Taiwan Water Corporation
AI acoustic diagnosis of underground pipe leaks
Taiwan Water's AI leak-diagnosis pilot achieved peak accuracy above 98% in excavation-validated cases, according to ITRI
Problem
Leak detection depended on experienced workers listening for faults in aging underground pipes. Limited inspection capacity delayed identification and repair.
Approach
Taiwan Water worked with the Industrial Technology Research Institute on an AI listening model combined with acoustic analysis and signal separation. The partners collected cases and checked diagnoses through excavation. Wider field testing remained a next step.
Results
Takeaway
The reported accuracy came from collected cases checked through excavation. Performance across a wider network remained unreported.
Categories
- Size
- Large enterprise
- Industry
- Energy and utilities
- Business function
- Production and quality
- Innovation type
- Process innovation
- AI technology
- Predictive analytics and ML
- Adoption
- Partner co-development
- Evidence
- Vendor case study
Sources
Source summaries
ITRI described its joint leak-diagnosis work with Taiwan Water and reported peak accuracy exceeding 98% after case collection and excavation checks. It identified wider field testing as future work.
Full details are in the original sources.