Using it artificial intelligence trying DADDI limit the problem of theft of electricity from the network, the costs of which are borne by other consumers who pay the cost of the stolen energy, as well as by the Operator itself.

DEDDIE recently organized an innovation competition aimed at developing algorithmic machine learning models for detecting rheumatism, in which 267 teams (companies and individuals) participated. The winners were announced last week: the teams AETHON Engineering, Silentech and Dataminds shared prizes totaling 30,000 euros.

Competitors attempted to discover services where electricity is (or is highly likely to be) stolen by examining a large amount of data provided by the Operator to participants for domestic and commercial use, in relation to:

1. Supply details such as their electrical power, their geographic location and their use.

2. Details of the supplier

3. Requests submitted to DEDDIE for change of supplier, interruption of electricity due to debt, etc. and dates of submission of requests.

4. Level of consumption and dates of receipt of indications.

5. Data to identify the fact of theft of electricity and the date of detection

Giorgos Lukos, director of DEDDIE Research and Innovation, explained to APE that the request was to develop an algorithmic model that would identify services where there is a high likelihood of power theft.

According to the results, the models developed by the winners of the competition had a success rate of more than 70%, that is, consumers identified as potential electricity thieves, in fact, in this percentage were “arrested” by DEDDIE services for stealing electricity. An evaluation element for selecting the best offers was also to identify services where electricity theft does not occur.

At a later stage, the best-performing models will be tested on larger population samples to confirm their robustness. The ultimate goal of DEDDIE is to identify, with the help of artificial intelligence, objects where there is a high probability of theft of electricity, so that more targeted and effective control can be carried out accordingly.

According to the Regulator, electricity theft in recent years corresponds to 4.5% of annual energy consumption, while before the financial crisis it was 1-2%. In addition, in the conditions of the energy crisis, the cost of “lost” energy has risen sharply. The agency has set specific targets to reduce rheumatic theft with a goal of eliminating it by 2031.

As Mr. Loukos points out, the HEDNO Datathon competition was the first Greek sandbox action presented by DEDDIE in the energy sector (that is, a limited controlled environment in which an innovative application can be tested), namely, as he emphasizes, an important tool for attracting and accelerating innovation.

The next area in which the Administrator will try to try out innovative solutions is the creation of a digital model of the distribution network in which they can, for example. to “run” supply and demand scenarios to identify solutions to potential problems. For example, how can cuts in renewable energy production be minimized during periods of low demand, a challenge increasingly faced by grids as green generation increases.

^{Source: APE-MEB}