AFR Winner

Employees: 500+

To build the Pole Top Fire model, we used Western Power data that analyses the likelihood of equipment failure on our network and pairs it with data from external weather sources. This helps identify the level of pollution and moisture that has naturally fallen on pole tops during various weather conditions over time. A range of predictive modelling techniques were tested, and the popular machine learning technique, Random Forest, was chosen to leverage the data and predict a fire risk. The model is automated to make it easy to use and analyse, with information displayed visually as a colour-coded dashboard. The interactive nature of the dashboard allows our crews and scheduling teams to engage with and explore the findings of the model. The model runs four times a day and automatically alerts staff when there are increased risks of fire.

Since launching in November 2018, our business has acted on data in real time, opening up new doors for how we manage the electricity network.Traditionally, we would need a heavier capital expenditure program to carry out preventative maintenance, but through this innovation, we have found new leading indicators that engineers are also starting to investigate.