Using Artificial Intelligence to "Green" America's Aging Energy Grid

Sean McEvoy, Senior Vice President of Energy, Veritone
Sean McEvoy, Senior Vice President of Energy, Veritone

Sean McEvoy, Senior Vice President of Energy, Veritone

In an increasingly connected world that depends on strong, reliable power sources, utility operators nationwide are scrambling to balance two seemingly incompatible goals.

First, grid operators are responsible for managing electricity supply in response to real-time consumer demand. However, this is incredibly challenging when serving a growing population and never-ending demand for power-hungry gadgets like computers, smartphones, and even electric vehicles (EVs). In fact, the US Energy Information Administration (EIA) predicts that global energy consumption will increase by 50% from 2019 to 2050.

Second, in response to climate change concerns, public demand, and stricter renewable portfolio standards (RPSs), utility operators are also responsible for “greening” the electricity network by investing in more renewables like solar photovoltaic (PV) technology and wind farms.

Unfortunately, a greener grid is inherently less reliable and harder to manage—leading to shortages, surges, outages, and price volatility. This is because renewables like solar and wind are intermittent power sources with unpredictable output. A sudden gust of wind, for example, can cause spikes in energy generation, and passing clouds can cause momentary dips in solar power production.

Worse still, this unpredictability will only become more pronounced for two reasons:

• According to the American Society of Civil Engineers (ASCE), most of the country’s 640,000+ miles of high-voltage transmission lines were installed during the 1950s and 1960s, with much of this infrastructure slated for decommissioning after just 50 years. ASCE also reports that America’s electricity network is already at maximum capacity—making it difficult to safely onboard more power even as energy demand continues to rise.

• A growing number of residential and commercial utility customers are investing in their own PV installations, with the Solar Energy Industries Association (SEIA) reporting that the PV sector has grown by over 40% in just the past decade alone. Similar growth trajectories exist across other green technologies that interact directly with the grid, including on-site battery solutions, electric vehicles, and even smart thermostats.

The First Tentative Steps Toward a Truly “Smart” Energy Grid

To overcome the above challenges, many utilities have turned to battery storage technology, distributed energy resource management systems (DERMSs), or a combination of both:

• Utility-scale batteries allow grid operators to store excess renewable energy when electricity demand is low, and dispatch this stored green power when grid demand goes back up.

• DERMS technology provides utility operators with real-time optimization and control of distributed energy resources like privately-owned PV installations, on-site solar batteries, and EVs. This is accomplished using sensors and receivers on edge devices that can communicate directly with the DERMS platform.

Although the widespread deployment of batteries and traditional DERMS technology are helping to mitigate the worst effects of grid unreliability, real-time coordination simply isn’t possible given the sheer scope of data that utility operators must collect and manage. Even with complete awareness of changing weather patterns, consumer demand, and energy production, the response is always reactive—with utilities forced to constantly course-correct to deliver power that is green, reliable, and affordable.

But of course, human decision-makers will never have a complete awareness of grid conditions. Doing so would require collecting, analyzing, and acting on many terabytes of historic and real-time data at the moment.

Fortunately, parsing Big Data is precisely where artificial intelligence (AI) excels.

An Intelligent Energy Grid Built for the 21st Century

  
​As utility customers increasingly go solar, have families, and buy power-hungry gadgets, our iDERMS solution can use and act on terabytes of historic and real-time data to guarantee greener, cheaper, and more reliable energy delivery—even when dispatching electricity across a rapidly aging grid that was originally designed for centralized power generation and distribution
   

 

At Veritone, we’ve developed the world’s first truly intelligent distributed energy resource management system, iDERMS, to help bring America’s aging grid into the 21st century. Powered by artificial intelligence and machine learning, this comprehensive solution is built around several complementary modules that work together to provide utility operators with unprecedented control over real-time grid management decisions:

• The Forecaster component of our iDERMS platform analyzes both historic and real-time climate, weather, and irradiance data to accurately predict renewable energy output, both for the green assets under a utility’s direct control and also for all the privately-owned distributed energy resources that connect to the grid.

• The Optimizer component uses Forecaster’s predictions to correctly identify the optimal balance of renewable energy generation, storage, and dispatch based on both current needs and on future demand.

• The Controller component uses Optimizer’s instructions to autonomously regulate distributed energy resources and other edge devices, allowing for seamless coordination across the entire utility network, even as grid conditions continue to evolve in real-time.

• The Designer component helps energy professionals plan, design, and analyze prospective green energy investments. Using historic and forecasted energy prices, load and generation profiles, and site-specific factors, Designer seamlessly evaluates thousands of project configurations and hones in on the optimal design and operational strategy to maximize ownership goals while meeting environmental and operational targets.

 Because Veritone’s AI-powered DERMS technology uses continuous machine learning, it’s able to continuously refine its predictions by matching real-time forecasts against actual grid conditions. This leads to improved accuracy over time, not despite the influx of more data, but because of it.

As utility customers increasingly go solar, have families, and buy power-hungry gadgets, our iDERMS solution can use and act on these terabytes of historic and real-time data to guarantee greener, cheaper, and more reliable energy delivery—even when dispatching this electricity across a rapidly aging grid that was originally designed for centralized power generation and distribution.