Fine-tuning predictions with AI
Farseer and AMART are part of AES' larger collaborative effort to expand its investments in renewable energy. For example, the company has acquired six wind farms from Valcour Wind Energy in New York, and is building and operating solar farms and storage systems in California, Arizona and other U.S. states, Brazil and Argentina. These investments come just as the company predicted in its 2023 annual report that demand from U.S. enterprise data centers will nearly double within the next three years as generative AI deployments expand.
But weather uncertainty poses a big challenge for AES as it adds renewable energy to its portfolio. Farseer's machine learning models represent a major advancement for the company because they analyze vast amounts of historical weather data to predict wind farm output with much greater accuracy than ever before, Reyes said.
“As we iterated with Farseer, we were able to improve the accuracy of the model because it was updated daily with reference to history,” Reyes points out. “We'd suggest what we'd like to generate the next day, and we'd get the actual value, which would then be fed into the history to continue to further refine what the model would suggest the next day.”
