OT observability · A VIEOX perspective
A signal becomes useful when its context travels with it.
Connect industrial signals and events with equipment identity, timing, units, operating states and quality information for a more useful OT investigation.
The practical question
What context should accompany industrial signals and events?
Preserve the equipment identity, meaning, unit, time reference, information quality and operating scope that a reviewer needs to interpret the item. Add relevant product, recipe, maintenance or inspection context when it is available and authorised. Keep the original observation distinguishable from any later explanation or calculated result.
Measurements, state changes and human accounts describe different aspects of the work. Their value in a review depends on whether a reader can understand what each item represents and how it relates to the others. This page explores the context that makes that connection possible.
Use an identity people can follow across records
A machine may be named differently in an operator log, a maintenance system and a production report. Agree how those names relate before comparing records. Preserve component and location context where it changes the meaning of the observation; a reading from the motor is not automatically a reading from the driven process.
Make changes understandable over time. If equipment moves, a sensor is replaced or a label is corrected, the old record still needs its historical meaning. A current name should not silently rewrite what an earlier observation described.
Explain when the event happened and when it became available
A value’s recorded time and the time a receiving system noticed it may differ. Use the relevant time for the question and preserve that distinction when it affects event order. Review clock differences, collection gaps and delayed updates before drawing a sequence of causes.
For a practical investigation, identify the time reference and any unresolved uncertainty near the timeline. A colleague should not need to infer the meaning from a chart label. If two events cannot be confidently ordered, keep both plausible sequences open.
Prepare information for re-use
NIST’s guidance on collecting, curating and re-using manufacturing data treats data management as part of the work of obtaining useful shop-floor information. It recommends broadly accepted open standards where possible. The practical implication is to plan how someone will interpret the collected record, not only how the first connection will be made.
Sources: [1]
Keep measurements, events and explanations distinct
A measured temperature, a machine state change and a technician’s interpretation can belong in the same review without becoming the same kind of claim. Label each clearly. Calculated values should identify the calculation’s purpose and the inputs needed to understand it.
Treat units and quality indicators as meaningful context. A value that is stale, outside its intended measurement conditions or unavailable should not look like an ordinary current observation. Explain what the team can reasonably compare and where specialist interpretation is needed.
Attach the work that was taking place
The same equipment can behave differently across products, recipes, load conditions and stages of a sequence. Keep those associations with the recorded period when they are available. Do not fill a missing association with the next convenient order or batch merely to complete a display.
Hypothetical example: two runs show different current patterns. One includes a changed part and a longer handling stage. The review records these differences before considering an equipment concern. The changed trace becomes a question about comparable work rather than an immediate diagnosis.
Make context a shared responsibility
Agree who can confirm an equipment mapping, resolve an ambiguous event and approve a change to a definition. Keep a route for operators and specialists to report a mismatch. A useful context model should become easier to maintain as people encounter real questions.
Start with one investigation and list the associations the team repeatedly recreates. Prioritise those that materially affect understanding. This keeps the work grounded in a reader’s need and gives future analytics or AI evaluations a more dependable starting record.
Put the question to work
A practical checklist
- Reconcile equipment names and component boundaries.
- State the time reference and any ordering uncertainty.
- Preserve units, freshness and information quality.
- Distinguish observations, accounts and derived values.
- Assign responsibility for maintaining important definitions.
Questions, answered
Should missing context be inferred automatically?
An inference can be useful if it is clearly identified and reviewed for its intended use. It should not silently replace a missing fact. Preserve the original gap and make the basis of any inferred association understandable.
Do all users need every technical field?
No. Present the information needed for the task, with an appropriate route to supporting detail. The context must remain available to authorised reviewers even when a daily view presents a simpler account.