TL;DR

NGED (National Grid Electricity Distribution) is trying out whether drone photography and machine learning can pick out ageing equipment on its network before it fails, The Engineer reported on 22 October. The PMT Scan project, run with data science firm Softwire and Team UAV, a drone operator, covers 500 pole-mounted transformers and roughly 3,000 images.

What the trial involves

Pole-mounted transformers step down the voltage on distribution lines to the level homes and businesses use. NGED wants to know whether automated image analysis could give it more consistent data on corrosion and other visible damage.

Team UAV has photographed each of the 500 transformers six times, across NGED’s licence areas and including rural sites. Softwire’s data scientists will label the pictures with NGED’s own engineers and train models to judge equipment condition. The trial will also look at how light, weather, location and season affect whether the models can spot faults reliably, and Softwire will build a prototype web app to present the findings.

If it works, NGED hopes the method would to help its engineers rank repairs, cut unplanned outages and reduce environmental harm from leaking transformer oil.

What the partners say

NGED innovation engineer Favour Eboh said the transformers play “a critical role in delivering electricity to homes and businesses, particularly in rural communities”. The project, Eboh said, will test whether “aerial imagery, AI, and computer vision” can automate parts of condition assessment and feed better data into asset health reporting and investment decisions.

Mark DeGiovanni, Team UAV’s technical lead, described the work as combining specialist drone operations with Softwire’s AI and software skills.

Looking forward

This is an early-stage trial on a defined set of assets, and NGED has framed it as a test of potential rather than a commitment to roll out. The interesting part is the attention to conditions: a model that works on clear summer images but misses corrosion in winter light would be of little use to a network operator. In our view, it is a good example of the kind of unglamorous AI deployment that matters to UK infrastructure, where the payoff is fewer power cuts in rural areas rather than a new product. The results will show whether drone and model inspections can stand alongside engineers’ own judgement in deciding where maintenance budgets go.