PA Consulting has co-developed iPredict, said to be the world’s first artificially intelligent system to prevent power outages by predicting asset failures weeks in advance. PA co-developed the system with San Diego Gas & Electric Company (SDG&E) and industrial Internet of Things and machine learning experts Toumetis.
By showing which assets are likely to fail within the next two weeks, iPredict lets utilities plan repairs and safely fix issues before a power outage. This allows a utility to reduce customer outages, improve customer satisfaction, reduce overhead and maintenance costs, and improve public and employee safety. Since its development, iPredict has an 80 percent success rate for predicting equipment failures two weeks out, and 90 percent success rate for predictions three days out.
When unplanned outages occur, utilities must make emergency repairs – install new equipment, dispatch crews and control traffic, and coordinate logistics – all with increased time and expense. With iPredict, utilities can predict and plan repairs, reduce customer service disruptions and give customers advanced notice about the planned improvements and outages. By providing advanced warning of disruption, utilities can ensure energy reliability to mission-critical services such as hospitals, fire stations and rescue services.
In real time, iPredict collects and analyses high frequency power quality data to identify the warning signs of an imminent failure of distribution system components. iPredict then applies machine learning techniques and algorithms to combine historical high-frequency sub-cycle data with data from Operational Systems including distribution connectivity models to identify the failures. SDG&E also expects to utilize iPredict for overhead equipment failures.
Gregg Edeson, Reliability Lead at PA Consulting and ReliabilityOne Program Director, said: “As a 40-year veteran of the power industry, I know first-hand the benefits a system like iPredict can bring. iPredict provides utilities with an opportunity to safely fix assets, lower risk for their employees, and minimize disruption for customers. This system is focused on improving reliability, employee safety, and giving utilities a necessary snapshot of the health or degradation of their assets, before the lights go out.”
“We’ve worked with PA on numerous initiatives, from their ReliabilityOne™ program to circuit risk data modeling," Tom Bialek, Chief Engineer at SDG&E, says. "We were eager to co-develop this first-of-its-kind program that could benefit the entire industry by proactively predicting underground and overhead equipment failures.”
PA has been focused on energy reliability for two decades, having created the ReliabilityOne program which recognizes US electric utility reliability performance. PA is an industry leader in utility systems, AI and machine learning, and data analytics and supports US utilities who are seeking ingenious ways to strengthen their power grids to ensure they are resilient, reliable, and safe. iPredict™ is another asset that utilities must have in their toolbox to prevent power outages and provide their customers with the highest levels of reliability.
