LS Electric is accelerating its power brokerage business after developing technology that predicts renewable energy generation using artificial intelligence.
The company announced Wednesday that it has developed an "AI-based wind power generation prediction model" and recently completed field testing. Wind power is among the most difficult renewable energy sources to forecast accurately due to large output fluctuations caused by weather variables.
When applied to a wind farm in Jeju late last year, the technology achieved approximately 92% prediction accuracy even in November, a month marked by high weather volatility. This reduced the typical prediction error rate from 10% to 8%—a level of accuracy that qualifies power generators for incentives under the Korea Power Exchange's "Renewable Energy Generation Forecasting System."
LS Electric succeeded in precisely predicting wind power output changes through a "dual prediction structure." This system uses machine learning and deep learning algorithms to integrate analysis of extensive weather data, regional characteristics such as terrain, altitude, and slope, as well as performance variations among individual turbines based on their specific attributes.
The company plans to leverage this technology to expand into power brokerage businesses including Virtual Power Plants (VPP).
A VPP is a system that uses digital technology to operate physically dispersed renewable energy sources as a single integrated power plant. It uses AI to predict generation from distributed resources and optimize supply strategies, offsetting the variability inherent in renewable energy.
Advancing AI-powered generation forecasting accuracy has become essential in the power brokerage business, as renewable energy operators receive additional compensation for lower prediction error rates.
When renewable energy operators participate in a VPP, they forecast solar and wind generation one day ahead under the "Renewable Energy Generation Forecasting System." If the error rate between predicted and actual generation meets the threshold the following day, they receive settlement payments.
"Through our advanced wind power generation prediction technology, we expect to achieve three benefits at once: solving the intermittency problem of renewable energy, contributing to the stabilization and efficiency of the domestic power grid, and maximizing profits for renewable energy generators," an LS Electric official said.