Energized, a tool to automate the detection of electrical fraud

Energizados, detección de fraudes eléctricos

Did you know that almost 2 million homes in the region could have been supplied with electricity throughout 2019, using the losses generated in the transmission and distribution systems[1]? To reduce these losses, the IDB has developed Energizados, a tool that uses artificial intelligence to automate the detection of electrical fraud.

For electricity to reach cities, communities, and businesses, electrical energy must be transported using transmission grids, transformers, distribution lines, and various equipment. However, there is a percentage that is lost in said transport. These losses are known as “technical losses” and are inherent to the transport of electricity. Additionally, there is another type of energy loss, the “non-technical losses. ” Latin America and the Caribbean loses approximately 15% of the total electricity supply[2].

Non-technical losses are those normally caused by electrical fraud and theft. The first (losses due to theft) is when an individual makes illegal connections to the electricity grid to avoid paying for consumption; the second (fraud losses) occurs when a person manipulates electrical or wiring meters in order to reduce energy consumption readings and pay less.

The consequences of non-technical losses

Consequently, energy theft does not only affect distribution and marketing companies, but also affects the entire community: illegal connections can cause explosions, fires, property damage, electrocutions and even death. In addition, the theft of energy causes direct losses of income and increases the operational costs of distribution, which in general are transferred to the population.

See also  AMP stories: let's discover this marketing tool

For example, in Brazil, it is estimated that in 2017, due to theft, approximately 62 million dollars ceased to be invoiced – only in the concession area of ​​CEEE, a distribution company. Although there is a tendency to believe that the greatest fraud occurs in the residential sector, the reality is that there are more economic losses due to fraud and theft in the commercial area.

Therefore, reducing non-technical losses is key to maintaining the financial sustainability of companies and the proper functioning of the electricity grid, avoiding overloads that can cause blackouts and deterioration.

In recent years, new technologies have emerged that allow tackling the problem of fraud in various sectors. Technologies such as Smart Meters, the Internet of Things (IoT), artificial intelligence (AI) and machine learning (Machine Leaning) make this inconvenience easier to identify and reduce. For example, banks make use of these cutting-edge technologies to detect behaviors that indicate a financial crime. Given the success in this sector, other sectors such as the electrical industry began to use them to identify non-technical losses.

Digital transformation, a pillar of the IDB’s Vision 2025

As part of the Digital Transformation activities under IDB Vision 2025 that the Infrastructure sector has been developing, Energizados was developed together with the energy distribution company CEEE. The goal is to reduce electrical fraud and potential damage to communities. CEEE distributes energy to 72 municipalities and provides electricity to 1.5 million customers in its 47,000 km of urban and rural networks. They came together to create a pilot that uses machine learning to attack the problem of non-technical losses.

See also  Lighthouse: Google SEO audit tool

Energized is an application that uses Machine Learning and statistical methods to help companies reduce fraud and theft in the electrical network. This solution began to be developed at the end of 2020 using data from users’ monthly electricity consumption and other descriptive variables to build prediction models.

How does Energized work?

To verify the assertiveness of the predictions made by Energized these were validated in the field. That is, it was physically inspected that the meters identified by Energized as fraudulent they were actually frauds. In this detection process, Energized It increased the capture of fraudulent consumer units by 5 percentage points, compared to the software currently used by the company, which obtained an accuracy between 17% and 23%. Additionally, this platform helped detect fraud in different areas of the city, while the software used by the company is biased to the same sites.

Energized suggests being promising for the reduction of non-technical losses since it contributes to carrying out inspections more efficiently, which leads to a reduction in their costs and an increase in the assertiveness of fraud detection. Furthermore, with the graphical interface provided to end users, Energized allows to obtain an overview that helps to make more assertive decisions to carry out inspections in the field.

See also  Creating payment channels in Bitcoin is made easier with this new tool

Energized is an application that helps to detect and reduce non-technical losses by reducing regularization times and increasing the precision of fraud identification, so we can say that: Energized is automating electrical fraud detection!

It is important to note that in addition to the use of technology, it is necessary to carry out awareness campaigns about the material and economic damages that this type of problems entails and thus also be able to contribute to the reduction of risks and costs.

Given the benefits that Energizados can provide to reduce the impacts caused by non-technical losses, the implementation of it in different countries of the region is being analyzed. Likewise, an adaptation is beginning to be implemented in the water and sanitation sector. Thus, in several sectors, Energizados is automating the detection of electrical fraud!


[1] Statistics of electricity losses from the OLADE-Sielac database assuming that these losses are reduced to their technical limit. The loss indicator considers both technical and non-technical losses in distribution. For more details of its methodology, consult: http://biblioteca.olade.org/opac-tmpl/Documentos/old0380.pdf

[2] OLADE, 2019.

Leave a Comment

Your email address will not be published. Required fields are marked *