Utility infrastructure monitoring and leakage detection
DataHex and anomaly-detection ML, IoT sensor platform
Challenge
Ten percent of water lost a year
A large water and waste network in New South Wales, running a distributed legacy SCADA network, was assessing a new geo-tagged, low-powered IoT sensor network, and needed a scalable near-real-time way to detect and locate leakages and major bursts, which typically account for around 10% of water loss a year.
Solution
Machine learning on pressure and flow
An anomaly-detection machine-learning algorithm to identify potential leakage from changes in pressure and flow; a platform to ingest sensor readings, invoke the algorithm, and present IoT sensor status on a map of the pipe network on AWS.
Visualization tools show historical trends and network status, with healthy sensors green and leakages red over the customer’s infrastructure map.
What DataHex delivered
- Anomaly-detection machine learning Identifying potential leakage from changes in pressure and flow.
- Sensor ingestion platform on AWS Readings ingested, the algorithm invoked, and IoT sensor status presented on a map of the pipe network.
- Visualization of network status Historical trends with healthy sensors green and leakages red over the customer’s infrastructure map.
Results
End to end in six weeks
- An end-to-end proof of concept from ingestion to machine learning to visualization delivered in six weeks
- The proof of concept surfaced issues in the existing sensor network, in message time-stamping and equipment, that were critical for a larger-scale rollout
- Business case support for a wider rollout
Talk to us
Ready to build your own data marketplace?
We will walk your team through what a launch looks like, from catalog and licensing to delivery and managed service.