Predict and prevent machine maintenance events. Ensure equipment uptime and materiel availability through the use of advanced data-driven and physics informed ML/AI. Our capability can be employed in the following frameworks:
- Descriptive (Monitoring): Describes the machine condition based or real and/or virtual sensors. Answers the question: What happened? What is happening?
- Diagnostics: Identifies the reason for a failure or degraded performance. Answers the question: Why did it happen?
- Predictive (Prognostics): Predicts the machine’s future behavior. Answers the question: What is likely to happen?
- Prescriptive: Prescribes how the machine should act based on the expected future behavior. Answers the question: How to Act in Response?
Case Studies
Remaining Useful Life
Thermomechanical Fatigue
Problem: Minimize equipment downtime related to hardware fatigue and failure.
Solution: Through the integration of governing equations into advanced machine learning algorithms, PredictiveIQ gives you full visibility into your asset’s remaining useful life.
Results: PredictiveIQ has reduced hardware product development and sustainment costs by 35%.
This modular, scalable architecture solution:
- Allows a seamless addition of new failure modes.
- Works with any telematics data stream.
- Reduces by 1,000X the amount training data
- Allows for real-time on Edge and on platform training.
Engine Health
Oil Quality & Wear Prediction
Problem: Reduce ground combat vehicles’ oil sample analysis frequency.
Solution: PredictiveIQ combines custom-made virtual sensing and vehicle telematics to provide real-time lube oil health status.
Results: PredictiveIQ has been proven to increase mission-focused equipment sustainment.
This modular, scalable architecture solution:
- Allows a seamless addition of new failure modes.
- Works with any telematics data stream.
- Reduces by 1,000X the amount training data
- Provides real-time monitoring of oil health while occasionally utilizing oil sample results for model calibration.
Vehicle Telematics
Anomaly Detection
Problem: Identify anomalies in engine telemetry data of a passenger vehicle.
Solution: PredictiveIQ developed a highly precise real-time PIML Classifier Software.
Results: Developed PIML model had a precision between 98.4% and 100% in detecting anomalies in incoming engine data.