How AI is changing the understanding of solar energy – a conversation with Todor Georgiev, CEO of Smart Energy Group

2026-02-193 min.

Change language: Albania Bulgaria N. Macedonia Serbia

Just a couple of years ago, cloudiness has been defined as a “problem” for solar energy – inaccurate forecasts, low efficiency and lost yield. Today however, the sector undergoes an interesting transformation during the prism of the other type of clouds – so called Cloud solutions. The advent of artificial intelligence in this recently purely physical industry, changes radically the focus on the number of panels and their position relevant to the precision of the data. To what extent do algorithms manage to balance the energy system, what is different in the management of photovoltaic centers and what are the next steps towards the symbiosis between artificial intelligence and the sun in the RES sector? We discuss the topic with Mr. Todor Georgiev, CEO of Smart Energy Group.

In the solar industry, traditional SCADA systems (monitoring and collecting data) have reached their limit. The problem with them is that they function only as a reactive instrument - they identify events and archive deviations without offering contextual interpretation. The new paradigm is predicative or the transition to systems, based on machine learning, which besides collecting data, achieve a kind of understanding (Deep Understanding) of the whole production process. "We are moving from observation to understanding the entire system. Until now, forecasts were made based on historical data and general meteorology. With artificial intelligence, they become dynamic and allow power plants to be much more optimized and efficient, which is of great importance for their financing and profitability," explains Todor Georgiev.

One of the most serious economic effects because of AI is most noticeable with the control of tracking systems. Standard software for these systems usually works in a static astronomical algorithm. It is “blind” regarding the actual atmospheric conditions and specific sightings on the terrain. However, AI can recalculate the angle of the panels every 5 to 15 minutes, also analyzing micro shading. “This software optimization together with the integration of AI can lead to between 8% and 10% higher annual yield,” the expert says.