 |
Description  |
|
|
TECHNICAL FIELD
The present invention relates to induction motor drives in general, and
more particularly to an induction motor torque and flux system, and to
current control apparatus for such system.
BACKGROUND OF THE INVENTION
Field-oriented control for an AC motor drive is well known. Based on a
reference frame which rotates at the speed of the rotor flux, a flux
component and a torque component of the stator currents oriented upon such
reference frame are calculated and used to control the motor flux and the
resulting torque. See for instance FIELD-ORIENTED CONTROL OF A STANDARD AC
MOTOR USING MICROPROCESSORS by R. Gabriel, W. Leonard and C. J. Nordby,
IEEE Trans. IA-16, pages 186-192, March/April 1980; INTRODUCTION TO FIELD
ORIENTATION AND HIGH PERFORMANCE AC DRIVES by D. W. Novotny and R. D.
Lorenz, IEEE Industry Applications Society, Oct. 6--6, 1985, Toronto,
Canada, Section 2, pages 2-1 to 2-65. The two afore-cited publications are
hereby incorporated by reference. The afore-mentioned W. Leonard and the
D. W. Novotny and R. D. Lorenz publications are hereby incorporated by
reference.
The assumption is that the motor flux .psi.* and T* demand signals can be
instantaneously satisfied under the further assumption that the
mathematical model used is accurate, that the parameter T2, namely the
rotor time constant, is known and that the specified direct and quadrature
current components i.sub.d and i.sub.q can be instantaneously injected
into the stator winding.
Direct and quadrature stator currents have been generated for control
according to the vector control method described in U.S. Pat. No.
4,456,868 of Yamamura et al. The purpose, there, is to improve the
response on the torque.
It is also known from U.S. Pat. No. 4,125,796 of Nagase et al. to generate
a desired torque by calculating a current pattern signal, also by
decomposing the motor current into a flux oriented direction and in
quadrature thereto.
U.S. Pat. No. 4,451,771 of Nagase et al. discloses the generation of a
current correction signal applied to the current control signal derived
according to the motor control method in an AC motor drive.
The object of the present invention is to achieve a speed regulator
providing dynamic control of both the motor speed and the magnetic flux
level, thereby to ensure that control is maintained over the
field-weakening operative range of the motor drive.
The present invention involves a speed regulator system wherein both the
torque and flux references are variables. The torque demand is derived
from the speed regulator error signal and the motor flux reference is a
predefined function of the motor speed.
As long as in the motor drive, the flux is held constant, or merely
gradually changing, the prior art technique of vector control can
accommodate speed regulation. If, however, the speed is called to
accelerate rapidly, or conversely, to decelerate rapidly, the problem
arises of dynamically forcing the flux to match such circumstance. Since
there are two variable current components, the problem translates itself
into how to selectively exercise the compensating effect on those two
components so as to cause the resultant vector to match the speed
requirements. The major obstacle with such rapidly changing demand is to
prevent the current from exceeding acceptable limits. Therefore, the
question arises as to how the total current should be limited to a safe
maximum value. Imposing constant limits on both components would
unnecessarily restrict one component in magnitude whenever the demand for
the other is low.
SUMMARY OF THE INVENTION
The invention relates to an AC induction motor drive including first means
responsive to a flux demand for generating a first signal representative
of a direct component reference current; second means responsive to a
torque demand for generating a second signal representative of a
quadrature component reference current; third means responsive to a speed
demand for generating a third signal representative of a position angle
characterizing the current resultant vector of said direct and quadrature
components; and fourth means responsive to said first, second and third
signals for generating three coordinate phase currents for the motor
drive. According to the invention, means is provided within the first
means for dynamically responding to the flux demand and first limiting
means is provided in response to said dynamically responding means for
limiting the first signal in magnitude within a predetermined maximum
value (LIM). Second limiting means is provided responsive to the first
signal and operative upon the second means for limiting the second signal
so that the resultant vector remain within the value ALIM=(LIM.sup.2
-i.sub.d.sup.*2), where i.sub.d.sup.* is said first signal.
Preferably, a microcomputer is used to compute ALIM=.sqroot.(LIM.sup.2
-i.sub.d.sup.2).
According to another aspect of the invention, pole-tying current control
apparatus is provided with a voltage-source inverter generating the
three-phase currents of the motor under a bang-bang technique, the
operation of which is enhanced by control means operated cyclically upon
two of said poles while connecting one phase of the motor to one of the
voltage-source terminals through the third of the poles, the roles of said
two and third poles being sequentially permutated during such cyclic
operation.
BRIEF DESCRIPTION OF THE DRAWINGS
Preferred embodiments of the invention will now be described by way of
example only, with reference to the accompanying drawings in which:
FIG. 1 is a block diagram representation of an AC motor drive embodying the
speed/flux control system according to the present invention;
FIG. 2 shows in block diagram the concept of induction motor torque control
using an impressed stator current vector as in the prior art;
FIG. 3 illustrates a prior art closed loop speed regulator system requiring
constant flux operation;
FIG. 4 is a block diagram of the speed/torque control system embodying
forced flux level control according to the present invention;
FIG. 5 shows the drive control system which is a preferred implementation
in the speed/torque control system of FIG. 4;
FIG. 6 is a hardware representation of the speed regulator in the system of
FIG. 5;
FIG. 7 is a schematic representation of the control software used in the
speed/torque control system of FIG. 4;
FIGS. 8A-8D are the power, rotor flux, torque/speed and voltage pkph/speed
motor drive characteristics, respectively, for steady-state maximum power
with the speed/torque control system of the invention;
FIGS. 9A-9D are simulated characteristics for the torque, the current, the
speed and the voltage, respectively, for an induction motor drive under
the flux-dominant speed/torque control system of the invention, when
accelerating through base speed;
FIGS. 10A-10D are simulated characteristics for the torque, the current,
the speed and the voltage, respectively, for an induction motor drive
under the flux dominant speed/torque control system of the invention, when
decelerating through base speed;
FIG. 11 shows a bang-bang current-control scheme as can be applied for
current control in the speed/flux control system according to the
invention;
FIG. 12 is a pole-tying current control apparatus according to the
invention which can be applied to the speed/flux control system of FIG. 4;
FIG. 13 is a block diagram illustrating the detection of motor emf as can
be used in the current control system of FIG. 12;
FIG. 14 shows waveforms providing a comparison between the prior art
bang-bang control technique of FIG. 11 and the pole-tying scheme used in a
bang-bang control system as shown in FIG. 12; and
FIG. 15 is a hardware implementation of the current regulator of FIG. 12.
DESCRIPTION OF THE PREFERRED EMBODIMENTS
Referring to FIG. 1 a speed regulator system is shown embodying dynamic
control according to the invention for both the motor speed and its
magnetic flux level. As explained hereinafter, within the speed/flux
control system SFCS the torque demand is derived from a speed regulator
error signal, while the motor flux reference is obtained according to a
predefined function of the motor speed. Both the torque and flux
references are used as variables, as explained hereinafter.
In the illustrated voltage source inverter and variable frequency AC motor
drive of FIG. 1, the rectifier RCT includes a combination of a GTO device
and a thyristor TH7 to ensure the passing of regenerative energy from the
inverter side, while providing a zero-current intermediate stage in the
commutation process and maintaining maximum voltage on the DC-link
capacitor C between successive such zero-current stages. This two-quadrant
power conversion aspect of a voltage-source inverter motor drive has been
described in U.S. Pat. No. 4,697,131 of Colin D. Schauder. For the purpose
of the description of this aspect of FIG. 1, the Colin D. Schauder patent
is hereby incorporated by reference.
Thus, from the three-phase AC industrial lines A, B, C, a rectifier RCT
provides, with a capacitor C, a DC-link voltage between DC terminals TA,
TB. The voltage source so designed includes a reactor L (as generally
known) and the DC voltage is converted by an inverter INV into a
three-phase AC power supply (U, V, W) for the AC motor M (as generally
known). The rectifier includes a thyristor bridge (TH1-TH6) and two
serially-connected networks, one having a nodal point connected to the
negative terminal TB and situated between a GTO device and a device D2,
the latter across the RCT bridge, the second having a nodal point
connected to the positive terminal TA and situated between a diode D1 and
a thyristor TH7, the latter across the RCT bridge. This is as described in
the incorporated by reference patent.
A thyristor control circuit TCC responding to the AC phase voltages derived
on lines L1, L2, L3 and to a power flow direction control signal applied
on line L.sub.cc, is generating gating signals for thyristors TH1-TH7 and
the GTO device. Depending upon whether the motor drive is in the forward
mode, or in the regenerative mode, the signal of line L.sub.cc is a Zero,
or a ONE, as outputted by a function generator FG responding to the value
of the voltage V.sub.c exiting across terminals TA, TB, as derived
between lines LA and LB.
As explained hereinafter, the invention provides for a speed/flux control
system SFCS responding to a speed reference on line 1 and to the actual
speed of the motor as sensed between lines 2 and 3 from an encoder ("type
"H25/encoder BEI") coupled to the stator and the rotor of the motor. Two
phase representative currents (ia, ic) are generated on lines 75 and 76 by
the speed/flux control system SFCS and are used by a current control
system CCS to provide control signals on lines 10, 11, 12, in relation to
the currents sensed on lines 6 and 7 on phases V and W of the stator of
the motor, so as to control a voltage-source inverter INV. As explained
hereinafter, the bang-bang technique is used by the current control system
CCS, preferably with the improvement according to another aspect of the
present invention consisting in cyclically tying one of the three poles
PU, PV, PW of the inverter, while controlling the two others, via their
corresponding gating circuits (GC1, GC2, GC3). This aspect of the
invention will be explained hereinafter.
Referring to FIG. 2, a block diagram illustrates the concept of vector
control in an AC motor drive. The reference signal representing the flux
demand .psi.* is applied by line 13 to a transfer function TF1 [function
(1+St2/M)] providing on line 15 the direct component i.sub.d.sup.* of the
current. The reference signal representing the torque demand T* is applied
by line 14, first to a scaling circuit SC1 accounting for the constant
2L2/3M n (where L2 is the three-phase rotor self-inductance, M the
three-phase stator/rotor mutual inductance and n the number of pole pairs
on the machine), then, to a divider DV1 having the output of line 20 from
a scaler circuit SC1 (2L2/3Mn) applied as numerator and the flux demand
derived from line 13 applied, thereto by line 17, as the denominator. The
result is on line 16 the quadrature component i.sub.q of the current. From
this two coordinate system, the vector current control system VCCS
converts the system into a three current system (ia, ib, ic, on lines 30,
31, 32, respectively) by reference to the direct and quadrature components
resultant vector angle .theta. derived on line 25. Angle .theta. is
obtained as follows: A speed angular velocity .omega..sub..mu. is derived
on line 23 from the motor; from the quadrature component i.sub.q.sup.* of
lines 19 and 19, after scaling by M/T.sub.2 (where T.sub.2 =L.sub.2
/R.sub.2, with L.sub.2 being the rotor inductance and R.sub.2 the rotor
resistance per-phase) and, after adjustment by taking a correction with
the inverse of the flux demand .psi.* (via divider DV2), a speed demand is
derived on line 22. The latter is then added (at S1) to the actual speed
of line 23 so as to provide at the output a value which is integrated (by
1/s within integrator INT) to convert the speed into a position angle
.theta. on line 25. From the resultant vector defined by orthogonal
components i.sub.d.sup.* and i.sub.q.sup.* (on lines 15, 16) and from the
resultant vector angular position .theta. (on line 25), are obtained, as
generally known, the three-coordinate currents ia, ib, ic of lines 30, 31,
32.
The problem arises as to how the total current should be limited to a safe
maximum value. Imposing constant limits on both components would
unnecessarily restrict one component in magnitude whenever the demand for
the other is low. According to the present invention the resultant vector
current in the motor is limited by establishing the flux-demand by
priority over the torque-demand. Up to a limit value (.+-.LIM) flux
control is prevailing, thereby ensuring optimum use of the current
available from the power source. To this effect, the instantaneous direct
component of the current i.sub.d is estimated within its normal limit
values (.+-.LIM) and a limit (.+-.ALIM) is established for the quadrature
component of the current i.sub.q so as not to exceed the assigned i.sub.d
limit (LIM). The quadrature current component limit ALIM is calculated in
accordance with the formula:
ALIM=.sqroot.LIM.sup.2 -i.sub.d.sup.2
considering that the square of the resultant vector is equal to the sum of
the square of its components. As a result, no current is allocated to
torque production unless the motor flux has reached the value prescribed
for each speed.
The direct current i.sub.d could contain large noise components due to the
time derivative of a quantized flux demand derived from a look-up table,
for example. As a precautionary measure, according to the invention
low-pass filter action is provided introducing a time lag between the flux
demand and the actual flux in the machine, and also adapted speed
regulation is provided with increased loop gain or decreasing flux, and
conversely.
Referring to FIG. 3, a block diagram illustrates a speed/torque control
system for induction motor expressed in polar coordinates drive and
operating under constant flux demand. The reference speed W.sub.m.sup.*,
applied on line 33, is compared by subtractor S2 with the actual
mechanical speed W.sub.m derived on line 23, thereby obtaining on line 34
a speed error. The actual frequency W.sub.m of line 23 is obtained from
the motor torque (line 27) and the load torque (line 26) to generate an
error passed into a transfer function TF2 involving integration with the
motor inertia J, as generally known. The speed error of line 34 is passed
into a proportional (K1) and integral (K2/S) circuit (TF3) providing on
line 35 the torque demand T* necessary to reduce the speed error. In order
to ensure that the outputted torque demand is correctly limited and that
the integral within TF3 is similarly limited, imposed in relation to both
the input of line 34 and the output of line 35, as generally known. The
torque demand of line 35 is scaled by a circuit SC2 involving the formula
2R2/3n .psi.*.sup.2, where R2 is the rotor resistance, n is the number of
pole pairs, and .psi.* the constant flux demand in this instance. The
result is (on line 36) the slip frequency Ws. Two transfer functions TF4
and TF5 are used in response to line 36 in order to provide, under polar
coordinates, the current vector i on line 38 and the vector angle .theta.
on line 48, which are converted by a vector current control system VCCS
into currents: i.sub.a =i.sin (.theta.); i.sub.b =i sin (.theta.-2.pi./3)
and i.sub.c =i.sin (.theta.+2.pi./3) for the three phases U, V, W of the
motor supply. Transfer function block TF4 involves the function
(1+W.sub.s.sup.2 T.sub.2.sup.2).sup.1/2 .times.Ym, where T2 is the L2/T2
rotor time constant, with L2 being the rotor inductance, and M the mutual
inductance between rotor and stator. Transfer function block TF5 involves
the arc tangent TAN.sup.-1 .rarw.(.omega..sub.s T.sub.2). The angle
.theta. is obtainedly, first integrating the signal of lines 36 and 42 to
provide on line 43 a demand angle, and by integrating the signal of lines
23 and 40 (which is the actual W.sub.m signal) thereby obtaining by
integration .theta..sub.m. Then, a summer S3 combines the angles of lines
43 and 41 to provide on line 44 a corrective angle for the output (on line
47) from transfer function TF5, via summer S4. All this is generally
known, and is provided only to illustrate the prior art.
Referring to FIG. 4, a speed regulator system implementing controlled
flux-forcing according to the present invention is illustratively shown.
FIG. 4 shows, like in FIG. 3, the generation of a speed error (on line 34),
the generation of a speed signal (on line 23) obtained from a motor torque
signal (line 27) and a load torque signal (line 26), and the generation of
a vector angle .theta. (on line 44). However, instead of generating a
polar coordinate vector current i (as in line 38 of FIG. 3), direct and
quadrature current components i.sub.d.sup.* and i.sub.q.sup.* are
generated on lines 15 and 16, like in FIG. 2.
In principle, the torque produced by an induction motor and the magnetic
flux level in the machine can both be controlled dynamically by correctly
controlling the stator current vector and its instantaneous slip frequency
relative to the rotor. FIG. 2 shows the relationships which must be
preserved in order for the instantaneous torque demand T*, and the
instantaneous flux demand .psi.*, to be satisfied. In this diagram, the
components i.sub.d.sup.* and i.sub.q.sup.* represent the stator current
vector in a reference frame at angle .theta. relative to the stator, and
T.sub.2 is the rotor time constant.
When a motor drive application requires constant flux operation (i.e.
.psi.* constant) this control concept can readily be used to configure a
closed loop speed regulator system as shown in FIG. 3. In such a case all
gain terms are constant and the speed control bandwidth applied to TF3
stays constant at all operating speeds. The motor phase current can be
limited by simply placing fixed amplitude limits on the compensated speed
error signal, as shown in relation to block TF3.
However, for some applications it is necessary to control the motor flux
level dynamically. Under constant horsepower operation, for example, the
motor flux must be forced-down while accelerating and forced up again
while decelerating, in order to ensure that a prescribed flux level is
achieved at each speed. If this is not done, it may not be possible to
meet the operating specifications continuously under the available supply
voltage. FIG. 2 shows that when .psi.* varies with time, the direct
current component i.sub.d must contain terms proportional to both .psi.*
and its rate of change.
The speed signal of line 23, by line 50 and function generator FG1, is
converted into a flux demand representative signal, the function being
.psi.*/M for both directions of rotation. Function generator FG1 defines a
speed range for which the flux demand .psi.* is constant. Outside the
range on either side thereof, depending upon whether the speed is positive
(forward), or negative (reverse), the flux demand is forced down as the
speed exceeds what can be called the "base speed" of the motor drive, and
conversely is forced back to the constant flux level if the speed is
decreasing toward "base speed". The invention comes into play in these two
instances: by i on the one hand allowing the direct current component
i.sub.d.sup.* to provide the necessary flux within the assigned limits
.+-.LIM, and on the other hand by 2) controlling the quadrature component
i.sub.q.sup.* within limits and within the capability to exert torque
controls with the quadrature component after the priority has been given
to the direct component. The outputted signal of line 51 is only an ideal
value. The latter is converted into an actual value by a transfer function
TF6 taking into account the time lag and introducing a time constant Ta
through a formula 1/(1+sT.sub.a).sup.2. The flux demand .psi.* on line 52
is then passed into a proportional-plus-differential transfer function
(1+sT.sub.2) within block TF7. The outputted current signal (on line 53)
is only ideal and needs to be maintained within practical limits LIM as
shown by the limiter LMT1 of FIG. 4.
According to the present invention where i.sub.d and i.sub.q are variables,
each calling for a resultant current required to be held between
acceptable limits. It is now proposed to treat the direct component
i.sub.d independently as a variable to be held between own limits .+-.LIM
which match the limits imposed to the resultant vector, but independently
of the resultant current vector, thereby forcing the flux from line 15 to
conform the required values within the function of FG1 for the particular
speed.
Having established the instantaneous value of i.sub.d, the value of i.sub.d
is used to instantaneously and continuously calculate which limits can be
imposed to the quadrature component i.sub.q so that, while maximizing the
use of i.sub.d, the resultant current vector, nevertheless, will not
exceed its imposed limit. This is achieved from the consideration that the
sum of the squares of the direct and quadrature components of current is
equal to the square of the resultant vector. Accordingly, the limit to be
imposed to the quadrature current ALIM is such that ALIM=.sqroot.LIM.sup.2
-i.sub.d.sup.2. In other words, while i.sub.d is allowed to be maximized
within the limits .+-.LIM assigned to the resultant vector, the limit ALIM
can be imposed to the quardature component, thereby never to exceed the
limit LIM for the resultant current vector. As shown in FIG. 4, this is
achieved with function generator FG2 responsive to the signal of line 15,
and line 54, thereby providing (on line 56) the variable limit .+-.ALIM to
be applied to controller CTL applying by lines 57, 58 to the (P+I)
controller TF3 of line 63 to line 16, a window of variable width, in
contrast to the fixed amplitude limits applied to block TF3 of FIG. 3. The
resulting signal is on line 16 the quadrature component i.sub.q.sup.*
which, with the direct component i.sub.d of line 15, determines the
current control signals of lines 30, 31, 32. The resultant vector angle
.theta. is obtained on line 44. To this effect from line 16 is supplied
the numerator of a divider DV4 having the output of a function
1/(1+sT.sub.2) defined within a transfer function TF8 (responsive to line
15) as its denominator. An integrator (TF9) embodying a function 1/sT2 is
used to convert the signal of line 64 outputted by divider DV4, which is a
speed signal, into a position (on line 43) angle signal to be added by
summer S3 to the actual position angle derived from line 23, via line 40
and integrator INT, and obtained on line 41. Therefore, line 44 (at the
output of summer S3) is the frame angle .theta. enabling the conversion
(by the vector current control system VCCS) of i.sub.d.sup.* and
i.sub.q.sup.* into phase currents i.sub.a, i.sub.b, i.sub.c. In other
words, the direct component i.sub.d has been given priority over the
quadrature component i.sub.q, thereby establishing "flux dominance" in the
speed controller. This means that no current is allowed to torque
production (i.sub.q.sup.* of line 16) until the motor flux has reached the
prescribed value for each speed according to function FG1.
Should there be no load torque, there will be a self-limiting effect on
i.sub.d since reducing motor torque reduces the derivative term in
i.sub.d. In practice, i.sub.d should never reach its allowable limits, and
the drive will accelerate, or decelerate, at an optimum rate which is
compatible with the prescribed flux characteristic (in FG1), the inertia
(J), and the set current limit (LIM). When the drive is loaded, the
situation is similar except that i.sub.d, now, conceivably can be driven
into limit by the accelerating, or decelerating, action of a load torque.
This would be analogous to a loss of control and the drive being stalled
by excessive load torque, which would not occur if the load
characteristics had been correctly anticipated.
It is recognized that for certain applications it might be difficult to
implement the (1+sT.sub.2) transfer function TF7 required for flux
control. In such a case some low-pass filter action will be provided. A
filter transfer function H(s) will be chosen to represent an acceptable
time lag between the flux demand and the actual flux in the machine.
However, thanks to the invention, such time lag can be made much shorter
than the T.sub.2 time constant which would prevail if no flux forcing was
attempted. It will also be chosen to meet a tolerable noise level.
Typically where the rotor time constant is T.sub.2 =458 msec, (considering
TFG as two cascaded first order lag filters, in the instance of each
having a 10 msec time constant), H(s) can be chosen to be two real poles
at s=-100.
In the proposed control system, the actual motor flux level is calculated
from the demanded value of i.sub.d and the resulting signal controls the
gain of the speed regulator loop. Since the gain would othewise vary in
proportion to the actual flux, the action of such adaptive control is to
increase the loop gain for decreasing flux and vice versa. The regulator
is tuned to operate under full flux conditions below base speed. It
should, then, have constant small signal bandwidth at all speeds. Should
the increased gain associated with low flux increase the noise level
propagated from the speed feedback transducer, the path will be adequately
filtered.
Referring to FIG. 5, the drive control system according to the invention is
shown to use a computer MCP (in this instance an INTEL microcomputer 8031)
for performing many of the functions disclosed in analog form in FIG. 4.
Thus, the ALIM function of FG2 is performed by the computer and supplied
by line 56 to the controller CTL. The i.sub.q.sup.* quadrature component
of line 16 is supplied to the computer, the i.sub.d.sup.* draft component
being generated within the computer. The computer responds on lines 70 and
71 to two shaft encoder signals .theta.1 and .theta.2, which the computer
uses as generally known to provide the signal .theta. of line 44 in FIG.
4. The computer generates the actual speed signal on line 72 which,
depending upon the sense of rotation (forward, or reverse), by switch SWS,
and as controlled by the computer, will cause on line 23 the speed signal
to be applied to summer S2, like in FIG. 4. The computer also generates
the signals of lines 61 and 56 (like in FIG. 4). Accordingly, are
outputted current reference signals i.sub.a.sup.* and i.sub.c.sup.* on
lines 75, 76, respectively. The third phase current signal i.sub.b.sup.*
being obviously: -(i.sub.a.sup.* +i.sub.b.sup.*).
According to a second aspect of the invention, and as shown in FIG. 5, a
bang-bang current controller BBC is used, in response to the reference
signals of lines 75, 76, to apply by lines 10, 11, 12 (like in FIG. 1)
control signals for the gating circuits GC1, GC2, GC3 of the respective
inverter poles PU, PV and PW.
The bang-bang method of controlling an inverter is generally known. See for
instance A. Kernick, D. T. Stechschulte and D. W. Shireman/"Static
Inverter With Synchronous Output Waveform Synthesized by
Time-Optimal-Response Feedback" in IEEE Transactions IECI Vol. 24, No. 4,
November 1977, pages 297-305; also, "Time-Optimal Response Control of
Two-Pole Single-Phase Inverter/M. A. Geyer and A. Kernick/Power Cond.
Spec. Conf. JPL, Pasadenia, Calif., Apr. 19, 1971; and, "High Performance
Torque-Controlled Induction Motor Drives/C. A. Schauder, F. M. Choo, M. T.
Roberts in IEEE Trans. IA-19, No. 3, May-June 1983. The bang-bang
technique used here calls for a measurement of the motor currents where
shunts SHU, SHV for two of the phases (U, V) provide on lines 83 and 84
the sensed current signals. The motor currents on lines 100, 101, 102 are
passed into three transformers TNF1, TNF2, TNF3 providing in the primary
the difference between two consecutive currents. Thus TNF1 receives the
difference between i.sub.a of line 100 and i.sub.b of line 101.
Transformer TNF1 generates at its secondary the rate of change d(i.sub.a
-i.sub.b)dt, mainly on line 82. Similarly, TNF2 provides on line 81 the
rate of change d(i.sub.b -i.sub.c)dt, and TNF3 provides on line 80 the
rate of change d(i.sub.c -i.sub.a)dt. Circuit BBC generates on lines 10,
11, 12 (like in FIG. 1) control signals for the gating circuits (GC1, GC2,
GC3) of the three poles PU, PV, PW of the inverter, respectively.
How the bang-bang current controller BBC, according to the invention, has
been provided with enhanced capability will be explained hereinafter by
reference to FIGS. 12 and 15. The speed regulator according to the
invention will be first described in its best mode of operation by
reference to FIGS. 5, 6 and 7.
FIGS. 6A and 6B show the computer MCP of FIG. 5 connected by 8-bit data
lines (DL) which are interconnected through ports #1 (solid state device
U9); #0 (U8); #2 (U10); #5 (U17); #4 (U16) and #3 (U15), as seen from
right to left in the drawing. The microcomputer MCP is an INTEL 8031.
Chips U9, U8, U10, U17, U16 and U15 are of the 7524 type. Associated with
the computer are an address latch U2, its decoder U5, and a PROM memory
U3.
Ports #0 and #1 output on pins 16, and lines 75 and 76, the current
references i.sub.a.sup.* and i.sub.c.sup.*, respectively. Port #2 includes
an A/D converter outputting, on pin 4 and line 16, the quadrature
component signal i.sub.q.sup.*. At port #5 are received the 8 bits from
the data lines DL relating to line 56 of FIG. 5 (pins 4 to 11 of V17).
Chip U17 involves the function TF3. To it are associated amplifiers A1 and
A2 which concur in accomplishing the speed regulator gain adjustment and
the function of controller CTL.
Port #4 receives the 8-bit lines 61 from the data lines DL (pins 4 to 11 of
U16). The functions of summer S2 and divider DV3 are performed by chip
U16, in relation to the speed reference of line 33 (received on pin 16)
and the actual speed of line 23 (on pin 1).
At port #3, the data lines DL (at 72) provides (pin 4-11) the speed
modulus, and the speed polarity is accounted for by lines 121, 122 from
the computer (polarity determination at A3) in accordance with pin 10 of
solid state device U1 (MCP). The speed feedback signal appears on line 23.
FIG. 7 is a software rendition of FIG. 4 where the lead lines and blocks
are matching with their numeral references the corresponding ones of FIG.
4. In addition have been added in block diagram the software
implementation of function FG1 with 1) a look-up table LKT1 outputting the
flux demand required for the speed of line 72; and 2) symbolic
representations of functions TF6 and TF7, leading to the limiter LMT1,
then, to the ALIM calculator, the latter using another look-up table LKT2.
The derivation of W.sub.m with the encoded values .theta.1 and .theta.2 is
illustrated with counters CNT0 and CNT1.
More generally, FIG. 7 is readily understood in the light of FIG. 4, for
the derivation of i.sub.d.sup.* on line 15, of i.sub.q.sup.* on line 16,
of .theta. on line 44, and of i.sub.a.sup.* and i.sub.c.sup.* on lines 75,
76 at the output of the vector current control system VCCS.
In FIG. 7, T represents the sample time (350 .mu.s), T.sub.a is the filter
time constant (10 .mu.s), K is the number of encoder counts per electrical
radian (256/2.pi.), and Q is a preset scaling constant which is the ratio
between the maximum current (LIM) and the steady-state value of i.sub.d
below base speed. The on-board counters of the 8031 computer simplify the
derivation of the shaft position, the speed and the direction of rotation.
The speed measurement is essentially done by pulse counting to give an
8-bit representation of the top speed modulus value.
Using a 12-MHz crystal, the 8031 computer performs the algorithm in 350
.mu.s. This time is acceptable although a shorter execution time can be
sought. At 240 Hz, which is the envisaged top or brake speed, the system
outputs about 12 samples per cycle. Typically, the computer software has
been written in ASM51 Assembler language, a way which is efficient in
terms of execution time. The program listing is included hereinafter in
Appendix A.
FIGS. 8A, 8B, 8C, 8D show curves at steady state (base speed) giving the
maximum power characteristics for power (FIG. 8A), rotor flux (FIG. 8B),
torque/speed (FIG. 8C) and voltage/speed (FIG. 8D) in an AC motor drive
not using the invention.
FIGS. 9A-9D show the effect of flux-dominant speed/torque control according
to the invention when accelerating through base speed. FIG. 9A gives the
torque, FIG. 9B the current, FIG. 9C the speed and FIG. 9D the voltage.
FIGS. 10A-10D are like FIGS. 9A-9D when decelerating through base speed.
FIG. 11 shows a conventional bang-bang current-control system. Between the
DC-link line terminals TA, TB and the three phase line inputs U, V, W of
the motor is connected an inverter INV with its three poles PU, PV, PW.
PWM modulation is obtained by controlling the inverter switches per pole.
The inverter switch operation in relation to terminals TA, TB is
symbolized by switches SW1, SW2, SW3 for the respective poles, controlled
at a certain high frequency in accordance with the control signals of
deadband comparators DB1, DB2, DB3 (for the respective switches and
poles). The phase currents are sensed from lines 190, 191, 192, and the
reference current signals i.sub.as.sup.*, i.sub.bs.sup.*, i.sub.cs.sup.*
are obtained on lines 90, 91, 92, respectively. The errors are derived on
lines 93, 94, 95 from subtractors S10, S11, S12, respectively, and applied
to DB1, DB2, DB3, respectively.
Referring to FIG. 12, the same technique is shown on the upper part of the
Figure. It will be explained now how the conventional bang-bang approach
is to be improved and why.
In vector control systems for high performance ac drive control algorithms
are used which define the required values of the motor phase currents at
all times. Controlling the motor currents to the reference values with
specified accuracy and bandwidth is a significant problem in the
implementation of such systems.
The constant voltage DC-line inverter is the preferred static power
converter for AC servo drives. The inverter is capable of producing six
different non-zero output voltage vectors and two null vectors (the latter
when the output lines are shorted). There are a number of schemes for
controlling the output current of this type of inverter. One of these
which is simple and very effective is the bang-bang system shown in FIG.
11. This controller acts in a non-linear way to keep the motor phase
current errors within an acceptable deadband around the target values. The
rate at which the inverter poles switch is not constant, but depends on DC
link voltage, back-emf of the motor, motor impedance and the size of the
error deadband. From a control point of view, this system produces optimum
response, but it suffers from a tendency to switch at excessively high
frequency on all inverter poles when the motor back-emf is low. This
effect may become so severe that in many cases it could cause failure of
the inverter hardware.
There are methods of switching this type of inverter on an open-loop basis
so as to generate desirable output voltage waveforms. The invention stems
from the observation that very high switching frequencies are not
"necessary" in order to produce acceptable currents in motors under low
back-emf. To this effect, an overriding current control scheme is now
proposed which does not suffer from the tendency to switch at high
frequency, while still retaining the rapid response quality of the
conventional bang-bang method.
The proposed scheme takes advantage of the fact that the motor has only two
independent current variables to be controlled via two independent
line-to-line voltages. At any time, therefore, two poles of the inverter
can be used to actively control the motor current, while the third pole
does not switch and simply establishes a reference potential on its
associated motor terminal. The selection of such "third pole" is cyclic
and occurs by permutation, it being determin | | |
