IBM’s Machine Learning Increases Solar Forecast Accuracy
There has been a recent surge in renewable energy investments from big name companies. Google, Microsoft, and Facebook, owners of some of the biggest data centers in the world have all invested millions of dollars in renewable energy to convert their data centers. The main types of renewable energy they have invested in are wind and solar power. Both of which, are limited by fluctuating weather patterns.
Weather can often be unpredictable, unfortunately this can cause an issue for companies that need a way to use wind and solar renewable energy. Now, thanks to Big Data and IBM’s machine learning, there is better analysis for solar and weather predictions. Since many companies are now choosing the green data center over traditional, this is exactly the type of technology they need to make things work.
On July 16, 2015, IBM research shared a video that explained how it had co-developed wind and solar forecasts using Machine learning and Big Data. The results showed that the forecast was 30% more accurate than previous forecasting technologies. This advancement in technology is highly beneficial, especially since there was a SunShot Vision Study which predicted that by 2050 27% of US electricity will come from solar power.
This greater accuracy is due to the fact that IBM Research had been able to collect an enormous amount of data. There have also been more than a dozen researchers collaborating on this project with the Department of Energy for more than 2 years, in order to develop this new technology. The data collected for this project came from local weather stations, sensor networks, cloud movement being tracked by satellites and cameras, as well as historical weather records going back decades. This information is then plugged in and the system will continuously track weather conditions as they change.
Throughout the next year, the main focus will be using solar forecasting technology to optimize renewable energy producer operations. Meanwhile, power companies all around the world are slowly investing in solar capacity at a steady pace. Back in 2013, solar was one of the most used source of renewable energy, only second to natural gas. The only reason solar and wind power hasn’t taken over is because it was difficult to predict, but IBM’s machine learning may have changed all of that.