Note: Descriptions are shown in the official language in which they were submitted.
~304775
METHOD AND APPARATUS FOR THE DIGITAL DETERMINATION OF THE FIELD
ANGLE OF A ROTATING-FIELD MACHINE
2 BACKGROUND OF THE INVENTION
3 Field of the Invention
4 This invention relates to a method for determining an
instantaneous field angle of a rotating field machine by
6 computing an electromotive force vector (EMF vector) from
7 values of the current and voltage of the stator of the machine,
8 and by means of integration of the EMF vector.
Related Art
11
12 In the interior of a rotating field machine (either a
1~ synchronous machine or an asynchronous machine) negative
14 feedback of the field angle takes place in accordance with its
physical structure. In such a case the field of the machine,
16 in the steady-state situation, is proportional to the component
17 of the stator current vector that is parallel to the field.
18 The torque (which is physically: the vector product of the
19 field vector and the current vector) with a constant field is
proportional to the component o' the stator current vector that
21 is perpendicular to the field. The individual winding currents
22 of the stator, the amplitude, frequency and phase of which can
2~ be impressed on the windings by a converter, can therefore be
2~ thought of as projections of a stator current vector given in a
stator-oriented coordinate system. The field-oriented vector
26 components which define the field and the torque may be
27 generated from these projections through a transformation. If
28 the input variables of the converter are therefore computed
29 from a field-oriented reference vector for the machine currents
and if this field-oriented reference vector is transformed into
,~
13047qS
1 the stator oriented coordinate system by a suitable
2 transformation, which represents a positive feedback with the
3 angle of the field vector with respect to the stator (i.e., the
4 "field angle"), a highly dynamic control of the torque and the
flux can be achieved. It is beneficial to be able to determine
6 the field angle.
7 Such a control is described, for instance, in
8 "Siemens Forschungs-und Entwicklungsberichte 3", 1974, pages
9 327 to 332 where it is stated that the field angle is, in a
first approximation, the angle of the stator voltage vector
11 phase-shifted by 90.
12 In "msr", I8 ap (1975), no. 12, pages 278 to 280, a
13 number of further methods for determining the field angle are
14 given. Unless Hall probes for directly measuring the field are
used, machine models are used which compute the field angle
16 using physical equations and different measured variables.
17 Computing inaccuracies, as well as an incorrect setting of the
18 machine parameters in the computing model, however, may lead to
19 inaccurate calculated values of the field using the model. For
different operating states, different machine models may be
21 more advantageous.
22 In the meantime, still further machine models, as
23 well as combinations of models, have been proposed. Thus, it
24 has been proposed in DE-OS 30 34 252 (U.S. Patent No.
4,447,787) to use an EMF detector which computes the EMF vector
26 from measured values of the stator current and the stator
27 voltage. In the steady state case, the magnitude of the flux
28 is proportional to the quotient of the magnitude and the
29 frequency of the EMF vector. In this case the direction of the
EMF vector with respect to the direction of the field is phase
2--
1 shifted by go. A more exact determination of the field in the
2 dynamic case is obtained by integrating the EMF vector.
3 However, the flux vector can also, in ways different from the
4 current and the rotor position, simulate the field excitation
in the rotor. Since, for both flux vectors to agree, the
6 machine parameters (which for the EMF calculation are: the
7 stator resistance and the stray inductance) must be consistent,
8 it is proposed to form a control deviation from the deviation
9 of the two flux vectors calculated in the machine model (or EMF
vectors). This control deviation is leveled out by a
11 correction control. This correction control is accomplished by
12 correcting corresponding machine parameters with a correction-
13 control output signal.
14 Such a balancing model furnishing the field angle
requires a feedback loop which further requires a complex
16 structure and time consuming controls. This is troublesome,
17 particularly if AC variables are to be processed and leveled
18 out.
19 A particular problem is raised here by the presence
of DC components in the measured actual value. Thus, for
21 instance, the measured values of current and voltage can have
22 DC components due to the inaccuracies of measuring
23 transformers, so that the EMF vector formed by the voltage
2~ vector which is itself derived after subtracting the ohmic and
inductive voltage drops, is no longer, in the steady state
26 case, a vector with constant magnitude. If the stator oriented
27 components of the field vector are calculated by integration
28 from the corresponding EMF components, offset errors and other
29 computing errors of the integrators that are used can lead to
further errors which are continuously further integrated. As a
1304775
1 result the model no longer operates in a stable state.
2 From DE-OS 34 18 573 (U.S. Patent No. ~,629,961), a
3 method is known in which, on the one hand, DC components in the
4 measurement errors are suppressed through a "weak" smoothing
~i.e., smoothing with a time constant which is large as
6 compared to the periods of the measurement value), but on the
7 other hand, are also corrected by a correction signal which is
8 ~ derived from a "volatile variable". This "volatile variable"
9 is a physical characteristic of the machine which assumes the
value zero in the steady state. The model value for the
11 "volatile variable" is also calculated in the machine model
12 which is incorrect if the value calculated deviates from the
13 value zero. This non-zero model value determines the magnitude
14 of a correction vector. The direction of this correction
vector i5 determined by shifting the incorrectly determined
16 model flux vector by a preset angle. .~s a rule, at least the
17 component of the model vector perpendicular to the incorrect
18 model value vector deviates from zero. The vectorial addition
19 of the correction vector to the EMF vector therefore requires
at least one addition point for the addition of this component.
21 The integration of the stator-oriented components of
22 the EMF vector requires two AC voltage integrators which, when
23 implemented with a digital microcomputer, have only limited
24 resolution. In DE-OS 34 18 640 (U.S. Patent No. 4,626,761) it
is proposed to transform the EMF vector, by means of a vector
26 rotator, into a rotating coordinate system and to subsequently
27 integrate the transformed vector. In the integration of a
28 vector in a rotating coordinate system, a rotary component must
29 be taken into account so that the transformed EMF components
are composed with the mentioned correction signal and with the
13047~5
1 rotary component to form the input vector of a suitable
2 integration circuit. If the rotating coordinate system is
3 coupled to the field vector with rigid phases, the magnitude of
4 the actual field vector is always identical with its field-
parallel component and the orthogonal component perpendicular
6 to the field has the value zero. Therefore, the corresponding
7 orthogonal component of the integrated vector is fed at the
8 output of the integration circuit to a zero-point controller.
9 The output signal of the zero-point controller represents the
frequency of rotation of the coordinate system which, after a
11 further integration, supplies the transformation angle for the
12 vector rotator.
13 The corresponding circuit thus contains an additional
1~ control loop for forming the transformation angle, hut the
frequency of rotation as well as the transformed components of
16 the EMF vector, the correction signal and the rotary EMF
17 components are now DC signals which are easier to integrate in
18 the microprocessor.
19 In DE-OS 34 18 641 (U.S. Patent No. 4,593,240) a
circuit for the integration of the EMF vector is described
21 which likewise transforms the EMF vector (after suppression of
22 DC components) by means of a vector rotator into a rotating
23 coordinate system. The system of this reference is complicated
2~ in that the formation of the flux vector is performed with a
total of two integrators and without a rotary EMF component and
26 without a control "transformation angle". Rather, according to
27 this system the magnitude of a field vector is formed through
28 the integration of the one transformed and corrected EMF
29 component. A divider furnishes, from the quotient of the other
EMF component and the field vector magnitude, the field
130477~i
1 frequency. The second integrator determines the transformation
2 angle from the field frequency. The transformation angle then
3 corresponds to the field angle and the transformation frequency
4 corresponds to the field frequency.
A field angle determined in one of the above manners
6 is required for the phase control of the stator current in
7 order to assure the timely firing of the valves of a feeding
8 converter as a function of the field angle. Therefore, a
9 suitable device for determining the field must make available
the instantaneous field angle as a steady-state or at least
11 quasi-steady actual value with high resolution (if possible,
12 only fractions of a degree). on the other hand, frequencies of
13 the winding currents, as required in many applications, can be
14 as high as 100 Hz. This requires that the instantaneous field
angle must be read at a field determination device with a clock
16 frequency that is so high that a digital determination of the
17 field angle does not seem to be realizable.
18
19 SUMMARY OF THE INVENTION
21 The present invention overcomes the difficulty
22 associated with applying a field angle to a phase control for
23 the stator current of a machine as described above. The
24 present invention selects from the multiplicity of known field
angle determinations, a method suitable in view of digitizing
26 with high time resolution, and carries out this method with a
27 computing effort as simple as possible and with a suitable
28 choice of the interfaces between required computer building
29 blocks.
This problem is solved by a method for determining an
130~5
1 instantaneous field angle of a rotating-field machine in which
2 values for a current and voltage of the stator of the machine
3 are measured and a value for the frequency of the EMF vector is
4 calculated. This frequency value changes discontinuously at a
slow rate.
6 A quasi-continuous angle signal for the field angle
7 is read from an integrator controlled by the discontinuously
8 changing value at a rate which is faster than the slow rate.
9 The method of the present invention may further
include the steps of producing the magnitude of the EMF vector
11 and transforming the magnitude by division with the
12 discontinuous value of the frequency into a computed value for
13 the magnitude of the flux.
14 The calculation of the value of the frequency
changing discontinuously is done by digitally differentiating a
16 direction angle of the EMF vector to form a value and
17 correcting deviations in this value by using a correction
18 signal, said correction signal which is produced by leveling a
19 deviation of the quasi-continuous angle signal from the
direction angle with a time constant greater than the frequency
21 of rate. The value of frequency may be subsequently smoothed.
22 The step of measuring may include the step of
23 sampling the current and the voltage of the stator of the
24 machine to synchronously produce the EMF vector.
The method may further include calculating the EMF
26 vector in Cartesian coordinates, filtering out DC components of
27 the EMF vector, and calculating the frequency of the EMF vector
28 from the filtered components of the EMF vector.
29 The problem is also solved by a method for
determining the instantaneous field angle of a rotating field
1304775
1 machine by calculating the EMF vector from values for the
2 stator current and the stator voltage of the machine, in which
3 the values for the stator current and the stator voltage and as
4 well as a fed-back value for the field angle are read into a
calculation circuit at a slow rate and, stator-related
6 components of the EMF vector from the values for the stator
7 current and stator voltage and fed-back value are calculated.
8 In addition the components of said EMF vector are integrated
9 and a value for the frequency of said integrated EMF vector
components is calculated, said value for the frequency changing
11 continuously the slow rate. The changing values for the
12 frequency are integrated by means of an integration circuit to
13 form an angle signal representing the field angle, and the said
1~ angle signal changes at a rate which is faster than the slow
lS rate of reading in. The angle signal is used as the value for
16 the instantaneous field angle and also as a fed-back value for
17 the field angle.
18 A device suitable for this invention is a digital
19 device for determining a flux angle of a rotating field machine
that includes a clock frequency transmitter and a
21 microprocessor that receives an output signal of the
22 transmitter as an input along with digital values representing
23 current and voltage. The microprocessor reads in the inputs at
2~ a slow rate and calculates a digital value for the frequency of
the EMF vector and provides this calculation result as an
26 output. An integrator, external to the microprocessor has as
27 its input the output of the microprocessor. The integrator
28 calculates a quasi-continuous value for the field angle. An
29 additional clock transmitter controls the integrator to read
out the field angle at a rate faster than the slow rate at
130A775
20365-2816
which inputs are read into the microprocessor.
Another digital device for carrying out the inven-
tion includes a clock frequency transmitter and a micro-
processor which reads in digital values representative of
stator current and voltage at a rate governed by the transmitter.
The microprocessor includes an EMF detector, a vector rotator
to rotate the output of the EMF detector and an integrator that
integrates the output of the vector rotator. Another
integrator, external to the microprocessor calculates a quasi-
continuous value for the field angle based on the output of themicroprocessor's integrator.
According to a broad aspect of the invention there
is provided a method for determining an instantaneous field
angle of a rotating-field machine having a stator, an EMF
vector being developed in the machine, comprising the steps of:
(a) measuring values for current and voltage of the
stator of the machine;
(b) numerically calculating from said measured values a
discontinuously changing value ~ for the frequency of the EMF
vector, said value ~ changing at a rate t~; and
(c) reading, as the field angle, a quasi-continuous
angle signal ~ from an integrator controlled by the discontinu-
ously changing value ~ at a rate t~ which is faster than rate
t,u ~
According to another broad aspect of the invention
there is provided a method for determining the instantaneous
field angle of a rotating field machine by calculating the EMF
vector from values for the stator current and the stator vol-
tage of the machine, comprising the steps of: reading into a
calculation circuit at a rate tu the values for the stator
1304'775
20365-2816
current and the stator voltage and a fed-back value for the
field angle; and numerically calculating within said calculation
circuit:
i) stator-related components of the EMF vector
from said values for the stator current r said stator voltage
and said fed-back value for the field angle; and
ii) a value ~ for the frequency of a flux vector,
said flux vector being calculated from said stator-related
components of said EMF vector, said value ~ for the frequency
changing at a rate t~, said value ~ being calculated in a
microprocessor;
feeding said changing value ~ for frequency into
an integration circuit, external to said microprocessor to form
an angle signal ~ representing the field angle, said angle
signal ~ changing at a rate of t~ which is faster than tu, and
usiny said angle signal as the value for the instantaneous field
angle and also using said angle signal as a fed-back value for
the field angle.
According to another broad aspect of the invention
there is provided a method for determining a field angle of a
rotating-field machine having a stator and an EMF vector,
comprising the steps of:
(a) measuring values for a current and voltage of the
stator of the machine;
(b) numerically calculating from said measured values
a discontinuously changing value ~ for the frequency of the
EMF vector, said value ~ changing at a slow rate tu;
(c) reading a quasi-continuous angle signal ~ for the
field angle from an integrator controlled by the discontinu-
ously changing value ~ of a rate t~ which is faster than rate
- 9a -
13~4~7~;
20365-2816
tu;
(d) detecting the frequency of said field angle and if
said frequency is greater than a threshold value repeating
steps (a) to (d) and if ~he frequency of said field angle is
equal to or less than said threshold performing the following
steps (e) to (k);
(e) reading into a calculation circuit at a rate tu the
values for the stator current and the stator voltage as well
as a fed-back value for the field angle;
(f) numerically calculating stator-related components
of the EMF vector from said values for the stator current,
stator voltage and fed-back value,
(g) integrating said stator-related components of the
EMF vector;
(h) calculating a value for the frequency of said
integrated stator-related components of the EMF vector said
value ~ for the frequency changing at rate tu;
(i) integrating said changing value ~ for frequency by
means of an integration circuit coupled to said calculating
circuit to form an angle signal for the field angle, said angle
signal changing at a rate of t~ which is faster than tu;
(j) using said angle signal as the instantaneous field
angle and also using said angle signal as a fed-back value for
the field angle, and
(k) detecting a frequency of said angle signal, if
said frequency is less than or equal to a threshold repeat steps
(d) to (k) and if said frequency is greater than said threshold
perform steps (a) to (d).
According to another broad aspect of the invention
there is provided a digital device for determining a flux angle
- 9b -
` i3~4~77S
20365-2816
of a rotating field machine having a stator and a measuring
device disposed between said stator and the digital device
comprising:
a) a microprocessor receiving digital values
representative of a current and a voltage of the stator of the
rotating field machine as inputs, said microprocessor reading
in said inputs at a slow rate t~ of a first clock signal,
identifying an EMF vector from said inputs and calculating at
said slow rate, a digital value for the frequency of said
identified EMF vector;
b) an integrator, external to said microprocessor,
having as an input said digital value for the frequency of a
detected EMF vector, said integrator calculating a quasi-
continuous value for the field angle that changes at a rate t~
faster than tp.
According to another broad aspect of the invention
there is provided a digital device for determining a flux angle
of a rotating field machine comprising:
a) means for producing a first clock signal;
b) a microprocessor receiving digital values
representative of the current and voltage of the stator of the
field machine as inputs, said microprocessor comprising:
i) means for detecting an EMF vector from said
input digital values;
ii) means for rotating said detected EMF vector;
and
ii.i) means for integrating said rotated detected EMF
vector to produce the digital value of the frequency of a flux
vector at a rate of tu controlled by said first clock signal;
and
-- 9c --
1304~75
20365-2816
c) an integrator, receiving as an input said digital
value for the frequency of the flux vector from the micro-
processor, for calculating a quasi-continuous value for the
field angle that changes at a rate t~ faster than tu.
BRIEF DESCRIPTION OF THE DRAWINGS
Figure 1 illustrates principle of a power section
of a rotating field machine according to an embodiment of the
present invention and of the control in a synchronous machine
fed via an intermediate-link converter.
Figure 2 illustrates an embodiment of a processor
module of Figure 1.
Figure 3 illustrates an embodiment of a structure
for flux computation of Figure 2 for determination of the field
angle for high frequencies.
Figure 4 illustrates an embodiment of a structure
for flux computation of Figure 2 for determination of the field
angle for lower frequencies.
DETAILED DESCRIPTION
-
Figure 1 illustrates the use of a field angle
- 9d -
1304775
1 determination according to the present invention for impressing
2 the stator current in, for example, a synchronous machine. The
3 synchronous machine SM contains two winding systems, offset
4 relative to each other, ~hich are fed via an intermediate-link
S converter. Each intermediate link converter consists of an
6 inverter (MSl and MS2 respectively) on the machine side which
7 is supplied via an intermediate link choke C1 and C2,
8 respectively) from an inverter (NS1 and NS2, respectively)
9 which is connected on the network side to a main supply network
NM with a controlled intermediate circuit DC current. The
11 rotor winding of synchronous motor SM is fed from a rotating
12 excited device ER which is connected via a converter ES on the
13 exciter side to an auxiliary supply network NH.
14 The machine is regulated and the control of the
converters takes place digitally. They are subdivided into
16 several function packets (FP) which are associated with
17 individual processor modules: namely, the function packets
18 FP-NS1 and FP-NS2 control the converters on the network side of
19 the synchronous machine (formed by NSl-Cl-MSl and NS2-C2-MS2,
respectively) and the function packet FP-ES controls the
21 converter ES on the exciter side of the synchronous machine. A
22 universal processor module includes function packet FP-NC which
23 operates as a speed control as well as a function packet for
24 the sequence control of the entire procedure. Finally,
peripheral subassemblies such as measured-value transmitters
26 and converters are associated with the individual processor
27 modules which include the function packets, arranged partially
28 on an interface module.
29 The two processor modules that have the function
packets FP-NSl and FP-NS2 are of identical design and they have
--10--
4775
1 the object of forming firing pulses for the valves of the
2 converters NSl and NS2 respectively on the network side in
3 order to control, by phase-gating, the current fed into the
4 respective DC intermediate circuit. Thereby, the amplitude of
S the currents to be fed to the stator windings are controlled to
6 a value which is supplied by the module FP-NC and corresponds
7 to the desired torque. For this purpose, the interface
8 electronics contain voltage/frequency converters for
9 determining the actual current value, a rectifier with
subsequent analog/digital conversion for forming the network
11 voltage as well as other components provided for the operation
12 of the network converter such as a hardware circuit (not shown)
13 to determine zero crossing of the current. The frequency and
14 digital conversions of the actual current values take place in
the hardware (HW) of the module, while most of the software
16 functions of the module are shown schematically in Fig. 1.
17 Additional software functions not pertinent to this invention
18 are represented by SW.
19 In FP-NSl a current control IR furnishes a voltage
reference value Ud* which is limited by a limiter BG1 to
21 permissible values. The voltage value is recalculated in a
22 linearizing stage LIN, that takes into consideration the
23 network voltage Un, into the control angle~ = cos l(Ud~/Un). A
24 control unit SS which is synchronized with the measured value
of the instantaneous network voltage, makes the firing pulses
26 for the converter valves available.
27 The processor module including the function packet
28 FP-ES furnishes, as a function of the current reference¦i*~ the
29 exciter current reference value for the exciter current
controller IER by means of a function generator FG. The output
--11--
~3~47'75
1 signal of IER is limited by a limiter BG2 and furnishes in a
2 control unit, the control signals for the converter ES on the
3 exciter side. Hardware components for voltage/frequency
4 conversion of the actual value of the exciter current in the
converter ES, as well as proper hardware at the processor
6 module for frequency/digital conversion of the actual current
7 value, are provided in the portion designated HW.
8 These processor modules further form, like the
9 processor module with the functional packet FP-MS according to
the invention, suitable monitoring quantities, acknowledgment
11 signals and call-back signals which are fed to the processor
12 module FP-NC. This latter msdule in turn provides start and
13 release signals to the function modules, controlling the
14 converters. It is shown separately in Fig. 1, that the module
FP-NC transfers the field frequency as the actual speed
16 substitute value to a software building block NR for speed
17 control, by which the current reference value¦i*¦is formed.
18 The actual speed value is formed by a startup generator ~G from
19 a reference value n* which is entered by the operator. A
sequence control TC obtains its starting and stopping command
21 likewise from the operator, and furthermore, the operator can
22 enter the parameter values, adapted to the respective machine,
23 for control, for the selection of the mode of operation, and
24 other data as well. The sequence control TC thereby represents
an interface between data to be entered by the operator and the
2~ control and regulation furnished by the manufacturer. In
27 addition, special operating characteristics, for instance,
28 speed, voltage, flux and EMF can be monitored here for assuring
29 adherence to given operating regions, and to initiate an
orderly shutdown if necessary, in the event of impermissible
i3~75
1 values for these operating characteristics.
2 Two-phase currents (iRl, iSl and iR2, iS2)
3 at least two interlinked voltages of a stator winding system
4 UsT and UTR are measured at the lead wires of each stator
winding system, and are made available via a voltage/frequency
6 converter to the corresponding processor module FP-MS which
7 therefrom forms the digital measurement values as illustrated
8 in Fig. 2.
9 In Fig. 2, the operation of this processor module
FP-MS is shown in greater detail, where the analog/digital
11 conversion, which is not the subject of the present invention,
12 is shown as a separate element distinct from the computing
13 operation of the processor proper. The entire control of the
14 converters MS2 and MS1 on the machine side of the converters
NS2-C2-MS2 and NS1-C1-MSl is accomplished here by a computer
16 module which consists of the processor itself, FP-MS' and a
17 trigger module FT which uses hardware that is separate from
18 that of the processor itself. In the processor itself, FP-MS',
19 a control angle calculation A-CAL occurs first so as to fix the
angle between the stator current vector to be impressed and the
21 EMF vector. This control angle can be calculated selectably
22 according to different parametric strategies. For instance,
23 one strategy might be operation with a fixed power factor, with
2~ a minimum quenching angle or it might be an angle given by
field-oriented control, if desired, as compared with the flux
26 vector. The sequence control TC determines the choice of
27 strategies to be used. In the simplest case, a constant
28 control angle is given in order to always commutate the stator
29 current in the predetermined firing cycle of the converter
valves on the machine side to the next winding when the EMF
-13-
~3(~477~;
1 vector or the flux vector reaches a predetermined position in
2 space.
3 The control angle~can advantageously be stored
4 temporarily in a separate hardware memory A-MEM so that it is
still available for a controlled inverter operation in the
6 event of a failure of the processor module FP-MS. The trigger
7 module FT contains a comparison point to compare the control
8 angle~ with the field angle (corresponding to the phase shift
9 between the EMF vector and the field vector phase shifted by
90). In the event that they are equal, a subsequent control
11 stage ST forms the firing points in time directly, by which the
12 valves of the converters MSl and MS2 on the machine side are
13 fired according to a firing sequence fixed for one revolution
14 of the current vector.
The trigger building block FT thus requires the
16 instantaneous value of the field angle ~ , which must be
17 present for an exact control with high time resolution, i.e.,
18 quasi-continuously. Of particular concern is the digital
19 formation of the field angle with high resolution.
For this purpose, a method is used which calculates
21 the EMF vector from the digitally available values for the
22 current and voltage of the machine stator, and determines
23 therefrom the instantaneous field angle, with at least one
24 integration being performed.
According to an embodiment of the invention, a value
26 for the frequency of the EMF vector ~J which changes at a slower
27 clock rate (i.e. "discontinuously") is calculated and, a value
28 changing at a faster rate is read out as a quasi-continuous
29 angle signal at the integrator which is arranged outside the
microprocessor and is controlled by the discontinuous value.
~:~04~75
1 This present invention assumes that as a result of
2 the inertia of the machine, the flux vector and the speed of
3 rotation of the rotor change only relatively slowly. It is
4 therefore sufficient if, in a microprocessor, the current and
voltage values are read in synchronously at a slower rate
6 adapted to the computing capacity of the processor. The
7 processor may form, in an EMF detector stage E-DET of Fig. 2,
8 the EMF vector _ which is present at the time t~Lof the
9 reading. A flux calculating stage F-CAL also of Fig. 2
determines the frequency which is obtained fram the motion of
11 the EMF vector resulting from the motion of the flux vector
12 relative to the respective read-in times t~ , t~, etc. or the
13 motion of the flux vector corresponding to this EMF vector.
1~ This frequency ~ is therefore a frequency value which changes
at the slower read-in clock rate ("discontinuously") and from
16 which a field angle can be formed by in~egration. The field
17 angle changes continuously if an analog integrator is used or
18 changes quasi-continuously if an integrator INTl is used which
19 operates at a high operating rate.
Since the integrand of this integrator deviates only
21 insignificantly from the true flux frequency during the period
22 of read-in time t~, the values of the integral (angle
23 signal ~) deviate only insignificantly from the true field
2~ angle, which values are preset to the clock pulses t~ of the
computing-clock transmitter of the integrator INTl. This
26 deviation can be corrected by feeding back the angle signal to
27 the microprocessor at a slow rate.
28 Advantageously, the reading in of the fed-back angle
29 signal into the microprocessor takes place synchronously with
the reading of the current and voltage values, i.e., all read-
-15-
~3047~75
1 in values are instantaneous values belonging to the s~me points
2 in time t~, so that no additional phase shifts occur thereby.
3 In the microprocessor, the frequency can then be readjusted so
4 that, for the reading-in times t~L , equality between the angle
signal and the field angle discontinuously formed in the flux
6 computer F-CAL at these reading-in times is achieved. This is
7 described as synchronization. Since the feedback effect of the
8 angle signal increases with rising frequency, but on the other
9 hand, the flux calculation becomes more and more accurate at
high frequencies, the synchronization may in some cases be
11 dispensed with completely or the correction control provided
12 for the synchronization can be equipped with a large time
13 constant.
14 Plane vectors must be described by two defining
quantities. For the hardware integrator INTl, however, only
16 the derivative ~Jin time of the polar angle coordinate of the
17 calculated flux vector is required of the flux vector
18 calculated for the clock times t~L. In Fig. 2, the computing
19 stage F-CAL also computes the magnitude~ of the flux vector
together with the frequency ~ from the microprocessor. The
21 magnitude 1~l is fed back to the sequence control TC for
22 monitoring purposes.
23 For describing the F~MF vector e, the orthogonal
24 Cartesian components e~ and e~ of the EMF vector in a stator-
related coordinate system are used advantageously. The
26 currents and voltages required for the EMF calculation in a
27 stator winding system flow in windings which are shifted 120
2B relative to each other. Therefore, a coordinate transformation
29 of skewed coordinates which are associated with the three
individual windings of a stator winding system into orthogonal
-16-
1304775
1 coordinates is necessary, which is known as a "three/two
2 conversion" or 120/90 conversion". It is immaterial at which
3 point in the input channel of the computing stage F-CAL this
4 conversion takes place. This conversion can be applied to the
measurement values of the stator currents and stator voltages,
6 so that orthogonal components of the current vector and the
7 voltage vector can be fed to the EMF detector E-DET.
8 In Fig. 2 this conversion is performed at the output
9 of an offset control stage O-REG. This control stage is
primarily advantageous when, through offset errors of the
11 preceding hardware (that is, the measuring stages and
12 analog/digital converters), the components of the EMF vector,
13 which are basically pure AC voltage quantities, have a DC
14 component. For the offset control, very weak smoothing is
sufficient here, i.e., for instance, by means of a feedback
16 integrator, the time constant of which is large as compared to
17 the period of the reading times t~L , as is illustrated in the
18 functional block O-REG of Fig. 2.
19 EMF detector E-DET of Fig. 2 determines the EMF. It
is sufficient in a machine with several stator winding systems,
21 to calculate the EMF vector in a single winding system by
22 subtracting from the voltage vector u of this winding system,
23 the ohmic voltage drop R.i, the ohmic resistance R of the
24 supply circuit and the ohmic voltage drop R.l corresponding to
the current vector 1, as well as the inductive voltage drop
26 Ls.( ~ dt). That is,
27
28 e = u - Ri - L , ( i/dt)
29
However, the individual stator winding systems are
-17-
1304~75
1 coupled inductively to each other so that, corresponding to a
2 coupling factor x, the inductive voltage drops belonging to the
3 stator current veGtor of the other winding system must be taken
4 into account. For the winding system (voltage vector u2,
current vector i2) fed from the converter MS2 on the machine
6 side, the following vector equation is obtained.
8 - e = u2 - R12 _ Ls tdi2/dt + X dil/dt)
This equation may also be implemented by E-DET. In
11 the example of Fig. 1, this vector equation in the form of two
12 scaler equations for the skewed-angle components of the vectors
13 are calculated.
14 Fig. 3 illustrates, an embodiment of flux calculation
circuit F-CALl for the high frequency cases. The embodiment is
16 labelled F-CAL1. The dashed lines ~/S indicate here the
17 interface between the software-controlled microprocessor and
18 the hardware at its input and output. Vectors which are
19 represented by two signals corresponding to their definition
quantities (according to Fig. 3, advantageously the Cartesian,
21 skewed angle or orthogonal components), are indicated as double
22 arrows. The structure of the EMF detector E-DET, as well as
23 that of the offset control O-REG can be chosen according to
24 Fig. 2. A Cartesian/polar converter K-P receives the EMF
vectors which are the outputs of offset control stage O-REG and
26 forms therefrom polar definition quantities. In the case of
27 orthogonal EMF control components e , e , the magnitude
28 signal is therefore
29
-
e = ~ e2 + e2~
-18-
~3~775
1 and the angle signal which changes discontinuously is
3 ~ = tan~1 te~ / ed )
The frequency of the EMF vector is derived using software with
6 a digital differentiator DIF which calculates the differential
7 quotient by the differential quotient
g ~/d~
~
11 where ~and ~ I are the polar angle coordinates of the EMF
12 vectors belonging to the reading times t~ and t~_l ; T~ ,
13 the period of the reading times and where TD is a time-
14 normalizing constant adjusted to the time constant of the
hardware integrator INT1 of the trigger building block FT of
16 Fig. 2. TD corresponds to the time-normalizing constant
17 corresponding to the nominal frequency.
lB A divider DIV1 forms therefrom the quotient e/~ as
19 the magnitude ~ of the calculated flux. The flux vector is
physically the integral of the EMF vector. The integration,
21 however, can be avoided at high frequencies by replacing it
22 with the magnitude component of the flux vector by e/~ while
23 the angle coordinate is replaced by ~ + 90.
24 In the stationary state, this approximate
determination of the flux vector with the determination of the
26 EMF magnitude and the EMF frequency is exact and it changes
27 only the dynamic state. In many applications, especially in
28 asynchronous machines, even computing errors such as are
29 caused, for instance, by rounding-off errors of the
differentiation stage DIF can be tolerated without problem. In
i30477S
1 other cases, however, it may be advantageous to follow the
2 differentiation stage DIF with a further smoothing member GL.
3 Thereby, however, a frequency-dependent phase error is
4 generated.
This phase error can be compensated by the provision
6 that the angle ~belonging to the respective reading times t~ is
7 synchronized with the angle~belonging to these points in time.
8 For this purpose, the correction control CR is provided, to
9 which the angle difference is fed and the output signal of
which can be added to the output signal of the smoothing member
11 GL.
12 Since reading-in the current, voltage and angle
13 signal takes place synchronously, the identity between the
14 discontinuously calculated angle~ and the guasi-continuously
calculated angle signal ~is ensured at these synchronous points
16 in time if the correction control CR is balanced. The hardware
17 integrator INTl thus serves as an interpolator for the
18 discontinuously calculated ~-values. Here, the differentiating
19 stage DIF is required in the microprocessor.
At lower frequencies it is often advantageous for the
21 calculation of the flux to use a structure with which an
22 integration of the EMF vector actually takes place, but no
23 differentiation is performed for forming the frequency ~J. To
24 this end it is merely necessary to switch, at a given limit of
the frequency ~ to another program in the microproces~r for
26 the computation of the flux. For instance, if ~Jfalls below a
27 certain threshold the output of offset control stage O-Reg is
28 fed to the program F-CAL2 of Fig. 4. To the extent that with
29 this new flux calculation, the starting values of integrators
or other computing elements are required, the values for the
-20-
i30477~
1 flux ~, the EMF e and the frequency 6~ can be taken from the old
2 prog~am F-CALl into the new program F-CAL2, as is indicated in
3 Fig. 3 by corresponding outputs of the program section F-CALl.
4 Vice versa, a corresponding starting value for the
S differentiation stage DIF as well as the value zero can be
6 preset for the correction control CR when swit~hing back to the
7 program section ~-CAL1.
8 No further timing problems arise when switching over
9 since the hardware integrator INT1 can run unsynchronized for
one or more computing cycles required for calculating the
11 discontinuous frequency value ~.
12 According to Fig. 4, the method for determining the
13 instantaneous field angle at lower frequencies utilizes a
14 calculation of the EMF vector (program section E-DET) which can
be followed by an offset control stage O-REG. The program
16 section F-CAL2 follows and contains at least one integration
17 stage INT2 which is in addition to the hardware integrator INTl
18 of the trigger building block FT of Fig. 2.
19 In this method also, the values for the stator
current and stator voltage are likewise read into the
21 microprocessor at a slower rate and with a fed-back value for
22 the field angle are read-in synchronously. From them, the
23 stator-related components e~ , e~ of the EMF vector and
24 therefrom, by means of a first integration, a value L~, changing
discontinuously at a slower rate for the frequency of the flux
26 vector (integrated EMF vector), is calculated. This
27 discontinuous value ~is integrated in the hardware integrator
28 INTl of FT to form a faster-changing angle signal ~ for the
29 field angle ("second integration") and the angle signal is
taken off on the one side as the value of the instantaneous
-21-
1304775
1 field angle and is scanned on the other hand as a fed-back
2 value for the field angle.
3 Advantageously, the read-in values are converted in
4 the microprocessor, by means of the offset-control stage O-REG
into smoothed stator-related orthogonal components of the EMF
6 vector. In a vector rotating stage VD of F-CAL2, these
7 orthogonal components are subjected to a coordinate
8 transformation with the fed-back value of the field angle as
9 the transformation angle. The transformation angle is
converted here advantageously into its angle functions cos ~,
11 sin ~ since the vector transformation then consists of only
12 four multiplications and two additions.
13 In contrast to the stator-related orthogonal
14 com~onents of the EMF vector, which represent AC signals, the
transformed orthogonal components are now represented by DC
16 voltage signals, the integration of which does not require a
17 high working cycle.
18 Advantageously, a correction vector is vectorially
19 added in a damping stage DAMP to the transformed vector,
corresponding to the German Offenlegungsschriften 34 18 573, 34
21 18 640, and 34 18 641, mentioned at the outset.
22 From these publications the structure of suitable
23 integration stages INT2 can also be seen.
2~ In Fig. 4, an integration stage INT2 realized in
software is indicated which requires only a single integrator
26 for forming the flux magnitude ~, while the discontinuous
27 frequency value is formed by means of a division stage.
28 This form of the integration stage INT2 therefore
29 does not require a differentiation stage for determining, from
the components of the transformed EMF vector, the discontinuous
-22-
130~
l value ~, which must be entered as the integrand into the
2 h~rdware inteqrat~r INTl.
11
12
13
14
16
17
18
19
21
22
23
24
26
27
28
29
