P r o m o t i n g E n e r g y C o n s c i o u s B e h a v i o u r
173
days. There is a peak around 7:00am; we can derive the family gets up at 7:00am. Then
the consumption starts decreasing till 4:00pm. At 6:00pm, the electricity and water
consumptions dramatically increase and reach the highest level of the day at 11:00pm.
The consumptions then decrease. It is worth noting that there is relatively high usage
from 8:00am -10:00am. Especially the peak is at 9:00am during the working days. We
may derive their working time may be after 10:00pm rather than normal working time
9:00am. During the public holidays, the energy usage of electricity, water and gas
between 11:00am-15:00pm is higher than other periods. The electricity usage keeps the
high level during the public holidays.
a) b)
Figure 5. The correlation of energy usage patterns in hours
The prediction of energy consumption based on the real usage
It is very important and useful to provide feedbacks to the occupants about the prediction
of future energy usage. We have used Artificial Neural Networks (ANN) (German and
Gahegan 1996) and produced the model for the prediction of energy consumption based
on the current usage. We showed the prediction for working days and public holidays.
For the purpose of the illustration, we only showed the results for Monday and Sunday,
which is shown in Figure 6 a, b and c, and Figure 7a, b and c. Note that we can easily
generate the prediction for other working days and holidays based on Neural Networks.
For prediction on Mondays, in total 43 weeks, we selected 28 week-Mondays as a
training dataset and used 13 week-Mondays as a testing set (excluding 3 week-
Mondays, which are public holidays). Figure 6 a, b and c show the prediction results for
electricity, water and gas consumption. The prediction results are fitted very well,
comparing with the real data.
With respect to the prediction for Sundays, similarly, we used 30 week-Sundays as a
training set and 13 week-Sundays as a testing set. The prediction results for electricity,
gas and water consumption are shown in Figure 7a, b and c. The prediction results tally
with the real data.
0
5
10
15
20
25
0
0.5
1
1.5
2
2.5
Hours
relative mean value
working day
electricity
water
gas
0
5
10
15
20
25
0
0.5
1
1.5
2
2.5
3
Hours
relative mean value
Holiday and weekend
electricity
water
gas
