 |
Description  |
|
|
CROSS-REFERENCE TO RELATED APPLICATIONS
Not applicable.
STATEMENT REGARDING FEDERALLY SPONSORED RESEARCH OR DEVELOPMENT
Not applicable.
BACKGROUND OF THE INVENTION
The present invention relates to motor drives for variable speed control of AC induction motors and more particularly to a method and apparatus for smoothly starting an AC motor drive into a rotating motor.
Induction Motors Induction motors have broad application in industry, particularly when large horsepower is needed. A typical induction motor includes a stator and a rotor. The stator includes a three phase stator winding which forms a
cylindrical stator cavity. A common rotor design includes a "squirrel cage winding" in which axial conductive bars are connected at either end by shorting rings to form a generally cylindrical structure. The rotor is concentrically mounted for rotation
within the stator cavity.
The rotor is forced to rotate within the stator cavity by providing three phase electrical voltages to the stator windings. The stator voltages generate stator currents which in turn cause a rotating magnetic stator field. The stator field
interacts with the rotor to cause rotation.
As with movement of any object, rotor movement requires both (1) interaction between the stator field and the rotor and (2) a force applied to the rotor. Without enough interaction, even an extremely large force cannot move a rotor. Similarly,
without enough force, even mechanical contact could not move a rotor.
The interaction required for rotor movement is provided as follows. As the stator field rotates about the cavity, stator field flux lines cut across rotor bars. If the stator field rotates at a speed which is slightly greater than a rotor speed
so that each bar is subjected to a slowly varying stator field, the stator flux induces rotor bar currents (hence the term "induction" motor). The term "slip" will be used hereinafter to refer to the difference between the stator field and rotor
frequencies.
The rotor currents cause a magnetic rotor flux field. Rotor field strength is related to the number of stator flux lines cut by the rotor bars and therefore the amounts of rotor and stator flux are both related to stator field strength. The
stator and rotor fields attract and hence provide the interaction required for rotor movement. Thus, the stator field has an "attraction" component which provides the interaction required for rotor movement. Because the stator field attraction
component causes flux which interacts with the rotor, the attraction component is referred to hereinafter as the "flux producing" component.
The force required for rotor movement is provided as follows. As the stator field rotates about the cavity, if stator and rotor field attraction is sufficient, the stator field "pulls" the rotor along thereby causing rotation. Thus, in addition
to the flux producing component, the stator field has a "shear force" component pushing tangent to the rotor surface tending to rotate the rotor. Because the shear force causes a rotating torque on the rotor, the shear force is referred to hereinafter
as the "torque producing" component.
Field Oriented Control
The stator field corresponds to a stator magnetomotive force (hereinafter "mmf"). Referring to FIG. 1, a rotating phasor 1 representing an exemplary stator mmf generally has some angle .alpha. with respect to a rotor flux represented by phasor
2. Torque is proportional to the magnitudes of phasors 1 and 2 but also is a function of angle .alpha.. On one hand, maximum torque (i.e. maximum shear force) and essentially zero rotor flux (i.e. zero attraction) is produced when phasors 1 and 2 are
at right angles to each other (e.g., .alpha.=90 degrees). On the other hand essentially zero torque and excessive rotor flux is produced if phasors 1 and 2 are aligned (e.g., .alpha.=0 degrees). Mmf phasor 1 may therefore be usefully decomposed into a
torque producing component 3 perpendicular to rotor flux phasor 2 (and corresponding to the stator field torque producing component described above) and a flux producing component 4 parallel to rotor flux phasor 2 (and corresponding to the stator field
flux producing component described above).
Mmf components 3 and 4 are proportional, respectively, to two stator currents i.sub.qe, a torque producing current, and i.sub.de, a flux producing current, which may be represented by orthogonal vectors in a two dimensional d-q rotating stator
flux frame of reference (synchronous frame of reference) having slowly varying magnitudes.
Accordingly, in induction motor control it is generally desired to control not only applied driving voltage frequency (hence the speed of the rotation of the stator mmf phasor 1) but also applied voltage phase relative to current flow and hence
the division of stator winding currents into i.sub.qe and i.sub.de components. Control strategies that attempt to independently control q and d-axis currents i.sub.qe and i.sub.de (and hence the balance between attraction and shear force) are generally
termed field oriented control strategies ("FOC").
To accomplish FOC the industry has developed field oriented controllers. A typical PWM controller receives a command rotor frequency signal indicating a desired rotor frequency and includes a plurality of feedback loops to provide current and/or
voltage feedback signals for control purposes. The processor uses the feedback signals and the command signal to adjust q and d-axis currents i.sub.qe and i.sub.de thereby causing the rotor to rotate at the commanded rotor frequency.
As indicated above, rotor flux can only be generated if slip is relatively minimal. This is because currents will only be induced in a rotor bar if the bar is subjected to varying stator flux over at least a short threshold time period. If the
stator field is rotating at an extremely high frequency relative to the rotor, the stator field cannot cause rotor flux. Similarly, if the stator field is rotating at a much lower frequency than the rotor very little if any rotor flux will be generated. In either case there is essentially no attraction between the stator field and rotor and hence the stator field cannot be used to control rotor rotation.
For this reason, for example, even where a commanded rotor frequency is 30 Hz, if the instantaneous rotor frequency is much different than 30 Hz, say 28 Hz, in order to increase the rotor frequency to 30 Hz, first the stator field must be
provided at approximately 28 Hz to "gain control" of the rotor. Thereafter, once flux (i.e. attraction) between the stator field and rotor is established, the stator field frequency can be increased thereby increasing the rotor frequency.
Initially, when a rotor is stationary rotor frequency is known to be zero. In this case it is relatively easy to start a drive into the motor as the initial stator field frequency will be just above zero. As rotor frequency increases toward the
commanded frequency, stator field frequency can be controlled to maintain an acceptable slip using standard field oriented control strategies. Thus, FOC strategies can be applied to precisely control rotor frequency during normal operation.
Unfortunately, under certain operating conditions FOC strategies can cause unintended and disruptive motor operation. One important control area wherein FOC strategies have not been extremely successful has been when a motor drive has to be
started into an already rotating motor.
In many industries motors are not driven continuously but rather are turned on and off sporadically. For example, in the wood cutting industry motor power is often disconnected during "power down" periods between cutting jobs to save energy and
minimize tool wear. After power is cut off, rotor speed decreases and eventually, after some deceleration period, the rotor stops. Ideally deceleration periods are short. However, in reality, often deceleration periods are longer than a desired power
down period and the motor driver has to be started into a rotating motor.
As indicated above, to start a drive into a rotating motor stator field frequency has to be essentially identical to the instantaneous rotor frequency. or this reason many motor control systems include hardware such as a tachometer or an encoder
for tracking rotor speed. However, many other systems are not equipped with such speed sensing hardware. In any event, such speed sensing hardware increases system costs and therefore should be avoided if possible.
One solution for determining rotor speed prior to starting a driver into a rotating motor is to use residual stator flux to determine rotor speed. In general, when motor power is disconnected and during a first stage of the deceleration period,
the motor rotor operates like a generator inducing a stator flux which varies along with the instantaneous motor frequency. Thus, in the example above, if motor frequency is instantaneously at 29 Hz, the stator flux is also at approximately 29 Hz. The
flux decays to essentially zero flux as a function of a rotor time constant. Until the flux decays to zero, the flux value can be used by an AC controller to identify instantaneous motor frequency. Unfortunately, after the flux decays to zero, motor
frequency cannot be identified by sensing a residual signal.
A second solution for starting a drive into a rotating motor is described in U.S. Pat. No. 4,958,117 entitled "Frequency Control Based On Sensing Voltage Fed To An Induction Motor" which issued to Kerkman et al. on Sep. 18, 1990. That patent
employs a voltage controlled current regulator to start a drive into a rotating motor. That patent reflects the realization that when a voltage controlled current regulator is used to start a drive into a rotating motor where stator field and rotor
speed are disparate, the voltage regulator causes excessive command currents. Thus, the command currents can be used to identify instances wherein there is an appreciable difference between the stator field and rotor frequencies.
When excessive command currents occur, a voltage error (i.e. difference between commanded and actual stator winding voltages) is used to modify a command stator field frequency. Once the stator field and rotor speeds are essentially identical
the command currents are no longer excessive and volts/hertz control can be used to accelerate the rotor to the commanded frequency. Unfortunately, this scheme does not address FOC.
In general reconnect schemes may be affected by a number of shortcomings. First, as a stator field frequency approaches the rotor frequency a number of different adverse conditions can occur which cause the rotor to jolt or may cause a safety
device to trip and cut motor power entirely until the device is reset. For example, one condition is an over current fault wherein, when a voltage is initially provided to a stator winding, the winding resistance is minimal and often the provided
voltage is not in phase with any winding potential. In this case a large winding surge current results which can exceed safe levels. Many control systems are equipped with safety devices which cut off motor power when surge currents are sensed to
protect system components. Thus, over current faults can cause unintended power cut off resulting in inefficient system operation.
Another condition is referred to as a bus overvoltage fault which occurs with non-regenerative drives (i.e. drives which do not return energy back to a utility supply). A non-regenerative drive typically includes a diode bridge rectifier linked
between three utility supply lines and the positive and negative DC rails of a DC bus for converting three AC utility voltages to a DC bus voltage on the rails. A DC bus capacitor is usually provided across the DC rails to help smooth out DC voltage
ripple. A PWM inverter then links the DC bus to three motor terminals for providing three phase AC voltage to the terminals for driving the motor.
In a non-regenerating drive power can only flow from the utility to the DC bus and cannot flow in the opposite direction from the bus to the utility. Power also flows from the DC bus to the motor during motoring and, can flow from the motor back
to the DC bus during braking (i.e. dynamic slowing of the motor by providing a stator field having a frequency slightly lower than an instantaneous rotor frequency). In a non-regenerative drive if the stator field frequency approaches the rotor speed
from below the rotor speed during a frequency search, the motor will operate like a generator just prior to the stator field and rotor frequencies being equal thereby returning energy back to the DC bus and causing excessive DC bus capacitor voltage.
Energy returned to the bus can cause voltage across the DC bus capacitor to exceed a safe level and may again trip a safety mechanism cutting off motor power.
Yet another condition referred to as pull out torque can result. During frequency searching the stator field has very little effect on rotor motion as relative motion between the rotor bars and stator flux is too fast to generate rotor currents
and flux. However, as the stator field frequency approaches rotor frequency, because field orientation has not yet been achieved, an appreciable amount of the stator mmf causes rotor torque which jolts the rotor instantaneously. This pull out torque
has been known to "roll" the rotor thereby hindering frequency identification.
Other conditions include "plugging" and "dynamic braking", terms of art which are well known in the industry and therefore will not be explained here in the interest of simplifying this explanation.
Second, many of the adverse conditions described above are exacerbated by the relatively high search current (i.e. an excessive current level has to be reached prior to frequency modification) employed by this second solution. For example, pull
out torque magnitude is a function of the search current magnitude and therefore this second solution causes excessive pull out torque. In addition, excessive current causes higher bus overvoltages and can cause greater plugging and braking forces.
Third, whenever any rotor disturbance occurs during frequency searching the processes of ultimately determining rotor frequency is extended. This is because any disturbance (e.g. a jolt, rolling, etc.) can result in oscillation about the steady
state component of rotor frequency and ultimate rotor speed can only be determined after the transient oscillation deteriorates to an essentially zero value. As discussed above there are many sources of disturbances using prior art rotor frequency
searching methods.
Thus, it would be advantageous to have a method and/or apparatus which could accurately and quickly determine rotor speed for the purpose of starting a drive into an already rotating motor.
BRIEF SUMMARY OF THE INVENTION
The present invention includes a method for smoothly starting a driver into an already rotating motor. To this end, using a FOC controller, it has been recognized that if the current used to search for a rotor frequency is limited to a d-axis
flux producing current component (i.e. the q-axis current is held at a zero value), frequency searching is much smoother and the time required to complete a search is appreciably reduced. By maintaining a zero q-axis torque producing current during a
frequency search rotor disturbances are either eliminated or minimized and both current surges and bus overvoltages are minimized. Because disturbances are minimized, less rotor oscillation occurs as the stator field frequency nears rotor frequency
resulting in a faster overall speed determination.
In addition, because current surges and overvoltages are minimized it is less likely that a safety device will cut motor power for these reasons resulting in more efficient overall motor operation. After rotor frequency is determined both d and
q-axis current components are provided according to conventional FOC strategies to increase rotor frequency to a commanded frequency.
Initially, at the instant when the d-axis stator current is first field oriented with respect to the rotor so that the d-axis current is being provided at essentially the same frequency as the instantaneous rotor frequency, the d-axis current
generates only q-axis flux. However, when the d-axis current is not field oriented with respect to the rotor so that the d-axis current is being provided at a frequency which is appreciably different than the rotor frequency, there will be a q-axis flux
error. Thus, the q-axis flux error can be used to identify when the stator field frequency is essentially identical to the rotor frequency.
The q-axis flux error can be measured by monitoring a d-axis feedback voltage at the motor terminals and comparing the feedback voltage to a commanded d-axis voltage signal. Where the feedback and commanded d-axis voltage signals are different,
a q-axis flux error exists and the stator field frequency is different than the rotor frequency. In this case the stator field frequency (i.e. the d-axis current frequency) must be altered.
The stator field frequency is dynamically altered until the error between the feedback and command d-axis voltages is essentially eliminated (i.e. the q-axis flux error is eliminated) at which point the instantaneous d-axis current frequency is
essentially identical to the instantaneous rotor frequency.
After the d-axis stator current frequency and rotor frequency are essentially equal, in one embodiment of the invention motor control is not initiated until after d-axis flux builds up. Because there is no q-axis current, d-axis flux is directly
related to q-axis voltage. Therefore, d-axis flux can be monitored by monitoring a q-axis feedback voltage and comparing the q-axis feedback voltage to a voltage level known to occur when d-axis flux is at an acceptable level to start motor operation.
At that point, the FOC controller is allowed to control the motor providing both d and q-axis currents at essentially the rotor frequency to take control of the rotor. The controller eventually increases the stator field frequency thereby
increasing the rotor frequency up to the commanded frequency.
In a second embodiment of the invention after the d-axis stator current frequency and rotor frequency are essentially equal both the d and q-axis current components are provided according to conventional FOC strategies to increase rotor frequency
to the commanded frequency. It has been recognized that, while this method may be slightly less smooth than the first method (i.e. where flux is first built up prior to FOC), in some applications this method would be advantageous as a faster start time
would occur. This is because, as d-axis flux is built up torque can simultaneously be applied to the motor thereby increasing motor speed.
Accordingly, one object of the invention is to provide a method by which a motor controller or driver can be started into an already rotating motor with minimal disturbance. To this end, by providing only d-axis flux generating current, as the
stator field frequency approaches the instantaneous rotor frequency, only very little current causes a pull out torque. In reality, some pull out torque still occurs because, during searching the d-axis current is not field oriented with respect to the
rotor. For this reason, some of the d-axis current actually acts like a q-axis torque producing current causing some, albeit a minimal amount, of pull out torque as the stator field frequency approaches the rotor frequency.
Another object of the invention is to achieve the aforementioned object with existing controller hardware. Although separate hardware could be provided to implement the present invention, the inventive method is preferably implemented in
software run by existing controller hardware and therefore is relatively inexpensive to implement.
Yet one other object of the invention is to identify rotor frequency quickly during a searching procedure. To this end, because only d-axis current is used during the search, current related disturbances are eliminated or their effects are
substantially reduced thereby enabling quick frequency identification. In other words, because rotor disturbances are reduced rotor frequency does not oscillate about its steady state frequency component appreciably when the stator field frequency nears
the rotor frequency and the steady state rotor frequency can be determined more quickly.
In addition, in this regard, because q-axis flux typically is a much smaller value than d-axis flux, q-axis flux error is eliminated when a d-axis current is provided much faster than a d-axis flux error would be eliminated if a q-axis current
was provided. Thus, by using the q-axis flux error instead of a d-axis flux error to determine rotor frequency, the frequency identifying procedure period is minimized.
In a preferred embodiment, during the first stage of the deceleration period after motor power is initially cut off the q-axis feedback voltage which is indicative of residual stator winding flux caused by rotor inductance is monitored. The
stator flux, and hence feedback voltage, is at the rotor frequency and therefore can be used to determine the rotor frequency if the stator flux is at least a threshold level. Only after the stator flux falls below the threshold level is the d-axis
current provided for frequency searching.
This residual flux tracking feature also speeds up frequency determination as, in some cases where a motor is re-powered shortly after motor power is turned off, rotor speed will be known immediately and the driver can be started into the motor
instantaneously.
Also, preferably, when the d-axis current is provided for rotor frequency searching, a relatively low search current is provided. To this end, the search current may be limited to a rated d-axis current or, more preferably, the search current
may be limited to a much lower current level (e.g. 1/5th the rated current level). By limiting the search current to a low value the effects of pull out torque and other current related disturbances which are common during frequency searching is
minimized even further. For example, referring to FIG. 2, a graph illustrating torque as a function of slip for four different d-axis search current values i.sub.de1 i.sub.de2, i.sub.de3 and i.sub.de4 is illustrated. Current ide1 is greater than
i.sub.de2, i.sub.de2 is greater than i.sub.de3 and i.sub.de3 is greater than i.sub.de4. Clearly the pull out torque generated by larger search currents is greater than the pull out torque generated by smaller search currents and therefore smaller search
currents are advantageous.
These and other objects, advantages and aspects of the invention will become apparent from the following description. In the description, reference is made to the accompanying drawings which form a part hereof, and in which there is shown a
preferred embodiment of the invention. Such embodiment does not necessarily represent the full scope of the invention and reference is made therefor, to the claims herein for interpreting the scope of the invention.
BRIEF DESCRIPTION OF THE
SEVERAL VIEWS OF THE DRAWINGS
FIG. 1 is a schematic diagram illustrating an exemplary relationship between a stator magnetomotive (mmf) force phasor, a rotor flux phasor and d and q-axis current components of the mmf phasor;
FIG. 2 is a graph illustrating torque as a function of slip for various search current levels which serves to illustrate pull-out torque;
FIG. 3 is a schematic view of a motor controller according to the present invention;
FIG. 4 is a schematic view of the motor model of FIG. 3;
FIG. 5 is a schematic view of the flux regulator of FIG. 3;
FIG. 6 is a schematic view of the frequency generator of FIG. 3;
FIG. 7 is a flow chart illustrating a preferred method according to the present invention;
FIG. 8 is a graph illustrating stator field search frequency and actual rotor velocity during a frequency search procedure according to the present invention;
FIG. 9 is a graph illustrating q-axis feedback voltage stator field frequency during a frequency search according to the present invention;
FIG. 10 is a graph illustrating flux error and stator field frequency during a frequency search according to the present invention;
FIG. 11 is a graph illustrating d-axis feedback voltage and stator field frequency during a frequency search according to the present invention; and
FIG. 12 is a schematic of the speed regulator of FIG. 3.
DETAILED DESCRIPTION OF THE INVENTION
In the description which follows, unless indicated otherwise, an "e" subscript denotes that a signal is referenced to the synchronous frame of reference, an "s" subscript denote that a signal is referenced to the stationary frame of reference, an
"f" subscript denotes a feedback signal, an "*" superscript indicates that a signal is a command signal, an "r" subscript denotes that a signal is a rotor signal, a "d" subscript denotes that a signal is referenced to the d-axis in a two dimensional d-q
reference frame, a "q" subscript denotes that a signal is referenced to the q-axis in a two dimensional d-q reference frame, and "a", "b" and "c" subscripts indicate that corresponding signals are referenced to "a", "b" and "c" supply lines.
While the following description details various blocks, steps, and functions, it should be remembered that all of these elements are meant to be implemented in software as computer programs and represent algorithms for execution by conventional
type digital processor adapted for industrial applications, such as a model 8096 Microelectronic processor as supplied by Intel Corporation of Santa Clara, Calif.
A. Theory
Some basic FOC understanding and system equations are helpful in understanding how the present invention operates to identify rotor frequency. As an initial matter, usually a FOC controller receives a signal indicating command rotor frequency
.omega..sub.r * which is a desired rotor frequency for a controlled motor. For example, the desired frequency may be 30 Hz. The controller uses the command frequency .omega..sub.r * and a plurality of feedback signals to generate voltages and current
to drive the motor at the commanded frequency. To drive the motor rotor, the controller provides a rotating stator field having a stator field frequency .omega..sub.e which is slightly faster than commanded frequency .omega..sub.r * (i.e. the difference
being a slip frequency .omega..sub.s). The difference in stator field and rotor frequencies induce rotor bar currents and hence rotor flux required for rotor movement.
To generate the stator field the controller provides both a d-axis command current i.sub.de * for generating "interacting" flux and a q-axis command current i.sub.qe * for generating a torque. As well known in the FOC art, production of any
given set of q and d-axis command currents i.sub.qe * and i.sub.de * requires command voltages V.sub.qe * and V.sub.de * as follows:
where:
rs=stator resistance;
.lambda..sub.de =d axis flux linkage; and
.lambda..sub.qe =q-axis flux linkage;
and where: ##EQU1## Fluxes .lambda..sub.de and .lambda..sub.qe can be represented as:
where:
L.sub..sigma. =transient motor inductance; and
L.sub.m =mutual winding inductance.
To ensure that commanded voltages V.sub.de * and V.sub.qe * are actually provided at motor terminals, most FOC controllers include at least one feedback loop providing d and q-axis feedback voltage signals V.sub.def and V.sub.qef for comparison
to commanded voltages V.sub.de * and V.sub.qe * Any error between voltages V.sub.qe * and can cause the controller to modify commanded current i.sub.de *.
Referring to Equation 2, according to the present invention, when only a d-axis current is commanded the resulting d-axis voltage is V.sub.de *=r.sub.s i.sub.de *. When no q-axis current i.sub.qe * is commanded, no q-axis stator flux component
.lambda..sub.qe is produced when the stator and rotor frequencies are equal. However, when frequency .omega..sub.e is not equal to the rotor frequency, a q-axis flux error occurs which is reflected in an actual d-axis voltage.
A feedback d-axis voltage equation similar to Equation 2 can be written as:
where V.sub.def is a feedback d-axis voltage indicating actual d-axis voltage. Equations 2 and 5 can be combined to yield: ##EQU2## where .lambda..sub.qe is a q-axis flux error. Hence, the difference V.sub.de *-V.sub.def can be used to identify
flux error .lambda..sub.qe and thereby recognize a difference between stator field and rotor frequency. Similarly, an equation for d-axis flux error .lambda..sub.de can be derived as: ##EQU3##
Once disparate stator field and rotor frequencies have been identified via flux error .lambda..sub.qe, the d-axis command and feedback voltage signals V.sub.de *, V.sub.def, respectively, are used by the controller to adjust stator field
frequency either up or down depending upon whether or not the stator field frequency is above or below the rotor frequency.
As well known in the controls art, d-axis feedback voltage V.sub.def is generated by first sensing three phase stator terminal voltages V.sub.as, V.sub.bs and V.sub.cs, converting voltages V.sub.as, V.sub.bs and V.sub.cs to two-phase stationary d
and q-axis voltages V.sub.dsf, V.sub.qsf, respectively, and then converting voltages V.sub.dsf and V.sub.qsf to synchronous d and q-axis voltages V.sub.def, V.sub.qef as a function of the instantaneous stator winding frequency .omega..sub.e.
Specifically, frequency .omega..sub.e is converted to a phase angle .theta..sub.e used to identify voltages V.sub.qef, V.sub.def according to the following equation: ##EQU4## On one hand, when stator field frequency is less than the instantaneous rotor
frequency, angle .theta..sub.e will be at a value which causes feedback voltage V.sub.def to be less than commanded d-axis voltage V.sub.de *. Thus, the difference V.sub.de *-V.sub.def can be used to increase stator field frequencies. On the other hand,
when stator field frequency is greater than the instantaneous rotor frequency, angle .theta..sub.e will be at a value which causes feedback voltage V.sub.def to be greater than commanded d-axis voltage V.sub.de *. Difference V.sub.de *-V.sub.def in this
case is used to decrease stator field frequency.
B. Hardware
Referring now to FIG. 3, the present invention will be described in the context of an exemplary current regulated pulse width modulated motor controller 10 for an induction motor 9. Generally speaking, induction motor 9 can be characterized by a
number of different motor variables including a stator resistance r.sub.s and a transient inductance L.sub..sigma.. Controller 10 is provided with a stator resistance value r.sub.s and a transient inductance signal L.sub..sigma. which correspond to
motor 9. Generally, controller 10 receives a command rotor frequency signal .omega..sub.r * indicating a desired motor rotor speed and generates three phase voltage signals V.sub.as, V.sub.bs and V.sub.cs at three stator terminals which cause three
stator currents (not illustrated) which together are intended to drive motor 9 at the commanded rotor frequency .omega..sub.r *.
To this end, controller 10 comprises a plurality of components including, among other things, a three-phase AC source 11, a power converter 12, a DC bus 13, a pulse width modulated (PWM) inverter 14, a two-to-three phase transformer 15, a current
regulator 16, a three-to-two phase transformer 17, a second three-to-two phase transformer 18, a stationary to synchronous transformer 19, a synchronous-to-stationary transformer 21, a flux regulator 22, a speed regulator 24, a frequency generator 26, a
motor model 28 and an integrator 29.
A power conversion section of controller 10 includes source 11, converter 12 and DC bus 13. AC power source 11 provides three-phase AC power at a line frequency of 60 Hz. The three phases of source 11 are connected an AC-DC power converter 12
which rectifies the alternating current from source 11 to produce DC voltage V.sub.dc of constant magnitude on bus 13 which connects to power inputs on inverter 14.
Controller 10 includes two separate feedback loops for providing feedback voltage and current signals. The first feedback loop, a current loop, includes three current sensors (e.g. Hall effect sensors, not illustrated), a separate current sensor
attached to each of three motor terminals for providing current feedback signals i.sub.af, i.sub.bf and i.sub.cf to transformer 17. Transformer 17 converts the three-phase current feedback signals i.sub.af, i.sub.bf and i.sub.cf to two-phase signals
referenced to a stationary d-q frame of reference in a manner well know in the art providing d and q-axis feedback current signals i.sub.dsf and i.sub.qsf, respectively.
The second feedback loop, a voltage loop, includes three voltage sensors (not illustrated), a separate voltage sensor provided at each of three motor terminals to provide voltage feedback signals V.sub.af, V.sub.bf and V.sub.cf to 3-to-2 phase
transformer 18. Transformer 18, like transformer 17 converts its three-phase inputs to two-phase d and q-axis stationary voltage signals V.sub.dsf and V.sub.qsf, respectively. Feedback voltages V.sub.dsf and V.sub.qsf are provided to
stationary-to-synchronous transformer 19. In addition, transformer 19 receives phase angle signal .theta..sub.e which is related to the stator field frequency. Transformer 19 uses its received signals to transfer stationary signals V.sub.qsf and
V.sub.dsf to synchronous signals V.sub.qef and V.sub.def according to Equation 9 above. Signals V.sub.qef and V.sub.def are provided to other controller 10 components as described in more detail below.
Inverter 14 includes a group of switching elements which are turned on and off to convert DC voltage on bus 13 to pulses of constant magnitude. The pulses formed by inverter 14 are characterized by a first set of positive going pulses of
constant magnitude but of varying pulse width followed by a second set of negative going pulses of constant magnitude and varying pulse width. The RMS value of this pulse train pattern approximates one cycle of a sinusoidal AC waveform. The pattern is
repeated to generate additional cycles of the AC waveform.
To control the frequency and magnitude of the resultant AC power signals to motor 9, AC inverter control signals are applied to inverter 14. Inverter 14 receives three balanced AC inverter control signals V.sub.as *, V.sub.bs * and V.sub.cs *
which vary in phase by 120.degree., and the magnitude and the frequency of these signals determines the pulse widths and the number of pulses in pulse trains V.sub.as, V.sub.bs and V.sub.cs which are applied to the motor terminals. Voltages V.sub.as,
V.sub.bs and V.sub.cs are phase voltage signals incorporated in the line-to-line voltages observed across the stator terminals.
The AC inverter control signals V.sub.as *, V.sub.bs * and V.sub.cs * result from a 2-to-3 phase conversion accomplished via transformer 15 in a manner well known in the art. Input signals V.sub.qs and V.sub.ds are sinusoidal AC voltage command
signals having a control signal magnitude and a frequency. These signals are related to the stationary d-q reference frame in which torque controlling electrical parameters are related to the q-axis and flux controlling electrical parameters are related
to the d-axis. The q-axis leads the d-axis by 90.degree. in phase difference.
Control signals V.sub.qs and V.sub.ds are output signals from a synchronous current regulator 16. Details of regulator 16 have been previously shown and described in U.S. Pat. No. 4,680,695 which issued to Kerkman, et al. on Jul. 14, 1987
entitled "Cross Coupled Current Regulator" which is incorporated herein by reference. As described therein, regulator 16 includes a proportional-integral loop (PI loop) with summing inputs. At one summing input, an AC current command signal for the
q-axis i.sub.qs * is algebraically summed with a q-axis feedback signal i.sub.qsf to provide a current error for the q-axis. At a second summing input, an AC current command for the d-axis i.sub.ds * is algebraically summed with a d-axis feedback signal
i.sub.dsf to provide a current error for the d-axis. A stator operating frequency f.sub.e is also input to regulator 16. Frequency f.sub.e is multiplied by 2.pi. to generate an angular frequency .omega..sub.e in radians. With these input signals
regulator 16 controls the AC voltage command signals V.sub.qs and V.sub.ds at its outputs in response to current error and further, regulator 16 maintains vector orientation of output signals to the d and q-axes.
Referring still to FIG. 3, synchronous-to-stationary transformer 21 receives both d and q-axis synchronous command current signals i.sub.de * and i.sub.qe * and also receives phase angle .theta..sub.e and converts synchronous command signals
i.sub.de * and i.sub.qe * to stationary signals i.sub.ds * and i.sub.qs * accordingly. To this end, synchronous command current signal i.sub.de * is provided by flux regulator 22 while synchronous command current signal i.sub.qe * is provided by speed
regulator 24.
Referring still to FIG. 3 and also to FIG. 5, flux regulator 22 includes a summer 34, a PI controller 36, a switch 43, a d-axis search current initializer 45, a minimizer 69, another switch 42, a flux error calculator 35 and a comparator 37.
Regulator 22 receives both q-axis command voltage signal V.sub.qe * and feedback voltage signal V.sub.qef and uses those signals to generate synchronous d-axis current i.sub.de *. To this end, summer 34 subtracts feedback voltage V.sub.qef from
command voltage V.sub.qe * generating a q-axis voltage error signal V.sub.qe. Error V.sub.qe is provided to PI controller 36 which steps up that error thereby generating current signal i.sub.de. Signal i.sub.de and signal i.sub.de ' from initializer 45
are both provided to minimizer 69 which selects and passes the lowest of the two input signals i.sub.de and i.sub.de ' as a search current i.sub.de ".
Switch 43 is a single pole, double throw settable/resettable (i.e. once set only changes state when reset and once reset only changes state when set) device having two inputs and a single output. Switch 43 is used to select between two separate
input current values, one value being d-axis current i.sub.de as provided by controller 36, the other being d-axis search current i.sub.de " provided by minimizer 69. Current i.sub.de ' is typically less than a rated d-axis current and for the purposes
of this explanation, it will be assumed current i.sub.de ' is 1/10th a rated d-axis current value. To select between currents i.sub.de " and i.sub.de, switch 43 includes a first input linked to minimizer 69 and a second linked to a controller 36 and a
single output which provides command current i.sub.de *. Switch 43 also includes both a set input and a reset input R which are linked to flag signals .zeta..sub.1 and .zeta..sub.2, respectively. Inputs S and R are control inputs. During normal
operation the switch 43 output is normally linked to the output of controller 36. However, when flag .zeta..sub.1 goes from low to high, switch 43 output switches to the output of minimizer 69 providing the minimum value of i.sub.de " and i.sub.de as
search current. Switch 43 remains so switched until flag signal .zeta..sub.2 changes from high to low. Flags .zeta..sub.1 and .zeta..sub.2 are described in more detail below.
During a frequency s | | |
