Information Technology Reference
In-Depth Information
10.6 Application
Having explained our approach, we are now in the position to present SOE, a
single-agent heating control system, that proposes optimized heating schedules that
aim to reduce the residential energy consumption. SOE computes the optimized
heating schedules based on manual adjustments of the residents and automatically
determined occupancy states. In addition, SOE enables the residents to monitor and
control their heating from remote using mobile devices.
In order to build a practical application we embedded the implementation of our
trained appliance model in our SOE agent using the Matlab to Java compiler. 1 The
SOE agent [ 15 ] is responsible for the heating control in a home and has access to
the aggregated energy signal using Smart Message Language (SML) 2 and the Multi
Utility Communication (MUC)(see footnote 2) interface.
Figure 10.4 illustrates the overall architecture of a SOE agent [ 15 ]. The resi-
dents are enabled to adjust the thermostat and create heating schedules for each
Residents
SOE Algorithms
adjust
Thermostat
create
Heating Schedule
optimize
Heating Schedule
Estimation
of Presence
Inferred Appliance Use
Mo
Tue
Wed
Thu
Fri
Sat
Sun
0 am
6 am
12 am
6 pm
Disaggregation
of Energy
Aggregated Power Consumption
Sleep
Breakfast
Work
Evening
[present]
[present]
[away]
[present]
2 kW
1 kW
Power
0 am
6 am
12 am
6 pm
Fig. 10.4 Overall architecture of SOE heating control system. The aggregated energy signal is
disaggregated using a Naive Bayes classifier to infer appliance usages. From such usage the occu-
pants presence is estimated and used to optimize the heating schedule. The whole system can be
controlled by the residents using tablets and smartphones
1 www.mathworks.de/products/javabuilder/ .
 
Search WWH ::




Custom Search