Predictive Maintenance for Automotive Stamping
AI-powered asset health monitoring for a Tier-1 automotive components supplier
A Tier-1 automotive stamping supplier was experiencing 6–8 unplanned press stoppages per month, each costing an average of $28,000 in lost production, tooling damage, and emergency labor. Maintenance was entirely time-based with no condition intelligence.
Peunier deployed an AI-powered predictive maintenance platform monitoring 14 high-value stamping presses—using vibration, acoustic, and current signature analysis to predict failures 24–72 hours in advance.
Failure Mode Analysis
Reviewed 18 months of maintenance records to identify failure patterns and select optimal sensor modalities for each press type.
Sensor Network
Deployed 96 sensors across 14 presses capturing vibration, acoustic emission, motor current, temperature, and tonnage data at 4kHz sampling.
ML Model Training
Trained anomaly detection and remaining useful life prediction models using historical run-to-failure data and domain knowledge.
Maintenance Workflow Integration
Integrated predictive alerts with CMMS for automated work order generation, eliminating manual interpretation step.
The system paid for itself in 9 weeks. We went from 7 unplanned press stoppages a month to barely one. The AI model accuracy genuinely surprised our maintenance engineering team.