U.S. Department of Energy (DOE) engineers will be among the first groups to test a new open standard smart grid chip based on machine learning software developed by Utilidata in collaboration with chip designer Nvidia.
The software-defined smart grid chip will be tested for use in commercialization of smart meters for real-time management of distributed energy resources, initially from consumers and eventually utilities.
Some of the smart meters could be deployed to individual homes or apartments to control and optimize the flow of power between home solar, battery and electric vehicle systems. Eventually the meters, or at least the underlying software and hardware, could be extended to use in power transformers, substations and transmission grid lines, an Nvidia spokesperson told Fierce Electronics.
Existing energy meters don’t have the computational power or communications capabilities for such real-time operations, said Marissa Hummon, chief technology officer at Utilidata.
Utilidata makes real-time grid software that is being combined with Nvidia’s Jetson edge AI platform and its AI Enterprise software to create the smart grid chip. Jetson modules have already been tested or used by 850,000 developers, according to Nvidia. In simple terms, Jetson is a series of embedded computing boards that run a Tegra processor integrated with an Arm CPU to provide low power capabilities for acceleration of applications including machine learning.
Engineers in DOE’s National Renewable Energy Laboratory (NREL) will conduct the actual testing of the smart grid chip in an effort to scale and commercialize Real-Time Optimal Power Flow (RT-OPF) technology. RT-OPF will be used for power load control at a local level for Distributed Energy Resources (DERs), which include solar and wind power.
The smart grid chip can be embedded in future smart meters to potentially enable wider adoption and commercialization of the technology. The move could “redefine the role of edge computing for DER integration and resilience” to benefit end customers and utilities, said Santosh Veda, group manager for automation and controls at NREL, in a statement.
Nvidia believes the collaboration with Utilidata and NREL will bring AI to grid edge operations to allow consumers to see reduced energy consumption and costs. Because the system will be open and software-defined, utilities and consumers will “no longer be locked into closed, proprietary systems that hamper innovation by limiting third-party applications,” said Marc Spieler, head of global energy business development at Nvidia.
DOE’s Solar Energy Technologies Office Technology Commercialization Fund has granted Nvidia, Utilidata and NREL $250,000 for the smart grid chip effort. The fund was created as part of a set of initiatives to foster partnerships among DOE and private companies.