AI & Analytics
Anomaly detection that connects alerts to action
Multi-algorithm anomaly detection for industrial assets, process equipment, and utility systems. TwinEdge connects abnormal signals to asset context, governed recommendations, work orders, field evidence, and replayable audit trails.
Downtime reduction proof
Anomalies matter when they turn into the right operational response.
TwinEdge connects anomaly detection to DataOps context, digital twin state, governed agents, AssetOps EAM, Field execution, API/MCP products, and audit evidence so teams do not stop at another alarm.
5
Detection methods
Statistical, ML, spectral, sequence, and correlation patterns.
Context
Before alert
Signals are interpreted with asset, process, history, and rule context.
Work
After insight
Recommendations can flow into EAM, Field, reports, and APIs.
Detection Algorithms
Five algorithms running in parallel, each with different strengths.
| Algorithm | Type | Strength | Latency |
|---|---|---|---|
| Isolation Forest | Unsupervised | Catches multi-dimensional outliers across correlated sensor groups | <50ms |
| Autoencoder | Deep Learning | Learns complex normal operating patterns; detects subtle deviations | <80ms |
| Statistical Process Control | Statistical | CUSUM and EWMA charts for gradual drift detection over time | <5ms |
| One-Class SVM | Semi-Supervised | Effective with limited training data; good for rare equipment types | <30ms |
| Spectral Residual | Frequency Domain | Detects periodic anomalies and unexpected frequency components | <20ms |
Detection Timeline
Core Capabilities
Earlier Warning Window
Detect changes before fixed threshold alarms where signal quality, history, and deployment conditions support early detection.
Cross-Sensor Correlation
Analyze relationships between vibration, temperature, pressure, and flow simultaneously. Catch issues no single sensor reveals.
Adaptive Baselines
Models retrain continuously on recent data. Seasonal changes, load variations, and process shifts are learned automatically.
Consensus Scoring
Multiple algorithms and operating context can be used before an anomaly becomes a recommendation, work draft, or alert.
Turn abnormal behavior into a response your team can trust.
Hours of early warning instead of seconds. Multi-algorithm consensus instead of single-threshold guessing.