Information Technology Reference
In-Depth Information
Table 3 Comparision of two rule extraction algorithms
Algorithms
FRENGA
FGS
TFGS
Operation
Iteration
50
100
65
Mutation probability
0.7
0.6
0.7
Cross over probability
0.4
0.7
0.6
Population size
65
60
65
Membership size
7
7
7
Time required to
Rule extraction
302.646775
501.212821
Rule extraction
time = 302.646775
Tuning
time = 459.895189
Total = 762.541964
Lowest error in the Rule extraction
0.1226
0.1445
0.0615
turbine control system in Senjyu et al. ( 2006 ). Figure 16 d, indicates output power of
this strategy in 180 s. The System response has sudden change, and hence is
sensitive to wind turbulence extremely. This method has saved power in same
places, but it has severe change when power reduces in long time that is an effect of
the wind downfall. Figure 16 e shows that proposed MLP algorithm has produced
controlled power with fewer drops than output power and torque of LQG algorithm
(Endusa and Aki 2009 ), for example, between 45 and 70 s. In those situations that
wind speed reduces intensely, output power downfall for a while. Note that this
event is unavoidable.
Figure 16 f, g indicates that FRENGA and GFS have better effect on output
power of wind turbine than MLP algorithm.
Figure 16 h shows hybrid controller output power that proposed in this chapter.
According to this
figure, proposed hybrid algorithms have produced power with
fewer drops than output power of Fig. 16 b
d, for example, between 45 and 70 s. In
addition hybrid method has better respond than FRENGA and MLPNN alone.
As a result in Fig. 16 i TFGS has improved responses better of them. Thus TFGS
has best response and controls turbine very well.
In the power curve, power reduction in some spots is distinguished that it
-
is result
of wind sudden reduction, but proposed algorithms with shift pitch angle almost
control it. Also FRENGA is compared by Fuzzy Genetic system without Neural
Network and TFGS in Table 3 . Despite the same and rules, TFGS extracts rules
with great accuracy and speed.
'
Search WWH ::




Custom Search