AI System in Russia Set to Detect Underreported Electricity Consumption

The analytical center of Setevaya Kompaniya has developed an AI system designed to detect unmetered electricity consumption.

The project Poisk Bezuchetnogo Potrebleniya (Unmetered Consumption Detection) is a web service that analyzes consumption at metering points equipped with smart meters. The AI automatically identifies cases where electricity consumption readings appear intentionally underreported. Such situations can lead not only to financial losses for energy companies but also to risks of exceeding maximum permitted load levels, which in some cases may result in accidents.

The system is based on a time-series classification model that analyzes consumption profiles and identifies anomalies. Data from electricity meters is automatically fed into the service, converted into time series, and classified by a machine-learning model.

Keeping Pace With Technology

The system is currently undergoing testing. Specialists regularly inspect suspicious facilities to refine the model using real-world data. In the future, the developers plan to speed up the service and incorporate external data sources.

The company is also actively using artificial intelligence in other areas. Last year it introduced a corporate GPT-based model called Tsifrovoy Nastavnik (Digital Mentor) to help employees navigate internal processes, a service that transcribes dispatcher communications and detects errors, a system that identifies comparable facilities to accelerate work in capital construction projects, and a tool that automatically generates responses to common consumer inquiries.

Setevaya Kompaniya is also developing a computer-vision system to monitor the use of personal protective equipment in order to improve worker safety.