Machine Learning’s Impact on Solar Energy

 

Wed, 07/29/2015 - 1:00pm Lindsay Hock, Editor IBM's solar forecasting technology. Image: IBM Research In 2013, solar was the second-largest source of new electricity generating capacity in the U.S., exceeded only by natural gas. A USA SunShot Vision Study suggests solar power could provide as much as 14% of U.S. electricity demand by 2030, and 27% by 2050. There are currently two main customers for renewable energy forecasting technologies: utility companies and independent system operators (ISOs). However, the difficulty in producing accurate solar and wind forecasts has required electric utilities to hold higher amounts of energy reserves as compared to conventional energy sources. Yet, solar power installations grow each day, and future solar penetration levels require increased attention to the value of more accurate solar forecasting. With better solar and wind forecasts, it’s possible that solar energy’s contribution to the U.S.’s energy will reach up to 50%. Until now, due to intermittency, solar energy won’t supply more than 20 to 30% of the U.S.’s energy. However, a collaboration between IBM and the U.S. Dept. of Energy (DOE) could double the accuracy of solar and wind forecasts within the next year with the help of IBM Research’s machine learning technology. “This collaboration could have a huge impact on the energy industry, as well as local businesses, the economy and the natural environment,” says Hendrik Hamann, Physical Analytics Manager, IBM Research. “Part of our goal is to help a wide range of industries and professions better understand how the world works so we can all make better decisions.” Announced on July 16, 2015, IBM Research revealed that solar and wind forecasts it’s producing using machine learning and other cognitive computing technologies proved 30% more accurate than ones created using convention approaches. The research program was funded by the DOE’s SunShot Initiative, and the results suggest new ways to optimize solar resources as they are increasingly integrated into the nation’s energy system. Machine learning for solar improvements So how does machine learning compare to conventional approaches for solar and wind forecasts? Think of it as big data meets science, says Hamann. While other solar forecasting systems take more narrow location and timeframe views, IBM’s approach incorporates a great number of weather and solar energy prediction models. “We use machine learning techniques to blend those using historical data as a function of weather situation, forecast horizon and location to create what we call a supermodel,” says Hamann. These advances are valuable for the future of alternative energy, and the machine learning model can generate accurate forecasts of solar energy from minutes ahead to several days. IBM Research worked with academic, government and industry partners for about three years to develop this self-learning weather model renewable forecasting technology, otherwise known as SMT. This technology uses a combination of machine learning, big data analytics and mathematical modeling of complex weather systems to continuously analyze, learn from and improve solar forecasts derived from a large number of weather models, including satellite observations, sensor networks and local weather stations and sky cameras. The system analyzes the data to forecast how much solar energy will be available at different locations and times. IBM’s approach provides a general platform for renewable energy forecasting, including wind and hydro. The SunShot Initiative “IBM’s goal is to help produce a more sustainable energy future by integrating solar power into the energy pipeline,” says Hamann. “Our collaboration with the DOE’s SunShot Initiative now makes this possible as the collaborative effort seeks to make solar energy fully cost-competitive with traditional energy sources before the end of 2020.” Through the DOE SunShot Initiative, IBM Research is working with a number of collaborators in academia, government and industry to grain various perspectives. “For example, National Renewable Energy Laboratory (NREL) and ISO-New England are two main partners who are helping evaluate the important metrics to determine accuracy of solar forecasting,” says Hamann. IBM Research is also providing foundational solar forecasts covering all 48 contiguous states at 5-km spatial resolution. This is primarily for government agencies, utilities and grid operators to evaluate how solar forecasting can impact the supply and demand, in addition to operations. source: http://www.rdmag.com/articles/2015/07/machine-learnings-impact-solar-energy