See the warning.
Plan the maintenance.
Detect developing equipment problems with physics-based edge monitoring. Turn changes in machine behavior into earlier maintenance decisions.
Local analysis. Critical equipment. Earlier insight.
Rotary asset dynamics
RSST methodology · Gear Technology India, July 2025.
Acoustic anomaly detection
Provisional application filed for decentralized detection.
Analysis at the edge
Physics-based feature extraction on ESP32 hardware.
Multi-node collaboration
Field validation of decentralized anomaly consensus.
The cost of not knowing early enough
Current approaches to condition monitoring often fall short, resulting in expensive unplanned downtime and catastrophic failure.
Calendar-based maintenance
Replacing parts on schedule regardless of condition.
MTBF assumptions
Statistical averages that don't reflect individual machine behavior.
Delayed inspection
Waiting for scheduled downtime to discover damage already done.
Cloud-only monitoring
Latency, connectivity gaps, and single points of failure in remote/marine environments.
How Pred-D works
Sense
Acoustic, vibration, and thermal signals acquired at the machine.
Interpret locally
Physics-based feature extraction on the edge node.
Compare behavior
Node-to-node collaboration detects deviation from healthy baselines.
Surface a signal
Actionable maintenance insight delivered before failure.
Published research
Random Shock Signal Testing for Rotary Asset Dynamics
RSST isolates fault-indicative transient events in non-stationary rotary signals. Validated under laboratory and controlled test-rig conditions; field generalization is ongoing.
Read the methodology & limitationsWhere it matters most
Critical assets in three demanding operating environments. Applications are proposed pilot scenarios.
Commercial Marine
Main propulsion bearings and auxiliary turbochargers on merchant vessels. Problem: in-voyage failure means emergency drydocking.
Explore application
Slurry Pipelines
High-velocity mineral slurry transport. Problem: hydro-erosion thins pipe walls silently until rupture.
Explore application
Power Generation
Supercritical boiler feedwater pumps. Problem: seal and bearing degradation under extreme temperature/pressure.
Explore applicationCommon questions
Does it work with existing sensors?
Designed to integrate with standard industrial vibration, acoustic, and temperature sensors. Can also deploy dedicated sensor nodes where needed.
Does it require constant internet connectivity?
No. Edge processing means each node operates autonomously. Connectivity is used for fleet-level insights and reporting, not for core detection.
How is it installed?
Sensor nodes mount directly on or near the monitored asset. Baseline collection typically takes 2–4 weeks of normal operation.
What happens when data is ambiguous?
The system surfaces uncertainty explicitly rather than suppressing it. Operators receive confidence-qualified alerts, not binary pass/fail.
Have a critical asset you need to evaluate?
Tell us about your equipment and operating environment. We'll assess whether Pred-D can help.