Environmental Engineering Reference
In-Depth Information
The salient feature of the fuzzy logic efficiency optimizer is that two inputs are
essential, last-change-in-power and last-change-in-
I
ds
. The last change in flux
command,
I
ds
, can be either positive or negative, while the last-change-in-power
can be described with more resolution, in this case by seven membership functions.
The change in power is related to change-in-flux, depending on whether the last-
change-in-flux was negative or positive as tabulated in Table 7.2.
Table 7.2 Hybrid M/G efficiency optimizer based on fuzzy logic
Last-change-in-power
Last-change-in-
I
ds
if negative
Last-change-in-
I
ds
if positive
NB
NB
PB
NM
NM
PM
NS
NS
PS
ZE
ZE
ZE
PS
PS
NS
PM
PM
NM
PB
PB
NB
The output of the fuzzy logic algorithm is a signal for the next change in
I
ds
as
shown (applied to M/G vector current controller). The change in flux command
resulting from the fuzzy rule set is then added to the previous flux command level
to become the new command to the M/G.
With the link power and efficiency optimization based on fuzzy rule set, the
M/G efficiency is globally optimized regardless of loss partitioning or operating
temperature. The optimum efficiency of the M/G can be predicted mathematically
by solving the machine model, in this case an IM, for torque and voltage for the
given speed. Then the M/G efficiency can be expressed as shown in (7.12), where
P
fe
0
is the no load core loss and
P
fv
0
is the friction and windage loss at speed
n
0
:
m
ð
n
p
=
30
Þ
ð
2p
f
=
P
Þ
m
þ
P
fe
0
ð
E
s
=
E
0
Þ
h
¼
ð
7
:
12
Þ
2
3
þ
3
I
s
R
s
þ
P
fv
0
ð
n
=
n
0
Þ
where
f
represents electrical frequency,
P
is the number of poles,
E
s
is the voltage
across machine core (e.g. the magnetizing branch of the single phase equivalent
circuit),
n
is the speed in rpm and
I
s
is the stator applied current magnitude. If (7.12)
were differentiated with respect to frequency,
f
, to find the maximum point it would
also be necessary to find the derivatives of
E
s
and
I
s
since these are also functions
of frequency. This rather convoluted approach can be circumvented by simply
sweeping the frequency in (7.12) and solving for
V
,
E
s
,
I
s
and
n
at each frequency.
Having the maximum value of efficiency from this procedure, it is then a simple
matter to set the vector current controller voltage
V
and frequency
f
accordingly.
The previous discussion is presented to illustrate the computationally intensive
algorithm that would be required to develop an efficiency optimized M/G drive
system online and in real time. The fuzzy logic algorithm requires more sensor
inputs, but it is much more computationally efficient. To summarize the necessary
