BACKGROUND OF THE INVENTION
The invention relates to a method for a frequency converter,
when the frequency converter controls a motor and operates torque-controlled partly
or entirely in a field weakening region.
By using frequency converters, the rotation speed of a
controlled motor can be increased considerably above the nominal frequency of the
motor. Increasing the rotation speed above the nominal frequency requires typically
a reduction in magnetisation, whereby the output frequency increases above a field
weakening point, i.e. the machine is said to be in a field weakening region.
Earlier, solutions based on direct torque control have
tried to control a flux phasor in the field weakening region in such a manner that
its head forms a circle in the spatial coordinates. At the same time they have tried
to avoid using zero phasors in keeping the voltage in the motor poles as high as
possible. A maximum voltage would be achieved, if the flux phasor formed a hexagon.
Each angle of the hexagon would correspond to the directions of voltage vectors.
With such a shape, it would be possible, in theory, to achieve a 20% increase in
the torque in the field weakening region as compared with a circular flux trajectory.
Eliminating the zero phasors has not, however, been successful with the earlier
implementations and with a hexagonal flux, and thus, it has not been possible to
utilise a maximum voltage.
In the earlier implementations, control is based on complex
calculations that due to the complexity need to be done in a slow time domain, whereby
the stability and dynamics of the control suffer at high rotation speeds.
Document DE 196 40591 C discloses a method for regulating
torque of an asynchronous motor. The regulation is based on actual flux value and
the sector of flux vector.
BRIEF DESCRIPTION OF THE INVENTION
It is an object of the present invention to provide a method
that avoids the above-mentioned drawbacks and enables the control of a motor torque
in a field weakening region more reliably than before. This object is achieved by
a method of the invention as claimed in the independent claim 1.
The invention is based on the idea that the change caused
by the voltage vector change to the torque can be predicted in a simple and reliable
manner. In the method of the invention, the torque can be predicted in the fastest
time domain of control, whereby the dynamics and stability of the control are ensured.
In addition, by means of the method of the invention, it is possible to utilize
in a field weakening region the entire voltage of the intermediate circuit of the
frequency converter, since the control implemented by means of the method uses no
BRIEF DESCRIPTION OF THE FIGURES
The invention will now be described in more detail by means
of preferred embodiments and with reference to the attached drawings, in which
DETAILED DESCRIPTION OF THE INVENTION
- Figure 1 shows the voltage vectors and flux circle of a frequency converter,
- Figure 2 shows a reference torque and an actual torque in a time domain,
- Figure 3 shows the prediction of a torque vector as a vector diagram, and
- Figure 4 shows a flow chart of the method of the invention.
When a frequency converter operates in a steady state constant
flux area, the flux vector draws a circular trajectory. In such a case, the torque
can be kept in a simple manner within hysteresis limits by using zero vectors at
required moments. When the frequency increases, less zero vectors are used on an
average, and the angle of the flux vector affects the number of the zero vectors.
The most zero vectors are used when the flux vector is close to the sector edges,
and correspondingly, the least when the flux vector is in the middle section of
the sectors. When the output frequency increases up till the field weakening point,
zero vectors are not used in the middle sections of the sectors. In this situation,
torque control cannot keep the torque within the hysteresis limits, and the field
weakening control should reduce the flux in such a manner that the torque remains
In the method of the invention, torque control is implemented
in such a manner that the used stator flux becomes smaller at the same time. In
the method of the invention, a torque estimate Tpred is predicted that
would be achieved when a voltage vector change occurred at the instant of prediction,
the predicted torque estimate is compared with a reference torque, the direction
of travel and sector of a stator flux vector is defined, and the voltage vector
change is implemented when the predicted torque estimate is smaller than the reference
torque and the stator flux vector is moving in a positive direction of travel or
when the torque estimate is greater than the reference torque and the stator flux
vector is moving in a negative direction of travel.
This is implemented in such a manner that at a most suitable
instant of time, a single voltage vector is selected that implements the reference
torque at a predicted instant. In addition, this voltage vector is maintained until
the reference torque is achieved, unless a second voltage vector should be selected
earlier. There is then no attempt to keep the flux circular. At the instant of prediction,
the flux detaches itself from the circle and attached again to it at the predicted
When the operating point of the controlled machine is in
the field weakening region or close to it, it is checked in the fastest time domain
of control whether the flux adjustment of the invention should be started or not.
Figure 1 shows the voltage vectors U1 to U6 of the frequency converter and the flux
According to a preferred embodiment of the invention, the
prediction of the torque estimate Tpred takes place on the basis of a
defined stator flux vector &PSgr;̅
and rotor flux vector &PHgr;̅
and the stator and rotor flux vector estimates produced by a possibly implemented
The rotor flux vector &PHgr;̅
can be defined in a reduced form as a function of the stator flux and current
is a constant term related to the controlled motor.
The torque that is typically calculated as a cross product
of the stator flux and current can also be calculated using the stator and rotor
Thus by predicting the behaviour of the stator and rotor flux, a predicted value
for the torque is obtained at the end of the adjustment. It is clear that the size
of the stator flux cannot be measured from the motor, but it is estimated in the
frequency converter by using a model made of the motor.
The execution of the adjustment and the operation of the
method of the invention will now be described with reference to Figure 3. The figure
shows the defined stator flux vector &PSgr;̅
and rotor flux vector &PHgr;̅
. The figure also shows a voltage vector U3, the selection of which moves
the stator flux to the desired direction. For the sake of symmetry, both the starting
point of the adjustment and its end point form the same angle with the edge line
OA of the sector. Thus, the predicted stator flux &PSgr;̅
is a mirror image of the defined stator flux and its direction can be calculated
at its simplest by defining a new set of coordinates that is attached to the edge
line OA of the sectors and by performing a coordinate transformation to the necessary
The coordinate transformation can be performed by defining
a unit vector s̅ that is parallel with the edge line of the sectors
and the position of which is defined in the stator coordinate system
The stator flux vector is in turn defined in the same coordinate system
The coordinates of the stator flux can now be calculated
in the edge line coordinate system as follows:
wherein fix is the x-directional component of the flux and fly is the y-directional
To calculate a predicted stator flux vector, the defined
stator flux vector should be turned to the extent of the angle 2*atan(ay/ax). The
predicted stator flux vector can, however, be calculated using a trigonometric function
simply by multiplying the stator flux vector twice by the vector ax + j*ay. The
predicted stator flux vector can be presented by using the auxiliary variables fx
and fy in the component form &PSgr;̅
= flx_pred + j*fly_pred:
To calculate a predicted rotor flux vector &PHgr;̅
, one should first define the time &Dgr;t that the adjustment to be used
requires. That is, the time during which the voltage vector selected at the prediction
instant is kept valid. Let us assume that during the adjustment, the voltage Uc
of the intermediate circuit of the frequency converter is constant, in which case
the amount &Dgr;&psgr;s of the stator flux change is
Then again, &Dgr;&psgr;s is defined in the edge line coordinate system
in Figure 3 as
By eliminating &Dgr;&psgr;s from the previous equations, the time
During the adjustment, no significant changes occur in
the angular speed of the rotor flux due to the inertial mass of the rotor. Therefore,
the rotor flux turns during the adjustment to the extent of the angle
&phgr; = &Dgr;t&ohgr;
, wherein &ohgr;s is an average output frequency of the frequency
converter during the adjustment. Thus, the turning angle of the rotor flux during
the adjustment is
Assuming that the amplitude of the rotor flux does not change during the adjustment,
the predicted rotor flux vector is obtained by turning the original rotor flux to
the extent of the angle &phgr;, i.e.
This way, the predicted rotor and stator flux vectors are
produced in a simple manner and at a sufficient accuracy so that the predicted torque
can be calculated from them according to equation (2)
When a torque estimate of a future time instant has been
predicted as described above, this estimate is compared with the reference torque
in accordance with the invention. If the result of the comparison is that the predicted
torque is smaller than the reference torque when the flux turns into a positive
direction, or if the predicted torque is greater than the reference torque when
the flux turns into a negative direction, the adjustment should be performed immediately
using the voltage vector used in making the prediction. Using this voltage vector
is the fastest and most optimal way of achieving the reference torque.
Figure 4 is a flow chart of the operation of the method
of the invention. The flow chart is implemented for instance in the fastest time
domain of control. When the method is started, a sector S, in which the stator flux
vector is at the time in question, is defined 40. In the next step, the sector is
checked 41 to see if it is the same as during the previous execution time. If the
sector has changed, a variable Sprev is given 42 the value of the current sector
S. At the same time, a state variable F is updated to the value 0 and the routine
moves to the end of the chart. When the state variable F has the value 1, the beginning
of the flux adjustment is ongoing (the sector has not yet changed). This is checked
in block 43. When the beginning of the flux adjustment is ongoing, the routine moves
to the end of the flow chart. If the sector has not changed (S = Sprev) and the
end of the adjustment is ongoing (F = 0), a prediction of the time instant in question
is calculated 44 for the torque. The calculation of the prediction is performed
as earlier described according to the equations (3) to (10).
Whether a new reference voltage needs to be implemented
is defined 45 on the basis of the defined torque prediction, reference torque and
stator flux turning direction. The definition is done by simply calculating the
difference between the prediction and the reference and multiplying this difference
by the angular speed of the stator. If this product is smaller than zero, the voltage
vector is changed 46, otherwise the routine moves to the end of the chart. When
changing the voltage vector, the state variable F is given 47 the value 1 as an
indication for the next round that an adjustment is being made, but the changing
point of the sector has not yet been reached.
Figure 2 shows how the actual torque resonates once as
the flux passes through the sector when using the method of the invention. The control
method is, however, stable, because if for some reason the actual torque after the
adjustment is smaller than predicted, the next adjustment starts earlier and causes
a reduction in the stator flux, and the difference between the actual torque and
the reference torque does not grow any more. This is based on the feature of the
invention that the control method continuously calculates the most optimum adjustment
instant. The system may have a permanent torque error due to torque prediction,
but the torque control is still capable of accurately following the reference torque.
A possible permanent error in the torque control should in any case be compensated
using an integrator, for instance.
In the manner shown in Figure 1, in partial field weakening,
there are points between adjustments, in which the stator flux follows the flux
circle defined by the reference flux. This occurs when the stator flux reaches the
reference value before the start of the next adjustment. When operating in full
field weakening, the stator flux touches the circle of the reference flux only at
points parallel to the voltage vectors, and the flux becomes hexagonal. The same
control principle works in both full and partial field weakening, since the predicted
torque equals the reference value after each adjustment. If the actual torque tries
to fall below the reference value, the same also occurs to the predicted torque,
in which case the control system of the invention automatically selects larger adjustments.
It is obvious to a person skilled in the art that while
technology advances, the basic idea of the invention can be implemented in many
different ways. The invention and its embodiments are thus not restricted to the
examples described above, but can vary within the scope of the claims.