Computer Science > Industry > UT Dallas CS Researchers Apply Power of AI To Forecast Energy Supply, Demand
Computer Science > Industry > UT Dallas CS Researchers Apply Power of AI To Forecast Energy Supply, Demand
UT Dallas CS Researchers Apply Power of AI To Forecast Energy Supply, Demand
The increasing use of renewable, fluctuating power sources, such as wind and solar energy, has made it more challenging than ever to forecast electricity prices.
In partnership with Vistra, one of the largest competitive power generators in the U.S., University of Texas at Dallas researchers are using artificial intelligence (AI) techniques to help the Irving-based company make more precise pricing projections.
In machine learning, computers analyze large amounts of data in order to learn to recognize patterns.
“The technologies of machine learning and deep learning provide valuable insights that can help an energy provider make more accurate forecasts that can save money for the company and its customers,” said Dr. Gopal Gupta, co-director of the center, professor of computer science and Erik Jonsson Chair. “Our partnership with Vistra is an example of how academia can work with businesses to solve real-world problems.”
The researchers completed the work in August 2020 and held classes from December through February to train Vistra employees in the background technologies.
“The technologies of machine learning and deep learning provide valuable insights that can help an energy provider make more accurate forecasts that can save money for the company and its customers. Our partnership with Vistra is an example of how academia can work with businesses to solve real-world problems.”
Dr. Gopal Gupta, professor of computer science and Erik Jonsson Chair in the Erik Jonsson School of Engineering and Computer Science
The project, which was funded by Vistra, will help the company project pricing for its Moss Landing Energy Storage Facility in Monterey County, California. The soon-to-be 400-megawatt energy storage system, which is one of the largest of its kind in the world, can power up to 300,000 homes during peak periods. Such battery storage plants offer the opportunity to store energy from renewable sources, such as wind and solar, for later use.
“This project, jointly done by the Center for Applied AI at UT Dallas and Vistra Corporation, was crucial to optimizing electricity pricing as our Moss Landing battery farm came online in early 2021,” said Rachit Gupta, vice president at Vistra and lead sponsor of the project. “The project was a tremendous success, and we are extremely happy that we availed ourselves of a great source of expertise that is present locally.”
Vistra’s Moss Landing Energy Storage Facility in Monterey County, California, stores energy from renewable sources, such as wind and solar, for later use.
Wholesale energy costs fluctuate constantly based on supply and demand. For example, extreme weather conditions, such as the winter storms that Texas experienced in February, can lead to skyrocketing demand and costs. Weather conditions also affect supply — solar panels generate less electricity on cloudy days and nothing overnight, while wind farms often produce most of their energy at night, when electricity demand is the lowest.
“AI can help a company like Vistra forecast future generation and demand on load, wind and solar energy, and optimize bidding, scheduling and deployment of energy to improve profitability and market participation,” said Dr. Feng Chen, associate professor of computer science at UT Dallas and the project’s principal investigator.
The researchers applied statistical and machine-learning methods to build models that can predict near-real-time bid and sell prices. The goal is to help the company buy electricity to charge the batteries at the lowest price and sell at the most opportune time.
In addition to Gupta and Chen, the project involved Dr. Doug DeGroot, CAIML director and professor of instruction in computer science. Other team members included Farhad Shakerin MS’16, PhD’20, now a senior software engineer at Microsoft Corp.; Abhijit Balaji, a computer science graduate student; and computer science doctoral students Yibo Hu MS’17 and Changbin Li. Dr. Anurag Nagar, assistant professor of instruction, and Ashutosh Senapati, a master’s student in computer science, conducted the training.
DeGroot said the center members are proud to have been selected for the project and look forward to teaming up with Vistra and additional industry partners in the future.
“Vistra’s innovative Moss Landing battery energy storage facility will play a critical role in supporting California’s energy grid,” DeGroot said. “Our machine-learning technologies for capturing and supplying that energy have helped optimize how that battery system is used.”
The UT Dallas Computer Science program is one of the largest Computer Science departments in the United States with over 3,315 bachelors-degree students, more than 1,110 master’s students, 165 Ph.D. students, 52 tenure-track faculty members, and 44 full-time senior lecturers, as of Fall 2019. With the University of Texas at Dallas’ unique history of starting as a graduate institution first, the CS Department is built on a legacy of valuing innovative research and providing advanced training for software engineers and computer scientists.
