Diagram Parsing

Diagram Parsing

The Diagram Parsing system enables users to onboard, interpret, and match physical diagrams (e.g., P&IDs, wiring diagrams, system layouts) with real-world Reliability Assets and Sensors. This connection enables richer metadata-driven monitoring and improves onboarding at scale.

Mermaid Diagram

---

Core Concepts

Concept	Description
ParsableDiagram	Represents a diagram ready to be parsed. Stores the raw source and an extracted image version. Tracks parsing status and extracted objects.
ParsedAsset	Represents detected equipment (assets) on the diagram. Connected to diagram extracts and bounding boxes for layout.
ParsedSensor	Represents detected sensors. Parsed text (e.g., label IDs) are linked to MasterSensor objects to validate accuracy.
MasterSensor	Canonical sensors in the system, allowing parsed sensor text to be validated and matched against known assets.
SourceParsableDiagram	Represents the raw source file uploaded for parsing. Supports triggering parsing workflows and creating ParsableDiagram entries.
LabeledDiagram	Contains annotations (bounding boxes, labels) for assets and sensors over the diagram image.
DiagramParsingMatchingHelper	Assists matching parsed sensor labels to existing sensors in the system using heuristics and confidence scores.

---

Key Processes

Process	Description
Upload Diagram	Upload P&ID or system diagram as a SourceParsableDiagram.
Trigger Parsing	Start an extraction workflow that identifies bounding boxes, parses text, and attempts auto-matching.
Parsed Sensor / Asset Review	Users validate matches or manually correct parsed sensors and assets.
Asset/Sensor Connection	Parsed items are linked to ReliabilityAsset and Sensor entities in the database.
Annotation & Review	Bounding boxes and extracted annotations are stored in LabeledDiagram for visual review and editing.
Parsing Status Tracking	parsingStatus field in ParsableDiagram tracks the lifecycle: Pending → Processing → Completed.
Handling Collisions and Duplicates	Colliding sensors and duplicate detections are flagged for manual review.

---

Important Fields

Field	Meaning
parsingStatus	Indicates parsing progress (e.g., Pending, Parsing, Parsed, Failed).
parsedText	The raw OCR output text for a parsed sensor candidate.
matchConfidenceScore	Numerical score representing the likelihood of correct sensor match (higher = better).
connectedSymbols	Set of symbol IDs in the diagram that are considered logically connected (e.g., piping flow).
boundingBox	Coordinates of parsed assets and sensors, supporting UI overlays for diagram visualization.

---

How This Supports Reliability

Diagram parsing accelerates the onboarding and configuration of Reliability systems, especially in environments with thousands of assets and sensors.
Instead of manually setting up every sensor, facilities can upload diagrams, semi-automatically parse asset/sensor data, verify matches, and link everything directly to Reliability monitoring models.

It drastically reduces onboarding time, increases data fidelity, and supports scalable deployments across facilities.