Samruk-Energy Raises Renewable Generation Forecast Accuracy To 86%

(MENAFN- Trend News Agency) BAKU, Azerbaijan, October 2. Samruk-Energy has increased the accuracy of renewable power generation forecasts to 86% through an artificial intelligence-based system, Head of the Digitalization and AI Office Aizek Zhupankhan said at AI & Digital Bridge 2026 in Astana.

“Considering the growth in the share of renewable energy sources - if I am not mistaken, from 7% to 15% by 2035 - high-quality forecasting is becoming more important than ever, since the balance of the entire energy system depends on it. At Samruk-Energy, we managed to increase forecast accuracy to 86%. The project is currently in pilot operation, after which we plan to scale it to all our renewable energy facilities,” Zhupankhan said.

He noted that artificial intelligence is also being used by Samruk-Energy for predictive diagnostics, computer vision, digital twins and other applications aimed at improving the efficiency and reliability of power generation.

Furthermore, Zhupankhan added that at Ekibastuz GRES-1, Kazakhstan's largest power plant, the company is connecting 56,000 sensors that generate almost 81 million signals per day. The data is analyzed by an AI system to identify developing equipment defects before failures occur, while a digital twin helps determine optimal operating modes and maintenance timing.

Samruk-Energy is also using AI-powered video analytics at its main facilities. According to Zhupankhan, the system analyzes data from thousands of cameras in real time and detects safety violations, including the absence of helmets and personal protective equipment, access to hazardous areas, falls, smoke, and fires.

Following the deployment of the system, the company reduced the number of safety violations at its main facilities by 87%, he said.

The company is also using drones for aerial monitoring, including automated measurement of coal stocks at its thermal power plants. The process, which previously took up to 10 days, can now be completed in no more than 20 minutes, Zhupankhan said.

MENAFN02102026000187011040ID1111750310