The Relynk Platform

Historian ingestion, 90-day baseline anomaly detection, and CMMS work order push. Three layers that work together without replacing the SCADA, historians, or maintenance systems you already run.

Relynk platform architecture
LAYER 1

Historian & Sensor Data Ingestion

Relynk ingests tag history and live data from OSIsoft PI, AVEVA Wonderware Historian, GE Proficy, and Ignition without custom middleware development. For assets without a central historian, the ingestion layer connects directly to PLCs via standard industrial protocols. Handles time-series normalisation and backfill for intermittent connections.

OPC UA

Native OPC UA client for direct server connection. Certificate-based authentication, TLS-encrypted sessions. Used by Ignition, AVEVA, and WinCC connections.

Modbus TCP

Polling-based Modbus TCP collection for PLCs and RTUs not covered by OPC UA. Configurable scan rates from 100ms. EtherNet/IP and Profinet also supported.

MQTT Broker

Built-in MQTT broker supports IIoT edge devices and Sparkplug B payloads. TLS 1.3 required.

Historian REST APIs

OSIsoft PI Web API, AVEVA Wonderware REST, and GE Proficy connector for historian pull. Inbound REST API also available for custom data feeds.

Sensor data ingestion
LAYER 2

90-Day Baseline Anomaly Detection

Relynk builds a rolling 90-day statistical baseline per sensor tag using historian data. A sliding-window comparison runs continuously against live readings for vibration, temperature, pressure, and flow signals. When a sensor drifts outside its tolerance band, the system generates an alert with the sensor ID, deviation magnitude, and trend context.

No rule-writing required. Baselines update automatically as equipment behaviour shifts seasonally or after planned maintenance. Target lead time before fault threshold: 4–72 hours.

  • Z-score deviation with configurable sigma thresholds (default: 2.5)
  • Rate-of-change detection for rapid escalation events
  • Correlated anomaly grouping across related sensors
  • Alert de-duplication with 15-minute suppression window
  • Severity classification: warning / critical / emergency
Anomaly detection
LAYER 3

SCADA & Historian Integration

Relynk reads from existing SCADA historians and systems via read-only connections. No modification to your control architecture. No new hardware on the plant floor. Your OT security team reviews one outbound read connection — that is the entire exposure surface.

OSIsoft PI (AVEVA PI)PI Web API + AF SDK
AVEVA Wonderware HistorianREST API, read-only
GE Proficy HistorianNative connector
Ignition by Inductive AutomationBuilt-in OPC UA server
Custom / other SCADAVia OPC UA or REST
SCADA integration

Work order integration

A drift alert that stays in a dashboard doesn't fix anything. Relynk pushes a structured work order to SAP PM, IBM Maximo, or Fiix the moment an anomaly clears threshold — with sensor readings, trend history, and asset location pre-filled.

SAP Plant Maintenance

Relynk creates PM notifications and orders via SAP RFC. Equipment number, functional location, and sensor readings are mapped automatically. Tested against SAP S/4HANA and SAP ECC deployments.

IBM Maximo

Work orders created via Maximo REST API with asset ID, location, and anomaly score pre-filled. Supports on-premise Maximo and Maximo Application Suite (MAS) cloud deployments.

Fiix CMMS

Fiix integration via REST API. Work order created with Relynk anomaly score, sensor reading trend, and asset reference. Supports Fiix cloud and on-premise deployments.

How a typical deployment goes

From first historian connection to production anomaly routing in under a week, for most facilities. Paper mills and chemical plants running OSIsoft PI or AVEVA typically complete onboarding in two to three days.

1

Connect your historian or PLCs

Point Relynk at your OSIsoft PI, AVEVA Wonderware, GE Proficy, or Ignition instance. For sites without a historian, direct PLC connection via Modbus TCP, OPC UA, EtherNet/IP, or Profinet. Read-only credentials only. No agent on your control systems.

2

90-day baseline builds from historical data

Relynk imports up to 90 days of historian data and builds a rolling statistical baseline per tag. You review the sensor list, flag any tags to exclude, and confirm the baseline window. Takes a few hours, runs automatically.

3

Anomalies route to your team and CMMS

Configure alert routing by asset type, severity, and shift schedule. Anomalies that clear threshold push a work order to SAP PM, Maximo, or Fiix with sensor context attached. Alert notifications via email, SMS, Slack, or mobile push.

Security and compliance

Sensor data from SCADA historians stays inside your network. Relynk's edge gateway runs in your OT DMZ, pushes outbound only, and requires no inbound cloud connections to the plant floor.

ISO 27001

Certified

TLS 1.3

All data in transit

AES-256

Data at rest

Pen Tested

Annual third-party

See it on your historian data

Send a tag export from your OSIsoft PI, AVEVA Wonderware, GE Proficy, or Ignition historian. We'll run the 90-day baseline analysis before the demo and show you what Relynk would have flagged in your own data.

Request a Demo