Optimization of In-House Energy Demand - Smart Information Systems: Computational Intelligence for Real-Life Applications - page 268

the task of the energy disaggregation algorithm would be to recognize and classify
the individual appliances consumption profiles within the aggregated energy signal
coming from the smart meter. Ralph assured Steven that this solution does not require
additional sensor infrastructure and enables recommendations for optimized heating
schedules based on automatically deduced presence states.
Steven was very pleased with Ralph's proposal and informed his wife Suzanne
about the good news. In about three months they would have a new energy-aware
heating control system, which will help them to cut future energy bills without any
effort. Suzanne was relieved by the idea that she will never again have to worry about
the kids going out of the house without turning the heating off or her coming home
early in winter finding the living room in a state of severe cold. Steven told her that
it would even be possible to access and monitor the heating control from remote via
her smart phone, which may be convenient when they are on longer vacation with
their kids. As usual both Steven and Suzanne were enjoying dinner together on their
garden terrace, talking about new family projects and listening to the local radio
station, which this time was broadcasting a debate about environmentally friendly
vehicles and new ways of transportation.
10.1 Introduction
The main goal of our study is to provide a framework for heating control and schedul-
ing which considers the occupancy states of residential homes. Since most solutions
for occupancy state identification involve complex sensor infrastructure and costly
hardware which cause high usage barrier [ 3 , 9 , 14 , 17 ], we aim to use given infor-
mation from available electricity smart meters. We propose to employ energy disag-
gregation to infer appliance usage which is, as we will show, beneficial to occupancy
state identification. In the following, we briefly introduce the value of appliance
usage information, before we explain how we use this information for the purpose
of heating control.
In the context of domestic environments, consumers vastly underestimate the
energy used for heating and overestimate the energy used for appliances that replace
manual labor tasks [ 4 ]. Numerous studies have identified that consumers get a bet-
ter understanding of their energy use by clear, concise, and direct feedback about
appliance-specific consumption information [ 13 , 19 , 23 ].
In regard to power grid operators and power suppliers, knowledge about the energy
consumption on appliance level is critical to the development of power system plan-
ning, load forecasting, billing procedures, and pricing models [ 4 , 19 ]. In addition, the
identification of electric appliances in domestic environments is important, because
the increasing number of renewable energy sources in the power grid requires electric
utilities to be able to quickly react to changes in supply and demand [ 18 ].
The growing need for accurate and specific information about domestic energy
consumption on device level has led to numerous studies on appliance load mon-
itoring [ 1 , 4 , 10 , 22 , 23 ]. Existing solutions for appliance load monitoring can