A relational graph of an electrical diagram is a map of a machine's connections: each component becomes a point, each link becomes a line, and the whole can be explored step by step. This map makes the machine queryable, where a diagram on paper stays fixed.
An industrial machine is rarely read on a single drawing. It spreads across dozens of electrical, pneumatic and hydraulic sheets, and the links between these drawings live mostly in the head of a veteran. The day that veteran leaves, the link is lost. Platforms such as Mimorian start precisely from these drawings to build the graph. Mimorian models equipment, structures failure diagnosis and captures the know-how of maintenance teams through a multi-agent AI architecture. This article describes what this graph is, why a diagram alone is not enough for diagnosis, and how AI links the three families of drawings into a single model.
What is a relational graph of an electrical diagram?
A graph is a simple structure: points, called nodes, and links between these points. Applied to an electrical diagram, each node represents a real component, a relay, a sensor, a motor or a protection device, and each link represents a connection, a wire, a command or a dependency between two components. The AI recognises the standardised symbols of the diagram, which the IEC 60617 standard sets for electrotechnology, identifies the components and the links, and rebuilds the network.
This graph does not describe the shape of the machine, it describes how it works. It is the basic building block of a functional digital twin. For the full framework of this twin, its uses and its differences with the visual or sensor-based twin, read our guide to the functional digital twin for maintenance.
Why is an electrical diagram alone not enough for diagnosis?
An electrical diagram is a two-dimensional drawing, split into sheets. A single signal can cross five pages before reaching the actuator it controls. On paper, following this signal means turning the pages, finding the cross-references from one sheet to another, and keeping the rest in mind.
The relational graph removes this limit. It takes the connections off the paper and makes them navigable. Instead of looking for a page cross-reference, the technician follows a link. Instead of rereading a whole sheet, they ask the model which components depend on the one they are examining. The diagram remains the official reference, the graph makes it usable at the moment of failure.
How does AI link electrical, pneumatic and hydraulic in a single graph?
A machine does not break down by separate families. An electrical signal controls a pneumatic valve, which drives a cylinder, which clamps a part. The three families follow on within the same function. So the AI reads the three sets of drawings, electrical, pneumatic and hydraulic, and links them in a single graph, from end to end.
Take a clamping unit that no longer holds. On separate drawings, each person searches within their own domain. On the graph, you follow the chain: the position sensor sees nothing, because the cylinder did not advance, because the valve did not receive its pressure, because a solenoid valve did not receive its electrical command. The cause moves from one domain to another, and the graph follows that passage. Pure mechanics stays outside this scope: the graph is built on the electrical, pneumatic and hydraulic diagrams.
| Family of diagram | What the AI reads in it | What the graph draws from it |
|---|---|---|
| Electrical | Relays, sensors, protection devices, commands | The commands and safety functions that drive the machine |
| Pneumatic | Valves, cylinders, pressure sensors | The driven actuators and their air supply |
| Hydraulic | Pumps, valves, cylinders, pressure relief valves | The forces, the movements and their points of failure |
What can methods and reliability teams do with it?
The graph is not only for day-to-day troubleshooting. Once in place, it becomes a base that methods and reliability engineers query. As each intervention links to a component and a cause in the graph, the recurring weak points become visible by equipment and by cause, instead of staying buried in free-form reports.
This reading rests on structured data. The ISO 14224 standard sets a reference taxonomy for reliability and maintenance data: equipment, failure cause, maintenance action. The graph files the field feedback within this framework, which makes the analyses comparable from one machine to another. Reliability plans then rest on measured facts, not on impressions.
The graph complements the CMMS here, it does not replace it. The CMMS plans and tracks interventions, the graph brings the structure of the machine and the reading of causes. The two work side by side.
Frequently asked questions
Do you need perfect diagrams to build the graph?
No. The AI works from the existing drawings, in PDF, DWG or even scanned image format. It proposes a graph, and the human validates the links, corrects a reading, adds a missing component. The quality of the graph improves as it is enriched, without waiting for perfect documentation to get started.
Does the relational graph replace the electrical diagram?
No, it builds on it. The diagram remains the official reference of the machine. The graph extracts its links and makes them navigable, to move faster from the symptom to the components involved. The two go together: the drawing for the proof, the graph for the route.
Conclusion
Three key points to remember.
- A relational graph turns a fixed diagram into a navigable map: each component becomes a point linked to its neighbours, and the machine becomes queryable.
- It links the three families of drawings, electrical, pneumatic and hydraulic: the cause of a failure often moves from one domain to another, and the graph follows that passage.
- It serves the field and the methods teams: a guided route to troubleshoot, structured data to analyse weak points, alongside the CMMS.
Next step for a methods manager: choose a critical piece of equipment that combines electrical, pneumatic and hydraulic, gather its sheets, and test the reconstruction of the graph on that scope.
For the full framework, read our guide to the functional digital twin for maintenance, and for the upstream of this modelling, our article on modelling an industrial piece of equipment with AI.
📚 Sources :