Industrial sites, such as oil rigs and power plants, require extensive monitoring for efficiency and safety because liquid, steam, or oil leakages can be catastrophic, costly, and wasteful. Global energy leaders, are using AI and machine learning to deliver a path to autonomous power plants. The company trains AI models using thousands of images and video streams from millions of onsite cameras and sensors to detect process anomalies.

These models are deployed at the edge in power plants and use real-time inferencing to identify leaks. Rig operators are using computer vision, deep learning, and intelligent video analytics (IVA) to monitor heavy machinery, detect potential hazards, and alert workers in real-time to protect their health and safety, prevent accidents, and assign repair technicians for maintenance.

Oil and gas enterprises rely on finding the best-fit routes to transfer oil to refineries and eventually fuel stations. Edge AI can calculate the optimal flow of oil to ensure reliability of production and protect long-term pipeline health. Using IVA, these companies can inspect pipelines for defects that could lead to dangerous failures and automatically alert pipeline operators. Further downstream, NVIDIA ReOpt uses GPU-accelerated solvers for logistics and route optimization, which can efficiently route fuel to fueling stations.

With proactive maintenance, utilities can accurately detect defects and reduce unplanned outages to better serve customers. FirstEnergy worked with Noteworthy AI, a NVIDIA Inception member, on a pilot project to automate utility pole inspections. Fixed camera systems powered by NVIDIA Jetson were secured to the roof of service trucks and collected standardized, high-resolution images of their utility poles, power lines, and pole-mounted assets.

The images were analyzed at the edge to determine if repairs or vegetation management was needed. Edge computing can help monitor the estimated 185 million utility poles in the United States, and reduce the tens of millions of dollars spent each year by utilities to manually track and maintain poles.

Intelligent forecasting using GPU-accelerated grid simulations combined with historical data on energy usage and weather can inform more efficient generation, distribution, and management of energy resources to consumers. AI helps manage the bidirectional flow of power in a grid, delivering reliable energy to residents and enterprises while automating the process for consumers to sell their additional energy back to the grid.