In many cases, however, smart meters are used merely as replacements for older analogue or digital electricity meters. Swiss electricity supply regulations envisage that 80 percent of all measuring points will be equipped with a smart meter by the end of 2027. We are thus faced with a situation in which extensive data is recorded and gathered without there always being clarity about what should or could be done with it. Aside from issuing an electricity bill.
Other industries are leading the way in this field. For example, credit card institutions have been able to predict consumers’ purchasing behaviour for many years now. Shoppers at Coop and Migros enthusiastically collect points, allowing both companies to analyse purchasing behaviour and target advertising to the individual consumer. The list is almost endless and also includes mobile phone providers and social media companies.
Where data is available, it can be evaluated – also smart meter data. Here at the iHomeLab we are applying cutting-edge analysis methods in a range of research projects, with the aim of deriving higher quality information from energy consumption data. Using breakdowns of this data we can, for example, identify when which consumers are present at a company and how much energy they use. Taking historical data as our starting point we can forecast future electricity consumption or offer advice on optimizing energy use. Alternatively, we can determine when the occupant of an apartment has retired from work.
What all of these examples have in common is that the data must be analysed in an appropriate manner. Recent developments have shown that algorithms from the fields of artificial intelligence or machine learning are promising options for this purpose. We too are using this technology, implementing recognition algorithms as so-called «deep neural networks». These are algorithms which are based on the way the human brain functions but are, however, unable to fully reproduce these functions.
The results gained from this approach for energy consumption breakdowns are good. They are not, however, limited to such applications. It is, for example, a promising option for determining changes in behaviour. This can be identified through the use of energy consumption data to draw conclusions about the activities which the occupants of a building are engaged in. Generally speaking, these activities will follow a certain pattern, which can vary from person to person or building to building. In all cases, however, a deviation from the normal pattern can indicate an unusual situation – e.g. that someone has fallen and is unable to get up. A very topical theme is whether this procedure can provide indications regarding the onset of dementia.
Such evaluations provide wide-ranging data about people, making compliance with data protection regulations a serious issue. Our key concern is to make people’s needs the focus of our activities. Data analysis should provide consumers with added value and everyone should be able to decide which data can be passed on and which not. Each individual must have the certainty that the data is protected from abuse.