AustraliaManufacturing2024 – ongoingRio Tinto
ReconAI locomotive fault classification
Rio Tinto reported 96% production accuracy for AI classification of locomotive faults
Problem
Engineers manually classified equipment faults using telemetry and maintenance notes. Breakdowns on Rio Tinto's remote autonomous railway could require staff to recover trains.
Approach
Rio Tinto built ReconAI to classify faults and help prioritize corrective work. It combined maintenance history with retrieved engineering documents and called Claude models through Amazon Bedrock. Engineers could inspect the retrieved material and classification justifications.
Results
Takeaway
Rio Tinto improved classifications by changing models and adding engineering documentation where errors revealed gaps. It did not disclose a quantified reduction in locomotive downtime.
Categories
- Size
- Large enterprise
- Industry
- Manufacturing
- Business function
- Production and quality
- Innovation type
- Process innovation
- AI technology
- Generative AI
- Adoption
- Built in-house
- Evidence
- Company's own claim
Sources
Source summaries
iTnews reported Rio Tinto's account of ReconAI's production deployment. The company reported 96% classification accuracy after iterative changes to models and supporting data.
Full details are in the original sources.
