Essential Energy using AI to improve safety

Artificial intelligence innovation turns field notes into safety intelligence

Essential Energy’s peer-reviewed innovation set to improve frontline productivity and strengthen network safety

An algorithm that can rapidly analyse notes made by employees to proactively detect and assess network safety insights to identify and prioritise material risks shows how Essential Energy is positioning itself among leading organisations in the energy industry using artificial intelligence (AI) to solve real-world problems.

Designed by the organisation’s data science team, the program can swiftly sift through thousands of unstructured text notes from field staff, classifying, and analysing the unstructured data and building patterns to support safer outcomes for network safety.

 According to Essential Energy’s Principal Data Scientist, Andrew Slack-Smith, the output of the program is designed to improve the accuracy and reliability of the information used to manage the electricity network.

“The work our team has done will make a huge difference because it automates the first pass of analysis from the field and creates actionable insights for human review. What they have created is important because it will enhance our ability to identify safety risks,” Andrew says.

The current rules-based checks rely on predefined pass/fail criteria and identify network safety data successfully 59 per cent of the time. By contrast, the new algorithm will lift this to 76 per cent, reducing the time people spend manually sifting through the data.

The algorithm is currently being further refined and tested before being rolled out.

The project’s success is underscored by its selection for peer-reviewed publication and presentation at the prestigious AMPEAK conference this week (April 12-15).

“In an area like AI, where there’s a lot of hype, being peer-reviewed and accepted into AMPEAK is a significant achievement for our data team and a celebration of their ingenuity and hard work,” Andrew says. “We have been leaders in innovation for some time, but we are integrating AI where it makes sense and adds value, not just because it is AI.”

While the algorithm will be used to analyse network safety data for the time being, Andrew expects his team’s creation will quickly expand into other areas of the network.

 “What we’ve created is an algorithm that can create structure out of unstructured data, from people writing notes in their iPads when they are out in the field. It is potentially scalable to hundreds of other data sources. All we need to do is find the data source we want to analyse and run an algorithm over it,” Andrew says.

Like many successful innovations, this solution evolved from a slightly different initiative. The original plan had been to create a program to analyse the structured data the field crews enter into the system when in the field, recording issues with infrastructure. “We were going through all the structured data when we realised that all the juicy data was under the comments section in the system, which isn’t that unusual. It’s where people were writing the notes that really make a difference because there were real insights,” Andrew says.

He sees a day where field crews and AI are working together to better identify network safety issues.

“The next step at some stage is going to be speech to text where crews in the field will be able to simply talk about what they are seeing and AI will not only help build out the full picture of the issue by using its knowledge of our assets, but also give them three succinct paragraphs that will fill in any gaps for them,” Andrew says.