Ausgrid: Infrastructural health assessment using AI

Challenge

Ausgrid has 5 million timber poles in service across Australia, accounting for 83% of the power and communication network infrastructure. Traditional methods for assessing pole integrity involve visual inspection and sounding tests, which can be inconclusive and inaccurate as they depend on an individual inspector’s interpretation of the information.

Annual spend on maintenance and asset management is approximately $40 million, and 80% of poles removed from service in Eastern Australia were still in good condition, resulting in significant operational and cost inefficiency.

Solution

The UTS Tech Lab team developed a novel screening prototype that comprises a sensor array (iBar) and AI technology.  The test is non-destructive and takes one person with a device only a few minutes to assess an entire pole and collect and upload field data to the cloud. The technology combines advanced sensing array, in-depth wave propagation theory, advanced signal processing and machine learning (AI) to identify the condition of the pole.

From this concept, two tools were developed:

1. A screening tool to determine the go/no-go for pole replacement.
   Mature and currently deployed in field

2. A diagnostic tool to precisely identify damage location, type and severity that leads to accurate evaluation of the remaining strength (safety margin).
   Conceptually proofed but under further development

Testing equipment includes an innovative iBar (integrated sensing, data acquisition and CPU system), a modal hammer and a smart phone.

The UTS team worked with Ausgrid to train contractors and staff, who implemented the field testing. The trial tested 10,000 timber utility poles in 2.5 months with an overall success rate of >88%. The contractors demonstrated improved efficiency in the field, with a record of 130 poles tested by one inspector.

The project has increased the safety and reliability of testing, while simultaneously reducing costs and labour. In addition, the database accumulated from historic information will improve testing accuracy and safety in the future.