWM) modulator illustrated in FIG.
1 operates.20
FIG. 4 is a diagram illustrating a configuration
example of a position estimator illustrated in FIG. 1.
FIG. 5 is a first diagram provided for explaining a
relationship between a switching period and a control
computation period in the first embodiment.25
FIG. 6 is a second diagram provided for explaining the
relationship between the switching period and the control
computation period in the first embodiment.
FIG. 7 is a diagram provided for explaining timing of
phase current detection in the first embodiment.30
FIG. 8 is a diagram provided for explaining a response
frequency in the first embodiment.
FIG. 9 is a diagram illustrating a configuration
6
example of a control device for a rotating machine
according to a second embodiment.
FIG. 10 is a diagram illustrating a configuration
example of a position estimator illustrated in FIG. 9.
FIG. 11 is a diagram illustrating a configuration5
example of a control device for a rotating machine
according to a third embodiment.
FIG. 12 is a diagram illustrating a configuration
example of a position estimator illustrated in FIG. 11.
FIG. 13 is a diagram illustrating a first hardware10
configuration that implements functions of each of the
control devices according to the first through third
embodiments.
FIG. 14 is a diagram illustrating a second hardware
configuration that implements the functions of each of the15
control devices according to the first through third
embodiments.
Description of Embodiments
[0011] With reference to the accompanying drawings, a20
detailed description is hereinafter provided of control
devices for rotating machines according to embodiments of
the present disclosure.
[0012] First Embodiment.
FIG. 1 is a diagram illustrating a configuration25
example of a control device 100 for a rotating machine
(hereinafter abbreviated as “control device” as
appropriate) according to a first embodiment. The control
device 100 according to the first embodiment includes a
voltage application unit 3, a current detector 4, a control30
unit 5, a PWM modulator 6, and a position estimator 7.
[0013] The voltage application unit 3 is connected
between a direct-current power supply 1 and a rotating
7
machine 2. The direct-current power supply 1 is a power
supply source that provides the rotating machine 2 with
driving power.
[0014] The rotating machine 2 is a three-phase motor in
which inductance varies with a rotor position. The5
rotating machine 2 includes a stator 2a with u-phase, v-
phase, and w-phase stator windings and a rotor 2b disposed
inside the stator 2a. Depending on an operating mode, the
rotating machine 2 also operates as a three-phase generator.
The rotating machine 2 is assumed herein to be, for example,10
a synchronous reluctance motor but may be a motor other
than the synchronous reluctance motor. A rotor direction
in which the inductance is maximized is defined herein as a
d-axis, and a direction in which the inductance is
minimized is defined herein as a q-axis. For the rotor15
position, the d-axis is used as a reference.
[0015] The current detector 4 is disposed between the
direct-current power supply 1 and the rotating machine 2.
The current detector 4 detects stator currents isu, isv, and
isw that flow between the voltage application unit 3 and20
the stator windings of the rotating machine 2.
[0016] The voltage application unit 3 applies
rectangular stator voltages to the rotating machine 2
through on-off switching of a plurality of switching
elements included for each of phases. The stator voltages25
are voltages that are applied to the stator windings of the
rotating machine 2. The voltage application unit 3 is
assumed herein to be a three-phase inverter.
[0017] The control unit 5 computes voltage command
values vsu*, vsv*, and vsw* on the basis of the stator30
currents isu, isv, and isw detected by the current detector 4
and the rotor position that serves as position information
of the rotor 2b. The voltage command values vsu*, vsv*, and
8
vsw* are command values for the stator voltages for driving
the rotating machine 2. The stator voltages that the
voltage application unit 3 outputs are controlled by the
voltage command values vsu*, vsv*, and vsw*.
[0018] In order for smoothed values of the rectangular5
stator voltages output by the voltage application unit 3 to
respectively match the voltage command values vsu*, vsv*, and
vsw*, the PWM modulator 6 generates gate signals gu, gv, and
gw that effect on-off control of the switching elements.
[0019] The position estimator 7 computes a rotor10
position estimate θ^r on the basis of the voltage command
values vsu*, vsv*, and vsw* and the stator currents isu, isv,
and isw. The rotor position estimate θ^r is an estimate of
the rotor position, which serves as the position
information of the rotor 2b. The rotor position estimate15
θ^r herein is a value converted into an electrical angle.
[0020] FIG. 2 is a diagram illustrating a configuration
example of a main circuit of the three-phase inverter used
as the voltage application unit 3 of FIG. 1. In FIG. 2,
the switching element 31 is a u-phase positive-side20
switching element, and the switching element 32 is a u-
phase negative-side switching element. Similarly, the
switching element 33 is a v-phase positive-side switching
element and the switching element 34 is a v-phase negative-
side switching element, and the switching element 35 is a25
w-phase positive-side switching element and the switching
element 36 is a w-phase negative-side switching element.
The switching elements 31 to 36 to be used are, for example,
insulated-gate bipolar transistors (IGBTs) illustrated but
may be switching elements other than the IGBTs. The30
switching elements other than the IGBTs are, for example,
metal-oxide-semiconductor field-effect transistors
(MOSFETs). A diode is connected across and in antiparallel
9
with one of the switching elements, correspondingly. The
term “antiparallel” refers to a connection configuration in
which an anode of the diode is connected to an emitter of
the IGBT, with a cathode of the diode connected to a
collector of the IGBT.5
[0021] A specific description is provided next of how
the control unit 5 operates. The control unit 5 includes a
current command value computation unit 501, a three-phase
to two-phase transformation unit 502, a rotating frame
transformation unit 503, a d-q current control unit 504, an10
inverse rotating frame transformation unit 505, and a two-
phase to three-phase transformation unit 506. A torque
command value T* is input to the control unit 5. The
control unit 5 computes the voltage command values vsu*, vsv*,
and vsw* in order for the rotating machine 2 to output15
torque corresponding to the torque command value T*.
[0022] The current command value computation unit 501
computes current command values isd* and isq* that are
command values for stator currents needed for the rotating
machine 2 to output the torque corresponding to the torque20
command value T*. The current command values isd*and isq*
are computed values in a rotating frame that rotates
synchronously with rotational speed of the rotating machine
2. The current command values isd*and isq* are computed so
that a root-mean-square current value is minimized for the25
torque, that is to say, copper loss of the rotating machine
2 is minimized for the torque.
[0023] The three-phase to two-phase transformation unit
502 transforms the stator currents isu, isv, and isw in a
three-phase frame to stator currents isα and isβ in a two-30
phase frame that is a stationary frame by three-phase to
two-phase transformation. A transformation matrix C32
shown in Formula (1) below is used herein for this
10
transformation operation.
[0024] Formula 1:
[0025] Using the rotor position estimate θ^r, the
rotating frame transformation unit 503 transforms the5
stator currents isα and isβ in the two-phase frame to stator
currents isd and isq in the rotating frame by rotating frame
transformation. A transformation matrix Cdq(θr) shown in
Formula (2) below is used herein for this transformation
operation.10
[0026] Formula 2:
[0027] The d-q current control unit 504 performs control
so that the stator currents isd and isq match the current
command values isd* and isq*, respectively, and computes15
voltage command values vsd* and vsq* in the rotating frame.
Proportional-integral control can be used for this control.
Control other than the proportional-integral control may be
used instead.
[0028] Using the rotor position estimate θ^r, the20
inverse rotating frame transformation unit 505 transforms
the voltage command values vsd* and vsq* in the rotating
frame to voltage command values vsα* and vsβ* in the two-
phase frame by inverse rotating frame transformation. An
inverse transformation matrix Cdq-1(θ^r) shown in Formula (3)25
below is used herein for this inverse transformation
operation.
[0029] Formula 3:
11
[0030] The two-phase to three-phase transformation unit
506 transforms the voltage command values vsα* and vsβ* in
the two-phase frame to the voltage command values vsu*, vsv*,
and vsw* in the three-phase frame by two-phase to three-5
phase transformation. A transformation matrix C23 shown in
Formula (4) below is used herein for this transformation
operation.
[0031] Formula 4:
10
[0032] FIG. 3 is a diagram provided for describing how
the PWM modulator 6 illustrated in FIG. 1 operates. FIG. 3
illustrates waveforms for the u phase as exemplary
waveforms for one of the phases.
[0033] In FIG. 3, a top section illustrates the waveform15
of the u-phase voltage command value vsu*, which is a
voltage command value for the u-phase, and a triangular
waveform of a carrier signal “c”. An upper-middle section
illustrates the waveform of the u-phase upper-side gate
signal gup, which is a gate signal for the u-phase upper20
side. A lower-middle section illustrates the waveform of
the u-phase lower-side gate signal gun, which is a gate
signal for the u-phase lower side. A bottom section
illustrates the waveform of a u-phase voltage vsu, which is
the stator voltage of the u phase. vdc represents a power25
supply voltage, which is the voltage of the direct-current
12
power supply 1. As illustrated in FIG. 3, half vdc/2 of
the power supply voltage vdc is a step width of the phase
voltage in this case, and the u-phase voltage command value
vsu* and the u-phase voltage vsu vary within a ±vdc/2 range.
[0034] The PWM modulator 6 compares the u-phase upper-5
side voltage command value vsu* with the carrier signal c.
When the u-phase upper-side voltage command value vsu* is
greater than a value of the carrier signal c, the PWM
modulator 6 sets the u-phase upper-side gate signal gup to
“H” and the u-phase lower-side gate signal gun to “L”.10
When the u-phase upper-side voltage command value vsu* is
less than or equal to a value of the carrier signal “c”,
the PWM modulator 6 sets the u-phase upper-side gate signal
gup to “L” and the u-phase lower-side gate signal gun to “H”.
Here “H” denotes “High”, and “L” denotes “Low”. When the15
u-phase upper-side gate signal gup=H and the u-phase lower-
side gate signal gun=L, the u-phase positive-side switching
element 31 of the voltage application unit 3 is turned on,
and the u-phase negative-side switching element 32 is
turned off. When the u-phase upper-side gate signal gup=L20
and the u-phase lower-side gate signal gun=H, the u-phase
positive-side switching element 31 of the voltage
application unit 3 is turned off, and the u-phase negative-
side switching element 32 is turned on. Operations for the
v-phase and the w-phase are identical with the operation25
for the u-phase.
[0035] The u-phase upper-side voltage command value vsu*
averages out in a switching period Tsw to the u-phase
voltage vsu, which is the actual output. The switching
period Tsw is equal to a carrier period, which is a period30
of the carrier signal c. Generally, when the positive-side
and negative-side switching elements are to be turned on
and off, a dead time, although not illustrated in FIG. 3,
13
is provided as a time during which both the switching
elements are off for the purpose of preventing both the
switching elements from being on at the same time. For
simplified description, a neutral point voltage that is an
average of the three phase voltages is ignored in the5
waveform of the u-phase voltage vsu illustrated in FIG. 3.
[0036] Adopted in the first embodiment is a technique of
synchronizing a switching frequency with an integer
multiple of a fundamental frequency fs of the rotational
speed of the rotating machine 2. The switching frequency10
is a reciprocal of the switching period Tsw. This
synchronization technique reduces lower-order harmonic
components even when the switching frequency is not
sufficiently high compared with the fundamental frequency
fs. As a result, the rotating machine 2 can be supplied15
with the stator voltage and the stator current that both
have low distortion. The phrase “when the switching
frequency is not sufficiently high” mentioned here applies
to, for example, cases where the switching frequency is 1
to 27 times as high as the fundamental frequency fs.20
[0037] Next, a description is provided of a principle of
estimation of the rotor position and the rotational speed
by the position estimator 7. To begin with, a rotating
machine model having characteristics of the rotating
machine 2 expressed mathematically is expressed in the two-25
phase frame by Formulas (5) and (6) below.
[0038] Formula 5:
Formula 6:
30
[0039] Here vsαβ represents stator voltage, isαβ
14
represents stator current, ψsαβ represents flux linkage, and
Rs represents winding resistance. The superscript “αβ”
indicates that the value is in the two-phase frame.
[0040] The inductance of the rotating machine 2 varies
with the rotor position. Above Formula (6) is an5
expression using a mean inductance component Lsavg for which
the inductance does not vary depending on the rotor
position and a variable inductance component Lsvar for which
the inductance varies at a frequency twice an electrical
angular frequency of the rotor position. The mean10
inductance component Lsavg and the variable inductance
component Lsvar are expressed respectively by Formulas (7)
and (8) below that use d-axis inductance Lsd and q-axis
inductance Lsq.
[0041] Formula 7:15
Formula 8:
[0042] As shown in Formula (9) below, an active flux
ψafdαβ with the d-axis as a reference can be extracted from20
the rotating machine model expressed by above Formulas (5)
and (6) by subtracting a product of the q-axis inductance
Lsq and the stator current isαβ from the flux linkage ψsαβ.
[0043] Formula 9:
25
[0044] The active flux ψafdαβ with the d-axis as the
reference is a component of the flux linkage ψsαβ that
rotates synchronously with the rotor position.
[0045] The stator current isαβ can be expressed by
Formula (10) below that uses its root-mean-square current30
value Iph and a current flow angle φi that is an angular
15
difference between the stator current isαβ and the rotor
position.
[0046] Formula 10:
[0047] Substituting above Formulas (6) and (10) into the5
right side of above Formula (9) gives Formula (11) below as
an expression of the active flux ψafdαβ with the d-axis as
the reference in the two-phase frame.
[0048] Formula 11:
10
[0049] As above Formula (11) shows, the active flux
ψafdαβ is the component generated by a product of the
variable inductance component Lsvar and the stator current
isd. Since the d-axis direction is used as the reference
for the active flux ψafdαβ of above Formula (11), the rotor15
position can be estimated by inputting this active flux
ψafdαβ into a publicly known observer.
[0050] It is to be noted that an active flux ψafqαβ with
the q-axis as a reference obtained by subtracting a product
of the d-axis inductance Lsd and the stator current isαβ
20
from the flux linkage ψsαβ, as expressed by Formula (12)
below instead of above Formula (9), can be used.
[0051] Formula 12:
[0052] As in the case of using the d-axis as the25
reference, substituting above Formulas (6) and (10) into
the right side of above Formula (12) gives Formula (13)
below as an expression of the active flux ψafqαβ with the q-
axis as the reference in the two-phase frame.
[0053] Formula 13:30
16
[0054] Since the q-axis direction is used as the
reference for the active flux ψafqαβ expressed by above
Formula (13), the rotor position can be estimated by
inputting this active flux ψafqαβ into a publicly known5
observer.
[0055] In the present embodiment, the rotor position is
estimated by inputting the active flux ψafd with the d-axis
as the reference into the above-mentioned observer
disclosed in Patent Literature 1. It is to be noted that10
the rotor position may be estimated with an observer other
than the observer disclosed in Patent Literature 1.
[0056] The observer expressed by Formula (14) in Patent
Literature 1 can be expressed by Formula (14) below that
uses the variable to be used herein.15
[0057] Formula 14:
[0058] In above Formula (14), ψ^safddq is an estimate of
the active flux with the d-axis as the reference. This
observer is expressed in the rotating frame synchronized20
with the estimated rotor position, and the superscript “dq”
indicates that the value is in the rotating frame. In
above Formula (14), ωr represents rotational angular speed,
and ωs represents rotational angular speed in the rotating
frame. Symbol J in above Formula (14) is a transformation25
matrix expressed by Formula (15) below.
[0059] Formula 15:
[0060] When an observer gain in the observer expressed
17
by above Formula (14) is set according to Patent Literature
1, the estimate of the active flux ψafd with the d-axis as
the reference can be obtained. Since the active flux ψafd
with the d-axis as the reference synchronizes with the
rotor position θr as shown in above Formula (11), the rotor5
position can be estimated by computing an arctangent of two
components from above Formula (11).
[0061] Above Formula (14) is the expression using the
observer but is fundamentally the expression that
integrates terms including stator voltage vsdq and stator10
current isdq. During high-speed rotation of the rotating
machine 2, the term with a product of the winding
resistance Rsdq and the stator current isdq is small compared
to the stator voltage vsdq and thus can be ignored here in
the computation of the flux linkage ψs that uses the15
integration. Voltage command value vsdq* can be used as the
stator voltage vsdq, and the detected values can be used for
the stator current isdq.
[0062] FIG. 4 is a diagram illustrating a configuration
example of the position estimator 7 illustrated in FIG. 1.20
The position estimator 7 can be configured to include
three-phase to two-phase transformation units 701 and 703,
rotating frame transformation units 702 and 704, an
observer 705, and a variable frequency notch filter 706.
[0063] The three-phase to two-phase transformation unit25
701 transforms the voltage command values vsu*, vsv*, and vsw*,
which are the command values for the stator voltages vs of
the phases in the three-phase frame, to voltage command
values vsα* and vsβ* in the two-phase frame by three-phase to
two-phase transformation. Using the rotor position30
estimate θ^r, the rotating frame transformation unit 702
transforms the voltage command values vsα* and vsβ* in the
two-phase frame to voltage command values vsd* and vsq* in
18
the rotating frame by rotating frame transformation. The
rotor position estimate θ^r used is an output of the
variable frequency notch filter 706, that is to say, an
output as feedback from the position estimator 7.
[0064] Similarly, the three-phase to two-phase5
transformation unit 703 transforms the stator currents isu,
isv, and isw in the three-phase frame to stator currents isα
and isβ in the two-phase frame by three-phase to two-phase
transformation. Using the rotor position estimate θ^r, the
rotating frame transformation unit 704 transforms the10
stator currents isα and isβ in the two-phase frame to stator
currents isd and isq in the rotating frame by rotating frame
transformation.
[0065] Using the observer mentioned earlier, the
observer 705 computes a rotor position estimate θ^r’ and a15
rotational angular speed estimate ω^r that is an estimate
of the rotational angular speed. The rotor position
estimate θ^r’ is a rotor position estimate that has not
undergone filter processing. According to Patent
Literature 1, not only the observer but also a phase lock20
loop is gone through for the estimation of the rotor
position and the rotational angular speed. The observer
705 herein, too, includes the function of the phase lock
loop. The position estimator 7 simply takes the values in
the three-phase frame as the inputs but is not limited to25
this. As illustrated in FIG. 1, the control unit 5
includes the three-phase to two-phase transformation unit
502 and the rotating frame transformation unit 503 and may
provide the values in the rotating frame as inputs.
[0066] The rotor position estimate θ^r’ computed by the30
observer 705 is input to the variable frequency notch
filter 706. Furthermore, the rotational angular speed
estimate ω^r computed by the observer 705 is input to the
19
variable frequency notch filter 706 as information
indicating a fundamental frequency component of the
rotational speed of the rotating machine 2. The variable
frequency notch filter 706 computes the rotor position
estimate θ^r on the basis of the rotor position estimate θ^r’
5
and the rotational angular speed estimate ω^r. The rotor
position estimate θ^r is the rotor position estimate that
has undergone the filter processing.
[0067] A supplemental explanation is provided here of
the computational processing according to the first10
embodiment. To begin with, suppose that Tpsi1 is a
computation period for the computation of the flux linkage
ψs that uses the observer. This computation period Tpsi1 is
not an integer multiple of half of the switching period Tsw.
Next, suppose that Tpsi2 is a computation period for the15
computation of the rotor position estimate θ^r that follows
the computation of the flux linkage ψs. This computation
period Tpsi2, too, is not an integer multiple of the half of
the switching period Tsw.
[0068] A description is provided next of a principle of20
the filter processing by the variable frequency notch
filter 706 according to the first embodiment. To begin
with, a transfer function of a notch filter that implements
the variable frequency notch filter 706 is expressed in an
analog domain by Formula (16) below.25
[0069] Formula 16:
[0070] In above Formula (16), ζr represents a damping
ratio. Furthermore, ω0 is a resonant angular frequency to
be removed by the variable frequency notch filter 706. In30
the processing described herein, the fundamental angular
frequency ωr corresponding to the fundamental frequency fs
20
is set. A relationship between the fundamental angular
frequency ωr and the fundamental frequency fs is expressed
by Formula (17) below.
[0071] Formula 17:
5
[0072] Expressing above Formula (16) in the form of a
digital filter by a bilinear transform gives Formula (18)
below.
[0073] Formula 18:
10
[0074] Coefficients a11, a12, b10, b11, and b12 in above
Formula (18) are expressed by the Formulas (19) to (23),
respectively.
[0075] Formula 19:
15
Formula 20:
Formula 21:
Formula 22:20
Formula 23:
[0076] In above Formulas (19) and (21), Tsmp is a
computation period of the filter processing. Based on25
these Formulas, a difference equation of the digital filter
is expressed by Formula (24) below.
[0077] Formula 24:
21
[0078] In above Formula (24), x represents an input
signal for the digital filter, and y represents an output
signal of the digital filter. By implementing the
processing using Formula (24), the variable frequency notch5
filter 706 according to the first embodiment is enabled to
implement its function. Above Formula (24) shows that
using at least two coefficients k1 and k2, the variable
frequency notch filter 706 can implement its function.
These two coefficients k1 and k2, are filter coefficients10
and are variables in the variable frequency notch filter
706.
[0079] In the first embodiment, the two variables of
above Formula (24) are precomputed and stored in a table
for each resonant angular frequency ω0, that is to say, for15
each fundamental frequency fs of the rotating machine 2.
Basically, the resonant angular frequency ω0 is set at a
value corresponding to the fundamental frequency fs of the
rotating machine 2, but with a lower limit. This is
because, when the resonant angular frequency ω0 is set20
below a response frequency for the position estimation,
interference between the filter processing by the variable
frequency notch filter 706 and the position estimation
processing by the position estimator 7 occurs, and response
for the rotor position estimate θ^r may be reduced and/or25
oscillation may occur. Here the lower limit for the
resonant angular frequency ω0 is set equal to or higher
than the response frequency for the position estimation
processing, where the rotor position estimate θ^r is
computed. If the lower limit for the resonant angular30
frequency ω0 is set higher than or equal to the response
frequency for the position estimation processing, the
22
operation is made possible without involving the above-
mentioned problem.
[0080] With reference to FIGS. 5 to 7, a description is
provided next of a relationship between the switching
period Tsw and a control computation period Tpsi. FIG. 5 is5
a first diagram provided for explaining the relationship
between the switching period Tsw and the control
computation period Tpsi in the first embodiment. FIG. 6 is
a second diagram provided for explaining the relationship
between the switching period Tsw and the control10
computation period Tpsi in the first embodiment. FIG. 7 is
a diagram provided for explaining timing of phase current
detection in the first embodiment. Here the computation
period Tpsi1 for the flux linkage ψs and the computation
period Tpsi2 for the rotor position estimate θ^r are equal,15
and the control computation period Tpsi, too, is equal to
each of the computation period Tpsi1 for the flux linkage ψs
and the computation period Tpsi2 for the rotor position
estimate θ^r.
[0081] Generally, command values are used in place of20
detected values in the control of a rotating machine as
values of the stator voltages. If the control computation
period Tpsi is an integer multiple of the half of the
switching period Tsw, the voltage command value and a
smoothed value of the actual voltage will be equal for each25
control computation period Tpsi. Note that when an actual
voltage average obtained by smoothing is used, the average
becomes substantially equal to the voltage command value.
[0082] Regarding the switching period Tsw and the
control computation period Tpsi, FIG. 5 illustrates a case30
where Tpsi=1×(Tsw/2), and FIG. 6 illustrates a case where
Tpsi=3×(Tsw/2). In each of the drawings, a top section
illustrates a waveform of the u-phase voltage command value
23
vsu* and a waveform of the carrier signal c, and a bottom
section illustrates a waveform of the u-phase voltage vsu.
The u-phase voltage command value vsu* takes on a sinusoidal
waveform.
[0083] In both the cases of FIGS. 5 and 6, it can be5
verified that the u-phase voltage vsu becomes substantially
equal to the u-phase voltage command value vsu* when
averaged over the control computation period Tpsi. At the
same time, it can be understood that if the control
computation period Tpsi is not the integer multiple of Tsw/2,10
the u-phase voltage vsu, which has been smoothed over each
control computation period Tpsi, does not match the u-phase
voltage command value vsu*.
[0084] FIG. 7 illustrates a relationship between the
three phase voltages and the u-phase current when the phase15
current is detected at peaks and valleys of the carrier
signal c. A top section illustrates waveforms of the
voltage command values vsu*, vsv*, and vsw* for the phases and
a waveform of the carrier signal c. A middle section
illustrates, from the top, a waveform of the u-phase20
voltage vsu, a waveform of a v-phase voltage vsv, and a
waveform of a w-phase voltage vsw. A bottom section
illustrates a waveform of the stator current isu. FIG. 7
also illustrates the timing of current detection being
synchronized with peaks and a valley of the carrier signal25
c. The timing of current detection may be considered
equivalent to control timing.
[0085] At the peaks and valleys of the carrier signal c,
values of the three phase voltages are all the same.
Therefore, at the peaks and valleys of the carrier signal c,30
a line voltage applied between lines of the stator 2a is
substantially zero. Consequently, at the peaks and valleys
of the carrier signal, changes in the stator currents are
24
small, and the three phase currents change gently, enabling
the current detection with no influence of ripple currents.
As described, at the peaks and valleys of the carrier
signal c, the stator currents can be detected during the
half of the switching period Tsw, with the influence of the5
ripple currents being removed.
[0086] As mentioned earlier, the switching frequency is
set at the integer multiple of the fundamental frequency fs.
It is to be noted here that the fundamental frequency fs of
the rotating machine 2 is not constant and changes from10
moment to moment. Therefore, a carrier frequency that is
equivalent to the switching frequency needs to be changed
in real time in accordance with the changing fundamental
frequency fs. Here in order to be the integer multiple of
the half of the switching period Tsw, the control15
computation period Tpsi needs to be changed sequentially in
real time by following typical rotating machine control.
Changing the control computation period Tpsi to realize this
while computing in the variable period leads to an
increased load of control computation and adds complexity20
to control design.
[0087] Therefore, in the first embodiment, the control
computation period Tpsi is a fixed value and is not
sequentially adjusted to an integer multiple of the half of
the switching period Tsw. In this way, the load of control25
computation is reduced, eliminating the need for an
expensive microprocessor or another such computing device.
Furthermore, the control design is made relatively simple.
In this case, the voltage command value vs* does not match
a smoothed value of the actual voltage. Consequently, the30
voltage command values vs* include errors with respect to
the actual voltages. Furthermore, with the control
computation period Tpsi not adjusted to the integer multiple
25
of the half of the switching period Tsw, the timing of
current detection is not synchronized with the peaks and
valleys of the carrier signal c. Therefore, the currents
detected by the current detector 4 also include errors with
respect to actual currents.5
[0088] As described above, when the control computation
period Tpsi is not adjusted to the integer multiple of the
half of the switching period Tsw, the stator voltages and
the stator currents can include errors. When the stator
voltages and the stator currents include the errors, an10
error occurs in the flux linkage ψs computed through the
use of these as well. Furthermore, since the computation
of the flux linkage ψs is fundamentally an integration
operation, influence of a direct-current component and
direct-current proximity components including low-frequency15
components is particularly significant. Errors in the
direct-current proximity components of the stator voltages
and currents, when transformed to the rotating frame that
rotates at the fundamental frequency fs synchronously with
the rotor position, become errors near the fundamental20
frequency fs. Since the computation of the position
estimate uses the flux linkage ψs in the rotating frame,
more precisely, the active flux ψafd with the d-axis as the
reference, an error near the fundamental frequency fs also
occurs in the rotor position estimate θ^r. If the rotor25
position estimate θ^r with the pulsating error is used in
the control of the rotating machine 2, the torque and power
pulsate. To deal with this, the position estimator 7
according to the first embodiment has the computed output
of the observer 705 go through the variable frequency notch30
filter 706 and uses the output of the variable frequency
notch filter 706 as the rotor position estimate θ^r.
Therefore, the position estimator 7 is capable of
26
estimating the position that removes the torque and power
pulsations resulting from the error near the fundamental
frequency fs.
[0089] Next, a summary of the effects of the above-
described control computation according to the first5
embodiment is provided. To begin with, in the first
embodiment, the switching frequency is synchronized with
the integer multiple of the fundamental frequency fs of the
rotating machine 2. As a result, even when the switching
frequency is low, the rotating machine 2 can be supplied10
with the stator voltage and the stator current that both
have low distortion. In the first embodiment, each of the
period Tpsi1 for the computation of the flux linkage ψs by
the observer 705 and the period Tpsi2 for the computation of
the rotor position estimate θ^r are not sequentially15
adjusted to the integer multiple of the half of the
switching period Tsw. In this way, the load of control
computation is reduced, eliminating the need for an
expensive microprocessor or another such computing device.
Furthermore, the control design is made relatively simple.20
Even with such a configuration, the error near the
fundamental frequency fs and the pulsation can be reduced
by the variable frequency notch filter 706 in the
estimation of the rotor position. Therefore, the control
device 100 can be configured to be position sensor-less and25
reduce torque and power pulsations without requiring the
expensive microprocessor, producing a notable effect non-
conventional.
[0090] As described above, the position estimator of the
control device for the rotating machine according to the30
first embodiment estimates, on the basis of the voltage
command values and the stator currents, the rotor position
through the variable frequency notch filter that removes
27
the frequency component of the fundamental frequency of the
rotational speed of the rotating machine. In this way, the
torque and power pulsations resulting from the estimation
error, which can be included in the estimate of the rotor
position, can be reduced.5
[0091] The estimate of the rotor position can be
computed from the phase of the estimate of the component
that, among the flux linkage components, rotates
synchronously with the rotor position. The flux linkage
used in this computation can be obtained by integrating at10
least the stator voltage command. In the computation of
the flux linkage by the integration, an offset component
can occur and result in an error in the estimate and
pulsation. However, the use of the method in the first
embodiment allows for a smaller error that could be15
included in the estimate and reduced pulsation.
[0092] In the control device for the rotating machine
according to the first embodiment, the PWM modulator
synchronizes the switching frequency at which the on-off
switching of the switching elements is performed with the20
integer multiple of the fundamental frequency of the
rotational speed of the rotating machine. As a result, the
rotating machine can be supplied with the stator voltage
and the stator current that both have low distortion.
Furthermore, in the computation of the flux linkage by the25
integration, an offset component can occur and result in an
error in the estimate and pulsation. However, the use of
this technique allows for reduction of such an offset
component, a smaller error that could be included in the
estimate, and reduced pulsation.30
[0093] The control device for the rotating machine
according to the first embodiment can enjoy its effects
when the computation period for the rotor position
28
estimation is not the integer multiple of the half of the
switching period. With the computation period for the
rotor position estimation not adjusted to the integer
multiple of the half of the switching period, the estimate
of the rotor position can include an error. However, the5
use of the method in the first embodiment allows for
reduction of such an error.
[0094] The control device for the rotating machine
according to the first embodiment can obtain its effects
when the computation period for the flux linkage10
computation is not the integer multiple of the half of the
switching period. When the computation period for the flux
linkage computation is not adjusted to the integer multiple
of the half of the switching period, the stator voltages
and the stator currents can include errors. However, the15
use of the method in first embodiment allows for reduction
of these errors.
[0095] In the control device for the rotating machine
according to the first embodiment, the lower limit is
preferably set for the frequency component to be removed by20
the filter. The lower limit is preferably equal to or
higher than the response frequency at which the rotor
position is estimated. Below that frequency, the response
frequency allows the control to follow. When a response
angular frequency in a first-order lag system in which the25
position estimation response is general is ωc, a frequency
characteristic of gain of the position estimation response
is illustrated as shown in FIG. 8, for example. In FIG. 8,
a vertical axis represents the gain. In FIG. 8, the gain
is substantially 1 at angular frequencies lower than and30
equal to the response angular frequency ωc, from which it
is verified that the position estimation system converges
sufficiently. If the above-mentioned resonant angular
29
frequency is set below the response frequency for the
position estimation, the interference between the filter
processing and the position estimation is conceivable,
which results in reduced response and the occurrence of
oscillation in the processing for obtaining the estimate of5
the rotor position. On the other hand, when the lower
limit for the frequency component to be removed by the
filter is set and further the lower limit is set equal to
or higher than response frequency for the position
estimation processing, such problems are avoidable.10
[0096] Second Embodiment.
FIG. 9 is a diagram illustrating a configuration
example of a control device 100A for a rotating machine
according to a second embodiment. Compared with the
control device 100 illustrated in FIG. 1, the control15
device 100A according to the second embodiment includes a
position estimator 8 in FIG. 9 that replaces the position
estimator 7. The configuration is otherwise identical or
equivalent to that of the control device 100, and identical
or equivalent constituent elements have the same reference20
characters and are not redundantly described.
[0097] FIG. 10 is a diagram illustrating a configuration
example of the position estimator 8 illustrated in FIG. 9.
The position estimator 8 can be configured to include
three-phase to two-phase transformation units 801 and 802,25
a rotating frame transformation unit 803, a first
computation unit 804, a first estimator 805, a second
computation unit 806, a variable frequency notch filter 807,
and a third computation unit 808.
[0098] The three-phase to two-phase transformation unit30
801 transforms the voltage command values vsu*, vsv*, and vsw*,
which are the command values for the stator voltages vs of
the phases in the three-phase frame, to voltage command
30
values vsα* and vsβ* in the two-phase frame by three-phase to
two-phase transformation. Similarly, the three-phase to
two-phase transformation unit 802 transforms the stator
currents isu, isv, and isw in the three-phase frame to stator
currents isα and isβ in the two-phase frame by three-phase5
to two-phase transformation. Using the rotor position
estimate θ^r, the rotating frame transformation unit 803
transforms the stator currents isα and isβ in the two-phase
frame to stator currents isd and isq in the rotating frame
by rotating frame transformation.10
[0099] A description is provided next of processing
details of the first computation unit 804 and the first
estimator 805. The first computation unit 804 computes a
flux-linkage inductance variation component. The first
estimator 805 provides an estimate of the flux-linkage15
inductance variation component.
[0100] To begin with, the flux linkage ψsαβ of the
rotating machine 2 in the two-phase frame is determined by
Formula (25) below.
[0101] Formula 25:20
[0102] An integration part of above Formula (25) is
expressed by a transfer function shown by Formula (26)
below.
[0103] Formula 26:25
[0104] Generally, when flux linkage is computed by
integration, an initial value is usually unknown.
Therefore, a high-pass filter (HPF) with a sufficiently low
cutoff frequency compared to a fundamental frequency30
component is used in computing flux linkages in the
31
stationary frames, which are the three-phase frame and the
two-phase frame. This method, which is a method of
computing the flux linkage in the stationary frame through
the use of the integration and the HPF, is called herein
“incomplete integration”. A transfer function of the high-5
pass filter used in this incomplete integration can be
expressed by Formula (27) below, where ωhpf is the cutoff
frequency.
[0105] Formula 27:
10
[0106] Applying the HPF expressed by above Formula (27)
to above Formula (26) gives Formula (28) below.
[0107] Formula 28:
[0108] Above formula (28) is a formula expressing flux15
linkage ψshpfαβ, with the HPF applied. Above Formula (28) is
changed into Formula (29) below.
[0109] Formula 29:
[0110] In the position sensor–less control of the20
synchronous reluctance motor, the method of utilizing the
incomplete integration for the flux linkage computation is
possible. With the method of utilizing the incomplete
integration, a computational load is low compared to when
an observer is used, allowing for the use of a less25
expensive microprocessor or another such computing device.
As in the first embodiment, during high-speed rotation of
the rotating machine 2, the term with a product of the
winding resistance Rs and the stator current isαβ in above
32
Formula (28) is small compared to a voltage command value
vsαβ* and thus can be ignored. In the computation of the
flux linkage ψshpfαβ of above Formula (29), the command value
vsαβ* is used as the stator voltage, and detected values are
used for the stator current isαβ. Furthermore, in the5
second embodiment, the period Tpsi1 for the flux linkage
computation utilizing the incomplete integration is not an
integer multiple of half of the switching period Tsw, and
the period Tpsi2 for the subsequent computation of the rotor
position estimate θ^r, too, is not an integer multiple of10
the half of the switching period Tsw.
[0111] The flux linkage ψsαβ of the rotating machine 2 in
the two-phase frame is expressed by above Formula (6).
When subjected to rotating frame transformation using the
rotor position estimate θ^r, this flux linkage ψsαβ can be15
expressed by Formula (30) below.
[0112] Formula 30:
[0113] In above Formula (30), the first term is a term
including the mean inductance component Lsavg, which does20
not vary with the rotor position, and the second term is a
term including the variable inductance component Lsvar,
which varies at the frequency twice that of the rotor
position.
[0114] The first computation unit 804 determines the25
component corresponding to the second term of above Formula
(30) by computation. Specifically, the computation is
performed according to Formula (31) below that is obtained
by changing above Formula (30).
[0115] Formula 31:30
33
[0116] The first term in the right side of above Formula
(31) is determined by subjecting the flux linkage ψshpfαβ
expressed by above Formula (29) to rotating frame
transformation. The second term in the right side of above5
Formula (31) shows the first term of above Formula (30). A
configuration example of the first computation unit 804 is
illustrated in FIG. 10 but is not limiting.
[0117] The first estimator 805, on the other hand,
directly estimates the component corresponding to the10
second term of above Formula (30). A configuration example
of the first estimator 805 is illustrated in FIG. 10. A
description is provided for a reason why the first
estimator 805 can be configured in such a simple manner.
[0118] To begin with, suppose that the second term of15
above Formula (30) is the estimate of the flux-linkage
inductance variation component in the rotating frame. This
estimate can be expressed as ψ^svardq by Formula (32) below.
[0119] Formula 32:
20
[0120] When the estimate θ^r of the rotor position is
approximated to be about equal to the true value θr of the
rotor position in above Formula (32), above Formula (32) is
simplified into Formula (33) below. A configuration of a
controller representing this Formula (33) is illustrated in25
FIG. 10.
[0121] Formula 33:
[0122] A description is provided next of processing
details of the second computation unit 806, the variable30
34
frequency notch filter 807, and the third computation unit
808.
[0123] To begin with, a cross product of the estimate
ψ^svardq of the flux-linkage inductance variation component
and the computed value ψsvar,calcdq is expressed by Formula5
(34) below.
[0124] Formula 34:
[0125] When the estimate θ^r of the rotor position is
approximated to be about equal to the true value θr of the10
rotor position in above Formula (34), that is to say, θ^r≈θr,
a rotor position estimation error “-(θ^r-θr)” can be
computed by Formula (35) below.
[0126] Formula 35:
15
[0127] In the above-described manner, the second
computation unit 806 computes the rotor position estimation
error “-(θ^r-θr)” on the basis of the computed value
obtained by above Formula (31) and the estimate obtained by
above Formula (33).20
[0128] The rotor position estimation error “-(θ^r-θr)”
computed by the second computation unit 806 is input to the
variable frequency notch filter 807 to undergo filter
processing and is then input to the third computation unit
808. The third computation unit 808 computes the rotor25
position estimate θ^r by performing, in converging the
rotor position estimation error “-(θ^r-θr)” to zero,
proportional-integral (PI) control on the rotor position
estimation error “-(θ^r-θr)” and integration thereafter.
Furthermore, the third computation unit 808 computes the30
35
rotational speed estimate ω^r in its process of converging
the rotor position estimation error “-(θ^r-θr)” to zero.
The variable frequency notch filter 807 to be used can be
identical or equivalent to the variable frequency notch
filter 706 described in the first embodiment.5
[0129] According to the second embodiment described
above, the control device for the rotating machine enables
the rotor position estimation method using the variable
frequency notch filter, which removes the frequency
component of the fundamental frequency of the rotational10
speed of the rotating machine, to be applied to the
configuration for the computation of the flux linkage in
the stationary frame. Since the flux linkage is computed
in the stationary frame by the integration in the position
sensor–less control of the synchronous reluctance motor as15
the rotating machine, an offset component occurs, easily
resulting in an error in the estimate and pulsation.
Therefore, the method in the second embodiment can be
suitably used in the position sensor–less control of the
synchronous reluctance motor.20
[0130] Next, a summary of the effects of the above-
described control computation according to the second
embodiment is provided. To begin with, in the second
embodiment, the period Tpsi1 for the computation of the flux
linkage ψs that uses above Formula (29) utilizing the25
incomplete integration and the period Tpsi2 for the
computation of the rotor position estimate θ^r are both not
the integer multiple of the half of the switching period
Tsw. In this case, the voltage command value vs* does not
match a smoothed value of the actual voltage. Consequently,30
the voltage command values vs* include errors with respect
to the actual voltages. Furthermore, since the timing of
current detection is not synchronized with the peaks and
36
valleys of the carrier signal c, the detected currents
include errors with respect to the actual currents.
Therefore, an error also occurs in the flux linkage ψs
computed through the use of these. In the second
embodiment, the computation of the flux linkage ψs does not5
use an observer that converges the flux linkage ψs to a
true value, but the incomplete integration. Therefore, the
error in the flux linkage ψs is greater, and the
convergence to the true value is relatively slow.
Furthermore, since the flux linkage computation is based on10
the integration operation, the error becomes greater in a
range from a direct-current component to a low-frequency
component. As a result, an error near the fundamental
frequency fs becomes greater in the rotating frame, and a
greater error occurs near the fundamental frequency fs in15
the rotor position estimate θ^r as well. To deal with this
problem, the position estimator 8 according to the second
embodiment has the output of the second computation unit
806 undergo the reduction at the variable frequency notch
filter 807 before being input to the third computation unit20
808. Therefore, in the rotor position estimation, the
position estimator 8 can reduce the error near the
fundamental frequency and pulsation. The position
estimator 8 according to the second embodiment does not use
the observer, but the incomplete integration for computing25
the flux linkage, thus allowing reduction of the
computational load compared with that of the first
embodiment. Therefore, through the use of the method in
the second embodiment, the control device 100A can be
configured to be position sensor-less and reduce torque and30
power pulsations without requiring an expensive
microprocessor, which produces a notable effect non-
conventional.
37
[0131] Third Embodiment.
FIG. 11 is a diagram illustrating a configuration
example of a control device 100B for a rotating machine
according to a third embodiment. Compared with the control
device 100 illustrated in FIG. 1, the control device 100B5
according to the third embodiment includes a position
estimator 9 in FIG. 11 that replaces the position estimator
7. The configuration is otherwise identical or equivalent
to that of the control device 100, and identical or
equivalent constituent elements have the same reference10
characters and are not redundantly described.
[0132] In the third embodiment, flux linkage computation
is performed without using integration in estimating the
rotor position and the rotational speed. Here a
description is provided first of a principle of the15
estimation of the rotor position and the rotational speed
by the position estimator 9. To begin with, a rotating
machine model having the characteristics of the rotating
machine 2 expressed mathematically is expressed in the
rotating frame by Formulas (36) and (37) below.20
[0133] Formula 36:
Formula 37:
[0134] Symbol J in above Formula (36) is the25
transformation matrix expressed by above Formula (15).
[0135] In the third embodiment, an inductance value is
computed. Therefore, above Formula (37) is expressed as
Formula (38) below.
[0136] Formula 38:30
38
[0137] In above Formula (38), Lsd,calc represents computed
d-axis inductance, and Lsq,calc represents computed q-axis
inductance.
[0138] Since induced voltage ωrJψsdq in the third term in5
the right side of above Formula (36) is also computed, this
induced voltage ωrJψsdq is represented by vemf,calc. When the
derivative term of above Formula (36), that is to say, the
second term in the right side of above Formula (36) is
ignored here, the induced voltage vemf,calc can be computed10
as the computed value by using Formula (39) below that uses
the stator voltage vsdq and the stator current isdq.
[0139] Formula 39:
[0140] The voltage command value vsdq* is used as the15
stator voltage vsdq, and detected values are used for the
stator current isdq.
[0141] An induced voltage estimate v^emf that is an
estimate of induced voltage vemf can be obtained by using
Formula (40) below that uses flux linkage ψs,calcdq computed20
by above Formula (38) and the rotational speed estimate ω^r.
[0142] Formula 40:
[0143] As described above, the rotational speed estimate
ω^r, which is the estimate of the rotational speed ωr, can25
be obtained by comparing the computed value obtained by
Formula (39) above and the estimate obtained by above
Formula (40) and performing proportional-integral control
to converge a difference to zero.
[0144] Dividing the induced voltage vemf,calc computed by30
above Formula (39) by the rotational speed estimate ω^r
39
gives a computed value of the flux linkage ψs, and further
dividing this computed value by the stator current is gives
the computed inductance value.
[0145] As shown in above Formula (6), the inductance
value varies depending on the true rotor position θr.5
Furthermore, as shown in above Formula (30), the inductance
value varies depending on a difference between the true
rotor position θr and the rotor position estimate θ^r.
Therefore, the rotor position can be estimated by checking
the computed inductance value against these inductance10
variation characteristics. Specifically, the inductance
value is computed by dividing the flux linkage ψsdq, which
includes the flux-linkage inductance variation component
generated by the product of the variable inductance
component Lsvar and the stator current is, by the stator15
current isdq. The rotor position estimate θ^r can be
obtained from a rotor position–dependent inductance
variation characteristic of the inductance value.
[0146] FIG. 12 is a diagram illustrating a configuration
example of the position estimator 9 illustrated in FIG. 11.20
The position estimator 9 can be configured to include
three-phase to two-phase transformation units 901 and 903,
rotating frame transformation units 902 and 904, variable
frequency notch filters 905 and 906, and a speed and angle
computation unit 907.25
[0147] The three-phase to two-phase transformation unit
901 transforms the voltage command values vsu*, vsv*, and vsw*,
which are the command values for the stator voltages vs of
the phases in the three-phase frame, to voltage command
values vsα* and vsβ* in the two-phase frame by three-phase to30
two-phase transformation. The three-phase to two-phase
transformation unit 903 transforms the stator currents isu,
isv, and isw in the three-phase frame to stator currents isα
40
and isβ in the two-phase frame by three-phase to two-phase
transformation. Using the rotor position estimate θ^r, the
rotating frame transformation unit 902 transforms the
voltage command values vsα* and vsβ* in the two-phase frame
to voltage command values vsd* and vsq* in the rotating frame5
by rotating frame transformation. Using the rotor position
estimate θ^r, the rotating frame transformation unit 904
transforms the stator currents isα and isβ in the two-phase
frame to stator currents isd and isq in the rotating frame
by rotating frame transformation. The rotor position10
estimate θ^r used is one of the outputs from the speed and
angle computation unit 907 that is provided as feedback.
[0148] The output of the rotating frame transformation
unit 902 goes through the variable frequency notch filter
905 before being input to the speed and angle computation15
unit 907. Similarly, the output of the rotating frame
transformation unit 904 goes through the variable frequency
notch filter 906 before being input to the speed and angle
computation unit 907. The speed and angle computation unit
907 computes the rotor position estimate θ^r and the20
rotational speed estimate ω^r according to what has been
described earlier.
[0149] Since the method in the third embodiment does not
use an observer or incomplete integration for the flux
linkage computation, a computation period may be long25
compared to when the observer or the incomplete integration
is used. For this reason, a computational load is lower,
allowing for the use of a less expensive microprocessor or
another such computing device. In the third embodiment, a
computation period Tpsi3 for the computation of the rotor30
position estimate θ^r and the rotational speed estimate ω^r
is not an integer multiple of half of the switching period
Tsw as in the first and second embodiments. In this case,
41
the voltage command value vs* does not match a smoothed
value of the actual voltage. Consequently, the voltage
command values vs* include errors with respect to the
actual voltages. Furthermore, since the timing of current
detection is not synchronized with the peaks and valleys of5
the carrier signal, the currents detected by the current
detector 4 also include errors with respect to the actual
currents. Therefore, errors also occur in the rotor
position estimate θ^r and the rotational speed estimate ω^r
that are computed through the use of these.10
[0150] In the rotating machine 2, lower-frequency errors
generate greater oscillation components in flux and torque.
A direct-current component and low-frequency direct-current
proximity components in the stationary frame translate into
errors near the fundamental frequency fs in the rotating15
frame. To deal with this, the position estimator 9
according to the third embodiment has the outputs of the
rotating frame transformation units 902 and 904 go
respectively through the variable frequency notch filters
905 and 906 before being input to the speed and angle20
computation unit 907. Therefore, the position estimator 9
is capable of the position estimation that removes torque
and power pulsations resulting from the error near the
fundamental frequency fs.
[0151] As described above, the position estimator 925
according to the third embodiment computes the flux linkage
without using the observer and the incomplete integration,
thus allowing for the low computational load compared with
those of the first and second embodiments. Therefore,
through the use of the method in the third embodiment, the30
control device 100B can be configured to be position
sensor-less and reduce torque and power pulsations without
requiring an expensive microprocessor, producing a notable
42
effect non-conventional.
[0152] With reference to FIGS. 13 and 14, a description
is provided next of hardware configurations for the above-
described control devices 100, 100A, and 100B according to
the first through third embodiments. FIG. 13 is a diagram5
illustrating a first hardware configuration that implements
the functions of each of the control devices 100, 100A, and
100B according to the first through third embodiments. FIG.
14 is a diagram illustrating a second hardware
configuration that implements the functions of each of the10
control devices 100, 100A, and 100B according to the first
through third embodiments. The functions of each of the
control devices 100, 100A, and 100B refer to the functions
of the control unit 5, the PWM modulator 6, and the
position estimator 7, 8, or 9 that are included in the15
control device 100, 100A, or 100B.
[0153] The functions of the control unit 5, the PWM
modulator 6, and the position estimator 7, 8, or 9 can be
implemented with processing circuitry. In FIG. 13, the
control unit 5, the PWM modulator 6, and the position20
estimator 7, 8, or 9 of each of the first through third
embodiments have been replaced by dedicated processing
circuitry 10. Corresponding to the dedicated processing
circuitry 10 used as dedicated hardware is a single circuit,
a composite circuit, an application-specific integrated25
circuit (ASIC), a field-programmable gate array (FPGA), or
a combination of these. The functions of the control unit
5, the PWM modulator 6, and the position estimator 7, 8, or
9 may be implemented individually or collectively with the
processing circuitry.30
[0154] In FIG. 14, the control unit 5, the PWM modulator
6, and the position estimator 7, 8, or 9 in the
configuration according to each of the first through third
43
embodiments have been replaced by a processor 11 and a
memory device 12. The processor 11 may be an arithmetic
means such as an arithmetic unit, a microprocessor, a
microcomputer, a central processing unit (CPU), or a
digital signal processor (DSP). The memory device 12 can5
be, for example, a nonvolatile or volatile semiconductor
memory such as a random-access memory (RAM), a read-only
memory (ROM), a flash memory, an erasable programmable ROM
(EPROM), or an electrically EPROM (EEPROM) (registered
trademark).10
[0155] When the processor 11 and the memory device 12
are used, the functions of the control unit 5, the PWM
modulator 6, and the position estimator 7, 8, or 9 are
implemented with software, firmware, or a combination of
these. The software or the firmware is described as15
programs and is stored in the memory device 12. The
processor 11 reads and executes the programs stored in the
memory device 12. These programs can be said to cause a
computer to execute procedures and methods for the
functions of the control unit 5, the PWM modulator 6, and20
the position estimator 7, 8, or 9. Usable examples of the
memory device 12 include the nonvolatile and volatile
semiconductor memories such as the ROM, the EPROM, and the
EEPROM, a flexible disk, an optical disk, a compact disk,
and a DVD, among others. The memory device 12 can store25
the above-mentioned two coefficients k1 and k2 for each of
those frequencies that the variable frequency notch filters
706, 807, 905, and 906 remove.
[0156] The functions of the control unit 5, the PWM
modulator 6, and the position estimator 7, 8, or 9 may be30
implemented partly with hardware and partly with software
or firmware. For example, the function of the PWM
modulator 6 may be implemented with dedicated hardware,
44
with the functions of the control unit 5 and the position
estimator 7, 8, or 9 implemented with the processor 11 and
the memory device 12.
[0157] While the synchronous reluctance motor is the
example of the rotating machine 2 in the second and third5
embodiments described herein, the rotating machine 2 may be
an induction motor or a permanent magnet motor. When the
rotating machine 2 is the induction motor, a method
disclosed in Japanese Patent Application Laid-open No. H11-
4599, for example, can be used. When the rotating machine10
2 is the permanent magnet motor, a method disclosed in PCT
International Publication No. 2002/091558, for example, can
be used. Part of the method in the third embodiment uses a
method described in Japanese Patent Application Laid-open
No. 2002-165475. Therefore, for details not described in15
the third embodiment, refer to contents of this publication.
[0158] While the voltage application unit 3 used and
described herein is the three-phase 2-level inverter, this
is not limiting. The voltage application unit 3 may be an
inverter with a different number of phases. The voltage20
application unit 3 may be a multi-level inverter such as a
3-level inverter or a 5-level inverter. These inverters
can even be used in implementing control devices for
rotating machines according to the present disclosure.
[0159] The switching frequency described herein as the25
example is 1 to 27 times the fundamental frequency fs.
Generally, when, for example, a common carrier signal is
used for the three phases, the switching frequency to be
used is one times as high as the fundamental frequency fs
or 3, 6, 9, ..., or 27 times as high as the fundamental30
frequency fs, where these numbers are multiples of 3. On
the other hand, when a fixed switching pattern is used
without utilizing a carrier signal, any multiple is usable
45
as long as the multiple is an integer multiple.
[0160] While the stator currents described herein are
set for the torque of the rotating machine 2 to minimize
the root-mean-square current value, this is not limiting.
The stator currents for the torque of the rotating machine5
2 may be set to minimize the flux linkage or maximize
efficiency of the voltage application unit 3 or the
rotating machine 2.
[0161] In the example shown in each of the first and
second embodiments described herein, the variable frequency10
notch filter 706 or 807 is inserted in series with the part
that performs the rotor position estimation processing. In
the example shown in the third embodiment, the variable
frequency notch filters 905 and 906 are inserted
respectively in series with the parts that output the15
stator voltages and the stator currents. These are just
the examples. The insertion can be at an appropriately
selected position based on where the error to be removed
occurs. Furthermore, the single variable frequency notch
filter for each position does not have to be the case;20
plural variable frequency notch filters may be inserted for
each position.
[0162] While the voltage command values are used herein
as the stator voltages to be used in the control
computation, detected stator voltages may be used instead.25
[0163] The above configurations illustrated in the
embodiments are illustrative, can be combined with other
techniques that are publicly known, and can be partly
omitted or changed without departing from the gist. The
embodiments can be combined with each other.30
Reference Signs List
[0164] 1 direct-current power supply; 2 rotating
46
machine; 2a stator; 2b rotor; 3 voltage application
unit; 4 current detector; 5 control unit; 6 PWM
modulator; 7, 8, 9 position estimator; 10 dedicated
processing circuitry; 11 processor; 12 memory device; 31
to 36 switching element; 100, 100A, 100B control device;5
501 current command value computation unit; 502, 701, 703,
801, 802, 901, 903 three-phase to two-phase transformation
unit; 503, 702, 704, 803, 902, 904 rotating frame
transformation unit; 504 d-q current control unit; 505
inverse rotating frame transformation unit; 506 two-phase10
to three-phase transformation unit; 705 observer; 706, 807,
905, 906 variable frequency notch filter; 804 first
computation unit; 805 first estimator; 806 second
computation unit; 808 third computation unit; 907 speed
and angle computation unit.15
47
WE CLAIM:
[Claim 1] A control device for a rotating machine, the
control device comprising:
a voltage application unit connected between a direct-
current power supply and a rotating machine to apply a5
rectangular stator voltage to the rotating machine through
on-off switching of a plurality of switching elements
included for each of phases;
a current detector to detect a stator current flowing
between the voltage application unit and a stator winding10
of the rotating machine;
a control unit to compute a voltage command value on a
basis of the stator current and a rotor position serving as
position information of a rotor of the rotating machine,
the voltage command value being a command value for a15
stator voltage to be applied to the stator winding;
a pulse-width modulator to perform on-off control of
the switching elements so that a smoothed value of the
stator voltage matches the voltage command value; and
a position estimator to estimate, on a basis of the20
voltage command value and the stator current, the rotor
position through a filter that removes a frequency
component of a fundamental frequency of rotational speed of
the rotating machine.
25
[Claim 2] The control device for a rotating machine
according to claim 1, wherein
the pulse-width modulator synchronizes a switching
frequency, with which the on-off switching of the switching
elements is performed, with an integer multiple of a30
fundamental frequency of rotational speed of the rotating
machine.
48
[Claim 3] The control device for a rotating machine
according to claim 1 or 2, wherein
an estimate of the rotor position is computed on a
basis of a flux linkage of the rotating machine, and
the flux linkage is obtained by integrating at least5
the voltage command value.
[Claim 4] The control device for a rotating machine
according to any one of claims 1 to 3, wherein
an estimate of the rotor position is computed on a10
basis of a flux linkage of the rotating machine, and
the flux linkage is obtained by integrating at least
the voltage command value in a stationary frame.
[Claim 5] The control device for a rotating machine15
according to claim 3 or 4, wherein
a computation period for computation of the flux
linkage is not an integer multiple of half of a switching
period, the switching period being a reciprocal of a
switching frequency with which the on-off switching of the20
switching elements is performed.
[Claim 6] The control device for a rotating machine
according to any one of claims 1 to 5, wherein
a computation period for estimation of the rotor25
position is not an integer multiple of half of a switching
period, the switching period being a reciprocal of a
switching frequency with which the on-off switching of the
switching elements is performed.
30
[Claim 7] The control device for a rotating machine
according to any one of claims 1 to 6, wherein
a lower limit is set for a frequency component to be
49
removed by the filter.
[Claim 8] The control device for a rotating machine
according to claim 7, wherein
the lower limit is equal to or higher than a response5
frequency at which the rotor position estimation is
performed.
[Claim 9] The control device for a rotating machine
according to any one of claims 1 to 8, wherein10
the position estimator includes a memory device that
stores a filter coefficient for implementing the filter,
for each of frequencies to be removed by the filter.
[Claim 10] The control device for a rotating machine15
according to any one of claims 1 to 9, wherein
the rotating machine has a variable inductance
component with which inductance varies depending on a rotor
position, and
the position estimator estimates the rotor position on20
a basis of a flux-linkage inductance variation component
generated by a product of the variable inductance component
and the stator current.
[Claim 11] The control device for a rotating machine25
according to claim 10, wherein
inductance of the rotating machine includes a mean
component that does not vary depending on the rotor
position and a variable component that varies at a
frequency twice an electrical angular frequency of the30
rotor position, and
50
the flux-linkage inductance variation component is
generated by a product of the variable component and the
stator current.