Harnessing visual recognition to help the energy and utilities sector function more efficiently

Share this post:

At IBM Client Engineering, we start by putting ourselves in the shoes of our clients, and then co-create solutions with them. Our recent project with National Gas Metering is a perfect example of this tailored approach. Together, we built an innovative visual inspection solution that matches the accuracy and knowhow of in-house experts, paving the way for smarter decision-making.

Under pressure

As the largest manager of traditional residential, industrial, and commercial gas meters in Great Britain, National Gas Metering is constantly engaged in a large-scale programme of asset upgrades and replacements. To decide where best to allocate its resources, the company wanted to know precisely which types of meter and pressure regulators are installed at customer sites.

National Gas Metering posed a question to IBM: is visual recognition technology mature enough to differentiate between its assets (considering they are usually grey and of a similar size and shape) and, if so, could this technology be of practical use?

Our answer was a resounding yes on both counts, and we set out to prove it.

Teaming up

First, we wanted to see the situation through the eyes of National Gas Metering. We set up a business framing session with key team members and invited them to articulate the details. An opportunity quickly emerged: repurpose an existing data source to create value. Each time a National Gas Metering engineer visits a site to fit a device on a meter, they take a photograph as evidence that they’ve completed the job. These photos are a rich data source that held the potential to address some of the company’s challenges.

In the next few years, National Gas Metering may see many subject matter experts (SMEs), who have deep knowledge of its assets, retire, or leave the business. Without their knowhow, engineers cannot easily differentiate between different asset types, which can lead to inefficient use of time on a job. In fact, one of the SMEs approaching retirement was at our initial business framing session, validating the business case with us there and then.

Our goal was to demonstrate that visual recognition technology can aid the development of an asset library and give National Gas an insight into the assets they have in the field.

Making it work

Next on the agenda was a discovery session, where we analysed the as-is process for distinguishing assets and worked out what the end solution should look like. From this, we targeted a visual inspection solution to be embedded in the company’s wider IBM Maximo asset management toolset.

Armed with a deep understanding of National Gas Metering’s processes and objectives, IBM embarked on a four-week build of a minimum viable product (MVP) solution. We took advantage of IBM’s market leading machine learning and AI capabilities, built into our Maximo Application Suite for Enterprise Asset Management.

In a relatively short time and with limited resources, we trained the model to recognise different pressure regulators and even spot potentially faulty installations. The results spoke for themselves: our solution achieved an accuracy level similar to an experienced SME at National Gas Metering.

If put into production, the solution has the potential to revolutionise decision-making at National Gas Metering, making it easier for the organisation to keep customer sites safe. The insights provided by the asset library will enable the company’s engineers to plan jobs more effectively, and ensure they make the best use of their time. By proactively recognising and responding to defective installations, the solution could also minimise liability and enhance safety.

Here to help

At IBM Client Engineering, we aim to understand our clients’ business first, and fit our technology solutions around their needs. We are always focused on what value means to our clients, and jointly prove out the value of our technologies in addressing some of their biggest opportunities and challenges. We are here to listen first.

To learn more about how Client Engineering can work with your business: join us at the upcoming Open Evening at IBM London, 16 November (link). To learn more about IBM Client Engineering, click here.