Industrial AI projects often begin with a model discussion before the team has confirmed that the required signals, history and operating context are available. A short readiness assessment prevents this mismatch and identifies the fastest path to reliable evidence.
Signal and history
Identify the measurements, states, alarms and events needed to understand the use case. Confirm update frequency, retention, missing periods, timestamp consistency and whether the source can be accessed without disrupting the control environment.
- Source system and protocol
- Sampling rate and retention
- Missing, stale and substituted values
- Time synchronisation
Operational context
Record the asset hierarchy, operating modes, product or recipe, maintenance events and known limits that explain the signal. Without this context, a model can learn correlations that are operationally misleading.
- Asset and process hierarchy
- Operating state and production schedule
- Maintenance and change history
- Quality outcomes and operator notes
Ownership and deployment
Define who owns each source, who can approve access and where the solution must run. The architecture should reflect data sovereignty, network segmentation, latency and degraded-connectivity requirements from the beginning.
A good readiness review ends with a small, testable scope and a list of evidence gaps that can be closed during the pilot.
Standards and primary sources used for context.
These links provide technical context. Their inclusion does not claim product certification or compliance unless explicitly stated elsewhere.
