Resources

Unfamiliar terms, explained simply.

Understand platform terminology in the context of equipment data and operational work.

Details

Scope and considerations

Tag: a named data item representing a measurement or state. A “Press 1 hydraulic pressure” tag can include its value, unit, timestamp and data quality. Reference: Inductive Automation
Edge: collecting and processing data close to its source, or the device that performs this work. An industrial PC beside a machine can collect and process signals before forwarding them to a central server. Reference: AWS
ETL/ELT: methods for extracting, transforming and loading data, with processing flows configured and managed for repeated execution. ETL transforms before loading; ELT loads first and transforms afterward. Standardizing equipment units and timestamps is one use. ETL: IBM · ELT: IBM

Lakehouse: an architecture combining a data lake’s ability to store diverse data with a warehouse’s data management and analytical querying. It can support analysis of equipment signals, inspection images and production history. Reference: IBM
Data catalog: an inventory of metadata describing available data, its structure, location and ownership. It helps an analyst find pressure data and check its units, source machine and responsible team. Reference: IBM
Data lineage: a record of data’s origin, transformations and downstream use. It lets a quality analyst trace a report figure back through calculations to the original equipment signals. Reference: IBM
Tenant: a customer or organization treated as a distinct unit within a service. Tenants provide a scope for managing access to users, data and settings; a manufacturer may organize them by customer or site. Reference: Microsoft
MDR (Metadata Relation): NAZARE’s term for managing relationships between metadata that describes the meaning of data. For example, pressure tags from different equipment vendors can be linked to a common hydraulic-pressure definition. MDR in NAZARE

Downtime: time when equipment stops producing during planned production time. Record unplanned breakdowns and planned changeovers separately; time with no scheduled production is a different category. Reference: Vorne
OEE (Overall Equipment Effectiveness): a measure of how effectively planned production time produces good parts. Availability × performance × quality combines losses from stops, reduced speed and defects. Calculation: Vorne
Cycle: one repetition of a machine or operator’s work; cycle time is the time needed to complete it. Define consistent boundaries, such as a press starting compression and completing its return, to compare durations and variation. Reference: Lean Enterprise Institute
Parts life: how long, or for how many operations, a component can be used. Replacement thresholds may use operating hours or cycle counts. Remaining useful life prediction instead estimates time or use left from condition data, with uncertainty. Remaining useful life: MathWorks

ML serving (Model Serving): deploying a trained model in an operational environment to return predictions for new inputs. A model might receive live pressure and vibration signals and return an anomaly score. Reference: TensorFlow
Ontology: an explicit, machine-interpretable structure of concepts, properties and relationships in a domain. Manufacturing equipment, parts and signals can be connected with operating criteria and response knowledge so AI can identify what a reading means and which asset needs attention. Reference: W3C
Tacit knowledge: experience-based skills and judgement that are difficult to fully express in words, such as recognizing an abnormal sound. Ingkle uses interviews and tools to capture the explainable conditions, relationships and response criteria for AI use. Reference: IBM
MCP (Model Context Protocol): an open protocol connecting AI applications to external data and tools. NAVIS uses it to access NAZARE capabilities; for equipment control, NAZARE receives an approved request and sends commands to the PLC. Reference: MCP documentation
HITL (Human-in-the-Loop): human participation in an automated process to review, correct or approve its output. In manufacturing, an operator can review an AI proposal and its evidence, approve or reject it, and allow execution within the approved scope. Reference: IBM

Start together

Tell us about your equipment, data and priorities to review a path from connection to operations.