Artificial intelligence and machine learning are becoming more widely used in predictive maintenance across the power industry as utilities seek to identify equipment deterioration before failures occur, according to research from GlobalData. The technologies are being applied to operational and sensor data to establish normal asset behaviour, detect anomalies and identify emerging failure patterns. GlobalData said their use is expanding across generation, transmission and distribution networks as operators increase the deployment of digital monitoring systems.
In its Strategic Intelligence: Predictive Maintenance in Power (2026) report, GlobalData identified Ørsted, Florida Power & Light and National Grid among companies using AI and machine learning alongside high-frequency sensor data, inspection imagery and historical operating data. These systems are intended to estimate failure probability and support maintenance and outage scheduling.
Predictive maintenance is also being used to manage changing operating conditions associated with renewable generation and shifting power flows, the report said.
Energy tracking is becoming another component of predictive maintenance systems. Rehaan Shiledar, power analyst at GlobalData, said comparing expected and actual energy performance can identify declining equipment performance before an outage or failure.
“By translating technical condition signals into expected energy loss under forecast demand, weather, and dispatch, it sharpens maintenance prioritization around risk-to-deliver and real economic impact,” Shiledar said.
Utilities are incorporating advanced metering infrastructure and grid-sensing data into asset health and reliability programmes. GlobalData cited Duke Energy and Southern California Edison as examples of utilities using load and voltage data to monitor stress on transformers and feeders and inform equipment replacement decisions.
Digital twins and augmented reality are also being combined with predictive maintenance systems. Digital twins provide virtual representations of physical assets that can be updated using live operating data, while augmented reality can display equipment status, diagnostics and maintenance instructions to technicians in the field.
GE Vernova is using digital twins for power generation equipment including turbines and boilers, alongside wearable and immersive systems for field technicians. Siemens is also combining digital twin technology with augmented reality across industrial operations, according to GlobalData.
The report also linked carbon pricing with the economics of predictive maintenance. Equipment degradation can increase fuel consumption, auxiliary loads and energy losses, potentially increasing carbon-related costs in markets where emissions are priced.
Predictive maintenance and digital asset management systems can be used to identify deterioration caused by factors such as fouling, leakage, component wear, control drift and insulation ageing. Avoiding forced outages and restarts can also reduce periods of inefficient operation and the need for higher-emitting backup generation.
GlobalData expects utilities and generators to expand predictive maintenance programmes as they manage ageing assets, operating and maintenance costs and growing renewable generation fleets. Shiledar said industrial internet-of-things sensors, edge computing and analytics are making condition monitoring easier to deploy across larger groups of assets, while remote monitoring is becoming more relevant for distributed wind and solar installations.
