6 ways AI technology is revolutionising our energy networks

Transitioning our networks to maximise on clean and renewable energy brings with it all kinds of new and unexpected challenges. This means we must constantly look to more innovative and imaginative solutions to keep the networks resilient, operating safely and efficiently as we journey towards net zero.

Here we look at some of the innovative and state-of-the-art artificial intelligence (AI) technologies we’re trialling or adopting to solve issues and evolve our energy systems.

1. Autonomous drones: performing pilot-free infrastructure inspections

Autonomous drones can fly ‘beyond visual line of sight’ to inspect overhead wires, pylons and substations without the intervention of an on-site pilot. They can be remotely accessed from anywhere in the world, so they can be automatically deployed, recovered and recharged.

These drones allow us to look at parts of our network that are more difficult or costly to reach, making inspections much easier and removing the need for power outages, scaffolding or cherry pickers.

A cloud-based AI model will then process the data and images collected by the drone, automatically detecting defects and producing near real-time asset condition reports. The drones also use AI to deal with unforeseen or unpredictable situations.

Our autonomous drone projects include:

Visual Inspection and Condition Assessment Platform (VICAP)

The drones gather high-definition close-up images of pylons, which are then processed using AI to indicate the health of the steelwork. The forthcoming VICAP 2 project will also use a refined AI model to predict future asset condition and make reports on recommendations for painting or replacement of steelwork.

Autonomous Aerial Thermal Inspection of Substations (AATIS)

This project is investigating how we can use drones and AI to automatically monitor the thermal condition of our substation assets. This will aid in maintaining the health of an ageing asset base efficiently and economically, plus improve understanding of the condition of our assets and failure modes.

> Read more about why we use drones

2. AiDash: satellite technology supporting vegetation management and biodiversity

The AiDash system provides both Intelligent Vegetation Management and Intelligent Sustainability Management. The system uses satellite imagery to locate problems on assets like power lines, coupled with an AI dashboard that makes predictions about which areas need immediate attention.

The satellite imaging can capture a top-down picture of the entire network, as well as imagery based on different angles.

Intelligent Vegetation Management

Near real-time imagery of tree and shrub conditions along the networks helps us to predict the probability of overgrown vegetation causing an electric circuit to malfunction or experience an outage. AiDash's Intelligent Vegetation Management technology has been applied to National Grid’s entire US business and a year after starting to use the information gained from the proof of concept we realised approximately $2 million in efficiencies.

Intelligent Sustainability Management

Using the same technology, Intelligent Sustainability Management can map the land type and measure the biodiversity level of habitats that our networks pass through, providing an alternative solution to on-the-ground surveying methods. In the UK, we’re in the early stages of using the Intelligent Sustainability Management technology to help us plan biodiversity enhancements and track our progress against our sustainability commitments.

> Read more about AiDash

3. Exodigo: mapping underground infrastructure without excavation

Exodigo’s non-intrusive subsurface imaging platform helps to map underground infrastructure by providing a digital, geolocated 3D map of buried assets. Combining AI with multi-sensor fusion dramatically improves accuracy and time to map, which reduces the costs associated with unnecessary excavation.

As part of efforts to increase grid resilience across our territories in the US, we planned to expand a substation in Rotterdam, New York. However, the surrounding area had no available records of what lay underground. To mitigate the risk of damaging any buried pipes or cables, we engaged Exodigo in late 2023 to locate all such objects in and around the substation and identify any risk areas.

Having completed the scanning the data was analysed and processed with AI and turned into a visual map of underground assets. Using the completed asset map, our NY electric team found 20 previously unknown electricity lines buried near the substation, helping them avoid project delays due to unforeseen excavation strikes.

> Read more about Exodigo

4. OceanBrain: Monitoring cables buried under the sea

With energy security more important than ever, we needed a robust system to monitor the health of our interconnectors – subsea electricity cables that allow energy to be shared between countries.

There are a number of ways to monitor the health of cables on land, but for those under the seabed it can be more challenging. One potential hazard for submarine cables is fishing trawlers that drag heavy nets and chains across the ocean floor to scoop up fish. While the cables are buried and designed to withstand impact, repeated trawling can increase the odds of strikes – or shift the seabed to leave them exposed. Climate effects can also move the sea floor.

With these complications in mind, National Grid Partners developed a digital tool called OceanBrain to modernise how undersea cables are monitored. It combines complex data sources (including cable location, burial depth and seabed type) with machine learning to automatically quantify the risk of potential damage.

> Read more about Oceanbrain

5. Helicopter LiDAR data: a birds-eye view for effective network management

The helicopter unit for our distribution network uses Light Detection and Ranging (LiDAR) data to comprehensively survey network assets and any adjacent vegetation. This makes the process of identifying potential issues that overgrown vegetation may cause – such as malfunctions or outages in electric circuits – much more efficient.

Previously, vegetation management in these areas relied on labour-intensive foot patrols to identify overgrowth, followed by manual dispatch of cutting teams – a time-consuming task given the extensive distribution network spanning hundreds of thousands of kilometres.

From reactive to predictive network management

Using LiDAR sensors on-board our helicopters has allowed us to transition from reactive to predictive vegetation and network management. The insights from LiDAR data, coupled with advancements in machine learning, enable us to forecast the future state of both the network and surrounding vegetation.

This proactive approach means our local teams can strategically prioritise tasks to manage assets and vegetation within their respective regions, as well as empowering us pre-emptively address potential issues, ensuring uninterrupted service for customers.

6. Cathodic Pipe Protection: machine learning that reduces compliance risks

This predictive model uses machine learning to estimate the probability that each test point in our underground US gas network would fail an inspection.

Built on decades of historical data, Cathodic Pipe Protection (CPP) detects past patterns to help us anticipate future work. The insights from machine learning present a reliable picture of pipeline health and allow our teams to prioritise work and keep gas flowing safely to customers’ homes and businesses.

Machine learning predicts issues before they can happen

Corrosion in pipework can present risks in both safety and reliability. However newly paved roads in Massachusetts are under 'guarantee' for five years, which prohibits digging and therefore prevents us from addressing compliance issues. Being able to predict when and where test points would fail allows us to guard against potential issues before the paving begins.

There is also substantial financial risk associated with failed test points – the possibility of fines for each failed inspection without intervention – which is reduced by CPP.

Backlogs reduced by 50%

In the first six months of use, we’ve leveraged the results of CPP to reduce the backlog of problem test points by 50% on Guaranteed Streets in Massachusetts. The Corrosion Control team has also used CPP to begin issuing proactive maintenance work orders for the first time, getting ahead of problems before they arise.

Collaborating with AI technology portfolio companies

Using AI technology is already bringing a variety of positive benefits that aid in the transition to clean energy; enabling us to collect better data at lower cost and with less environmental impact, as well as helping to make maintenance activities even safer and more efficient.

Many of the technologies listed above are portfolio companies of National Grid Partners (NGP), our Silicon Valley-based corporate venture capital and innovation group.

> Find out more about National Grid Partners