Massive batteries

A massive new solar generation and battery storage facility in Southern California's Mojave Desert is employing machine learning (ML) models powered by Amazon Web Services (AWS) is helping to predict when and how the project’s battery unit should charge and discharge energy back to the grid. The Baldy Mesa facility, operated by AES, will be managed by software using Amazon's SageMaker machine learning service, analyzing more than 33 billion data points annually. The software will use this data to decide when to buy, store, or discharge energy to the grid. In pilots, this system was able to anticipate a statewide heat wave and help stabilize the grid by delivering stored energy. As a major customer, Amazon will draw on the zero-emission electricity produced to power fulfillment centers, office buildings, and data centers. And as more data on the factors that govern supply and demand are collected over time, newer and more sophisticated AI models can be deployed as a software-only upgrade that improves performance and effective capacity in existing facilities.

This development points towards a future where intelligent microgrids work with regional grids that are also becoming more resilient and intelligent, creating multiple layers of resilience and enabling innovations for microgrids.