Overloaded US power grid: AI is becoming a necessity for management, IEEE launches training program

According to the US Department of Energy, the US power grid is operating at the limits of its capacity. A combination of rapid growth in industrial consumption, more frequent extreme weather events and a record increase in electricity consumption is straining infrastructure designed decades ago for a more predictable world with centralized generation from coal- or gas-fired power plants. Data centers for AI and cloud computing account for a substantial share of new demand – according to a CNBC report, the largest transmission utility in Texas has recorded 220 gigawatts of new connection requests, largely due to these facilities.

Alongside growing demand, more intermittent renewable sources such as wind and solar are being added to the grid, which, according to the source, creates a volatile environment where supply and demand must be balanced in real time to prevent outages. The grid also faces physical threats (extreme weather, such as the Texas freeze crisis or heat waves that overload transformers) and digital risks associated with the transition to smart meters and control systems, which are more vulnerable to cyberattacks. According to the source, organizations concerned with grid reliability that conduct security exercises such as GridEx therefore emphasize the need for a smarter, more agile and automated grid.

According to energy experts cited in the source, deploying AI to manage power systems has become an operational necessity rather than an experimental technology – traditional planning methods are described as too slow to balance volatile renewable energy in real time. According to a study by McKinsey & Co., integrating advanced data and automation into infrastructure networks can reduce the number of system design errors, reduce equipment downtime through predictive maintenance by up to 50 percent and extend the lifespan of energy equipment by up to 40 percent.

To bridge the gap between AI research and its practical deployment, IEEE, in collaboration with IEEE Power & Energy Society, launched the online course Artificial Intelligence for Power and Energy Systems, aimed at power engineers, utility managers and data scientists. The program was created by Fangxing "Fran" Li from the University of Tennessee in Knoxville and consists of five modules: AI fundamentals for grids, accelerating grid control through reinforcement learning, forecasting and data analytics, physics-informed and safe AI, and generative AI and emerging technologies. According to the source, the course emphasizes safety, asset protection and strict reliability standards rather than uncontrolled AI deployment. Details about the final module are missing from the source article.