Environmental Engineering Reference
In-Depth Information
The training sets are used to calculate the required statistical quantities and the
new reference persistence model is then applied to the test data sets. The resulting
error measures are plotted against look-ahead time in Figure 6.6. By comparing
with the previous result for the basic persistence forecast model in Figure 6.3,
which also used the same six-month set of measured test data, it is clear that the
performance at short look-ahead times has been maintained, while the poor per-
formance of the basic persistence model at longer look-ahead times has been
improved upon considerably. It is also clear that there is a substantial improvement
Single wind farm
0.45
0.40
0.35
0.30
0.25
0.20
0.15
MAE
RMSE
SDE
0.10
0.05
0.00
0
6
12
18
24
30
36
42
48
(a)
Look-ahead time (h)
15 geographically dispersed wind farms
0.45
0.40
0.35
0.30
0.25
0.20
0.15
MAE
RMSE
SDE
0.10
0.05
0.00
0
6
12
18
24
30
36
42
48
(b)
Look-ahead time (h)
Figure 6.6
Error measures in per unit of installed capacity versus look-ahead
time in hours for (a) single wind farm and for (b) 15 wind farms over a
six-month test period using the new reference persistence model
Search WWH ::
Custom Search