Integration

AI Integration with Ignition for Supervisory Control

Augment Ignition's control scripting and real-time data fabric with AI-driven decision logic for adaptive setpoint adjustment, multi-variable optimization, and intelligent alarm management without replacing your PLCs.

Get in touch Learn more

Cinematic overhead of a WeWork creative suite room with multiple curved monitors showing AI decision dashboards, executives in casual attire reviewing data, dramatic pendant lighting.

ARCHITECTURE PATTERNS

Where AI Fits into Ignition's Supervisory Control Layer

Ignition's supervisory control layer is the ideal real-time data fabric for injecting AI-driven decision logic without disrupting existing PLC code or operator workflows.

AI models integrate directly with Ignition's tag system and scripting engine. Instead of replacing deterministic control logic, AI acts as a supervisory copilot, analyzing real-time streams from PLCs, sensors, and SQL databases to recommend or trigger adaptive adjustments. Key integration surfaces include:

UDT (User-Defined Type) Tag Values: AI inference outputs can be written to specific tags, making predictions (e.g., predicted quality score, optimal setpoint) available to any Ignition screen, script, or alarm.
Gateway Scripting (Python/Jython): AI model calls are embedded within gateway event scripts (e.g., on tag value change, on timer) to perform inference on incoming data batches and update system states.
Perspective Session Scripts: For operator-facing AI assistants, session scripts call APIs to provide contextual guidance, anomaly explanations, or approval prompts directly within the HMI.

A typical implementation wires an AI service (hosted on-premise or in a private cloud) to Ignition via a REST API or Message Queue. For example, a model predicting equipment failure might consume a window of time-series data from Ignition's Historian or a SQL Bridge query. The resulting health score is written back to a tag, which can trigger a Standard Alarm with a recommended action. For adaptive control, an AI agent analyzing multiple process variables could calculate an optimized setpoint and write it to a tag that is then passed to a PLC via an OPC UA or Modbus write—all within Ignition's secure, logged transaction. This keeps the core control loop intact while adding a layer of intelligent optimization.

Governance and rollout require careful design. AI-driven setpoint changes should typically route through an Approval Tag or require a manual confirmation step in the HMI before being sent to the PLC. All AI inferences, input data, and actions should be logged to a dedicated SQL table via Ignition's scripting for full auditability and model performance tracking. Start with a pilot on a non-critical process line, using Ignition's Redundancy and Alarming systems to monitor the AI service's health. The goal is to augment the operator, not automate them out of the loop—providing data-driven recommendations that reduce cognitive load during alarm floods or complex multi-variable adjustments.

SUPERVISORY CONTROL

Ignition Modules and Surfaces for AI Integration

The Real-Time Data Fabric for AI

Ignition's SCADA engine and tag system form the primary integration surface for AI-driven supervisory control. This is where AI models connect to live IIoT streams.

Key Integration Points:

Tag Historian: Time-series data from PLCs, sensors, and instruments provides the training and inference dataset for predictive models.
UDT (User-Defined Types): Complex data structures representing equipment (e.g., Pump_UDT with status, pressure, temp) allow AI to reason about entire assets, not just individual tags.
Alarm & Event System: AI can prioritize, suppress, or enrich alarms by analyzing the multivariate context of the alarm flood.

AI Use Case: A model consumes real-time pressure, flow, and vibration tags from a Pump_UDT. It predicts bearing failure 72 hours out and writes a predicted_health_score tag. This score triggers a pre-alert in Ignition's alarm system, prompting preventive maintenance.

AUGMENTING REAL-TIME CONTROL LOGIC

High-Value AI Use Cases for Ignition Supervisory Control

Ignition's flexible architecture provides a powerful real-time data fabric for industrial operations. These use cases detail how to augment its supervisory control scripting with AI-driven decision logic, moving from reactive monitoring to adaptive, predictive control.

Adaptive Setpoint Optimization

Inject AI models into Ignition's control scripting to dynamically adjust PID loops and setpoints based on real-time process conditions, material properties, and predicted outcomes. This moves control from static recipes to context-aware optimization, improving yield and consistency in batch or continuous processes.

Batch -> Adaptive

Control Paradigm

Intelligent Alarm Flood Management

Use AI to analyze and prioritize Ignition's alarm streams, suppressing consequential alarms and highlighting root-cause events. This reduces operator cognitive load during upsets and can be integrated with Ignition's alarm notification system to route actionable alerts to the right personnel via SMS or email.

Hours -> Minutes

Mean Time to Diagnose

Predictive Quality Gate Control

Connect AI inference to Ignition's HMI or Perspective modules to create intelligent quality gates. Based on upstream sensor fusion (vision, vibration, temperature), the system can predict quality deviations and automatically route material, hold a batch, or adjust downstream parameters before defects occur.

Inspection -> Prevention

Quality Focus

Multi-Variable Constraint-Based Scheduling

Leverage Ignition's SQL Tag Historian and real-time equipment states to feed an AI scheduler. This model evaluates energy cost, maintenance windows, material availability, and skill sets to recommend or automatically execute optimal production sequences through Ignition's order management modules.

1-2 Day Lookahead

Schedule Horizon

Anomaly Detection for Unplanned Downtime

Deploy unsupervised learning models on Ignition's IIoT time-series data to detect subtle equipment anomalies weeks before failure. Integrate findings with Ignition's reporting or directly trigger work orders in a connected CMMS like Maximo, turning unplanned stops into planned maintenance.

Weeks of Lead Time

Failure Prediction

Operator Copilot for Complex Procedures

Build a conversational AI assistant embedded in Ignition Perspective HMIs. Using context from active alarms, running batches, and equipment states, it provides operators with step-by-step troubleshooting guidance, fetches relevant SOPs from connected systems, and logs actions for continuous improvement.

Same-Shift Resolution

Complex Issue Handling

IGNITION INTEGRATION PATTERNS

Example AI-Enhanced Supervisory Workflows

These workflows illustrate how AI models can be embedded into Ignition's supervisory control layer to augment operator decisions, automate routine adjustments, and manage complex alarm scenarios. Each pattern leverages Ignition's scripting engine, tag system, and database connectivity as the integration fabric.

Trigger: A new batch production order is released in the ERP and synchronized to Ignition via its database bridge.

Context Pulled: The Ignition script retrieves:

The target product recipe and its nominal setpoints (temperature, pressure, agitation).
Historical quality data (yield, purity) for the last 20 similar batches from the SQL historian.
Current raw material lot properties (viscosity, potency) from the LIMS integration tag.
Real-time ambient conditions (cooling water temperature) from PLC tags.

AI Agent Action: A lightweight regression model (deployed as a containerized service) is called via a REST script function. It receives the context and returns optimized setpoint adjustments (e.g., +2.5°C phase 1 temperature, -5% agitation rate) to maximize predicted yield while staying within safety envelopes.

System Update: The Ignition script writes the adjusted setpoints to a dedicated AI_Recommended_Setpoints tag group. A supervisory screen prompts the operator to review and apply. Upon approval, the script pushes the values to the PLC control loops via Ignition's OPC UA or driver tags.

Human Review Point: The operator must acknowledge the AI recommendations on the HMI before they are sent to control. All recommendations, operator actions, and resulting batch KPIs are logged to a dedicated AI_Decision_Audit table for model retraining.

FROM REAL-TIME IIOT TO ADAPTIVE CONTROL

Implementation Architecture: Data Flow and Model Orchestration

A practical blueprint for injecting AI inference into Ignition's supervisory control layer to enable adaptive setpoint adjustment and intelligent alarm management.

The core integration pattern treats Ignition as a real-time data fabric and execution engine. AI models are deployed as external services that subscribe to Ignition's tag change events and alarm streams via its REST API or MQTT connector. For control optimization, a lightweight model ingests a window of time-series data from key process tags (e.g., temperatures, pressures, flow rates, motor speeds) and returns optimized setpoint suggestions. These suggestions are written back to Ignition as new tag values, which then propagate to PLCs via Ignition's native OPC UA or driver-based communication. This creates a closed-loop where AI provides recommendations, but final execution and safety interlocks remain under Ignition's deterministic control scripting.

For alarm flood management, the architecture introduces an AI inference layer between Ignition's alarm generation and the operator HMI. Alarm events are streamed to a model that performs real-time clustering and root-cause analysis, using historical alarm logs and current process state. The model returns a prioritized, condensed alert—for example, 'Primary root cause: Pump P-101 cavitation'—which is displayed in a dedicated AI Insights panel within an Ignition Perspective module. This reduces cognitive load by suppressing consequential alarms and providing immediate diagnostic guidance, all while maintaining the full, auditable alarm log in Ignition's historian.

Rollout is typically phased, starting with a read-only monitoring agent that provides recommendations without writing back to control tags. Governance is enforced through Ignition's built-in audit trails and scripting security, ensuring all AI-suggested setpoint changes require operator acknowledgment or follow a defined approval workflow in the HMI. Model performance is monitored by logging inference results alongside actual process outcomes in Ignition's SQL database, creating a feedback loop for retraining. For a deeper dive on connecting AI to industrial data sources, see our guide on IIoT Data Pipelines for AI.

IGNITION INTEGRATION PATTERNS

Code and Payload Examples

Real-Time AI Inference on Tag Values

Ignition's tag system is the primary data fabric. AI models can subscribe to tag value changes for real-time inference. A common pattern is to use a Python script within an Ignition Scripting Function to call an external AI service when a tag value updates, passing a window of historical data for context.

This is ideal for adaptive setpoint adjustment. For example, an AI model analyzing temperature, pressure, and flow rates can calculate an optimal new setpoint. The script updates a target tag, which is then written back to the PLC via the OPC-UA or Modbus driver.

python
# Example: Ignition Python Script for Setpoint Optimization
def optimize_setpoint(current_temp, current_pressure, historical_window):
    """Calls an external AI model API."""
    import requests
    import json
    
    payload = {
        "features": {
            "current_temp": current_temp,
            "current_pressure": current_pressure,
            "historical_trend": historical_window
        }
    }
    
    response = requests.post(
        "https://your-ai-service/infer/optimize-setpoint",
        json=payload,
        timeout=2.0
    )
    
    if response.status_code == 200:
        result = response.json()
        return result.get("recommended_setpoint")
    else:
        # Fallback logic or raise alarm
        return current_temp  # Default to current value

AI-AUGMENTED SUPERVISORY CONTROL

Realistic Operational Impact and Time Savings

This table shows how augmenting Ignition's scripting and control logic with AI-driven decision models creates tangible operational improvements without replacing core PLCs or existing HMI investments.

Control Workflow	Before AI (Manual/Static)	After AI (Adaptive/Assisted)	Implementation Notes
Alarm Flood Management	Manual prioritization by operator during high-volume events	AI-suppressed nuisance alarms, root-cause grouping	Uses Ignition's alarm journal and tag history to train models; reduces cognitive load by 40-60%
Multi-Variable Setpoint Adjustment	Weekly/Monthly review by process engineers; static recipes	Real-time, constrained optimization suggestions every 15-30 minutes	AI model ingests Ignition tag data, suggests optimal setpoints; engineer approves via HMI button
Batch Recipe Parameter Optimization	Trial-and-error adjustments over multiple batches	Predictive parameter tuning for next batch based on incoming material properties	Connects Ignition SQL Bridge to material database; recommendations appear in batch launch screen
Predictive Quality Gate Decisions	Post-production lab testing (hours to days delay)	Real-time inline quality scoring with hold/release recommendation	AI correlates sensor fusion data from Ignition to final quality; triggers Andon alert for suspect batches
Anomaly Detection in Continuous Processes	Threshold-based alerts after deviation occurs	Multivariate pattern recognition for early warning (15-30 min lead time)	Unsupervised models run on Ignition Historian data; creates pre-alarm in Perspective module
Dynamic Scheduling for Shared Resources	Fixed schedule with manual overrides for breakdowns	AI-rescheduled queue for non-critical work based on real-time OEE and constraints	Integrates with Ignition's MES modules; suggests reschedules to supervisor for approval
Energy Consumption Optimization	Monthly utility bill review; manual setpoint adjustments	AI-driven setpoint and start/stop sequencing for non-critical loads	Uses Ignition's energy monitoring tags; executes via Ignition's scripting to PLCs, saving 5-15% on energy

ARCHITECTING CONTROLLED AI FOR INDUSTRIAL CONTROL SYSTEMS

Governance, Security, and Phased Rollout

Deploying AI in a supervisory control environment requires a deliberate architecture that prioritizes safety, auditability, and operator trust.

AI governance in Ignition starts with a secure inference layer that sits adjacent to, not inside, the core control scripting. We architect AI agents to consume real-time data from Ignition's Tag Historian and SQL Bridge modules via secure, read-only API connections. All AI-driven recommendations—such as adaptive setpoint adjustments or alarm flood filters—are first written to a secure audit log and presented to operators as contextual suggestions within a dedicated HMI screen in Ignition Perspective. Control actions are never executed autonomously without a human-in-the-loop approval step, ensuring the operator retains ultimate supervisory authority. This pattern maintains the integrity of the original PLC logic while layering intelligence on top.

A phased rollout is critical for building confidence and measuring impact. We recommend a three-phase approach:

Phase 1: Observational AI. Deploy models for multi-variable analysis and predictive alerting only. AI provides early warnings on equipment health or process drift via Ignition's alarm notification system, with zero control authority. This phase validates model accuracy and builds operator familiarity.
Phase 2: Advisory AI. Introduce AI-driven setpoint recommendations for non-critical processes. Operators see suggested adjustments (e.g., temperature, pressure, flow rate) in their HMI with a one-click "accept" function. All accept/reject decisions and reasoning are logged for model retraining and audit trails.
Phase 3: Conditional Automation. For validated, high-confidence workflows, implement approved control actions with automated, rule-based limits. For example, AI could automatically adjust a cooling loop setpoint within a pre-defined safe operating envelope during a specific batch phase, with real-time notifications to the operator. Each phase includes defined success metrics, such as reduction in manual alarm acknowledgments or improvement in key quality parameter consistency.

Security is enforced through role-based access control (RBAC) integrated with Ignition's built-in security model. Only authorized engineers can modify AI model parameters or approval workflows. All data flowing to and from the AI inference service is encrypted in transit, and model inputs are sanitized to prevent injection attacks. A continuous feedback loop ensures model performance is monitored for drift, with retraining triggered by deviations in prediction confidence or operator override rates. This structured approach ensures AI augments Ignition's capabilities safely, turning supervisory control into adaptive, intelligence-driven operations without compromising the deterministic safety of the underlying industrial automation system.

Enabling Efficiency, Speed & Accuracy

Intelligent Analysis, Decision & Execution

We build AI systems for teams that need search across company data, workflow automation across tools, or AI features inside products and internal software.

Talk to Us

Search across company data

Give teams answers from docs, tickets, runbooks, and product data with sources and permissions.

Useful when people spend too long searching or get different answers from different systems.

Enterprise searchRAGPermissions

Automate internal workflows

Use AI to route work, draft outputs, trigger actions, and keep approvals and logs in place.

Useful when repetitive work moves across multiple tools and teams.

AI agentsWorkflow automationGovernance

Add AI to products and internal tools

Build assistants, guided actions, or decision support into the software your team or customers already use.

Useful when AI needs to be part of the product, not a separate tool.

AI integrationDecision supportModel routing

AI INTEGRATION WITH IGNITION

Frequently Asked Questions

Practical questions about embedding AI-driven decision logic into Ignition's supervisory control and data acquisition (SCADA) environment for adaptive setpoint adjustment, optimization, and alarm management.

AI models connect to Ignition primarily through its Python or Jython scripting environment and its tag system.

Typical Integration Pattern:

Trigger: A scheduled script, a tag value change, or an alarm event initiates the AI call.
Context Pull: The script gathers relevant real-time and historical tag values (e.g., temperatures, pressures, flow rates, setpoints) from Ignition's internal or connected SQL databases.
Model Action: The script calls a deployed AI model via a REST API or a local inference library. The model receives the context and returns a recommendation (e.g., a new optimal setpoint, a control adjustment, an alarm priority score).
System Update: The script writes the AI's output back to an Ignition tag. This tag can then:
- Drive a PID loop's setpoint directly.
- Update a value on an HMI screen for operator review and manual acceptance.
- Trigger another automation script or a database log entry.
Governance: All AI inferences, input data, and resulting actions are logged to a dedicated audit table in Ignition's database for traceability and model performance monitoring.

About the author

Prasad Kumkar

CEO & MD, Inference Systems

Prasad Kumkar is the CEO & MD of Inference Systems and writes about AI systems architecture, LLM infrastructure, model serving, evaluation, and production deployment. Over 5+ years, he has worked across computer vision models, L5 autonomous vehicle systems, and LLM research, with a focus on taking complex AI ideas into real-world engineering systems.

His work and writing cover AI systems, large language models, AI agents, multimodal systems, autonomous systems, inference optimization, RAG, evaluation, and production AI engineering.

Limited slotsGet a Free AI Consultation

How We Work

Custom AI workflows for your Business

One-fit-all AI don't work for modern businesses. At Inferensys, we aim to understand your business & custom requirements; which we use to define most efficient agentic workflows, the data, and the tools for your business.