Learn

Industrial data and asset performance glossary

The vocabulary you meet across these guides, defined plainly. Where a word means something specific here, the entry says so rather than quietly redefining it. Where the industry genuinely disagrees on a definition, the entry says that too.

Every term links up to the guide that covers it in depth. Start from the Learn hub if you would rather read a topic end to end.

A

Agentic AI

Software that uses a language model to plan and carry out a multi-step task rather than answer a single question, calling tools and reading the results as it goes. In industrial settings the useful version is narrow and supervised: it drafts work a person reviews, and it does not act on equipment.

In depth: Agentic AI in industrial operations

Alert rule

A saved condition evaluated continuously against incoming readings, raising an alert when it holds, for example a temperature above a threshold for five consecutive minutes. The rule is the definition; the alert is the event it produces.

In depth: Asset performance management for renewable fleets

Alert storm

A flood of alerts from a single underlying cause, such as a lost connection making every signal on a site look abnormal at once. Storms are the main reason operators stop trusting alerts, so most alert tuning is about suppressing them rather than adding rules.

In depth: Asset performance management for renewable fleets

Approval boundary

The line between what an automated system may do on its own and what a person must approve first. Fleetera uses the term for a specific rule: the agents that build the KPIs a fleet is measured by, in early access, propose and never apply, so a person approves before anything takes effect.

In depth: Agentic AI in industrial operations

Asset model

A structured description of the equipment on a site: what each asset is, how assets nest, and which named variables each one carries. It is what turns a list of raw signal addresses into something you can ask fleet-wide questions of.

In depth: Industrial digital twin

See also: Binding, Variable, Digital twin

Asset performance management

Also: APM

Using equipment data to decide what to inspect, repair, or run differently, judged by how assets actually perform rather than by a maintenance calendar. The term is used for very different scopes, from vibration analysis alone to full maintenance planning, so it is worth asking which one is meant.

In depth: Asset performance management for renewable fleets

Asset template

A reusable asset definition listing the variables a class of equipment carries, applied to many assets so a fleet is modeled once rather than site by site. In Fleetera a template is a first-class object rather than a starting point you copy: the variable set for a class is defined once and applied to every asset built from it.

In depth: Industrial digital twin

Audit trail

An append-only record of who or what did something, when, and what the result was. For any automated or remote action it is the difference between a system you can operate under a compliance regime and one you cannot.

In depth: Agentic AI in industrial operations

Availability

The share of a period an asset was able to run, expressed as a percentage. Definitions differ on what counts as unavailable: grid outages, planned maintenance, and adverse weather are sometimes excluded and sometimes not, so two availability figures compare only if their exclusions match.

In depth: Defining renewable energy KPIs that survive real data

B

Backfill

Recomputing a metric over a past window, usually because its definition changed or because readings arrived late. A KPI that is never backfilled quietly disagrees with its own history after any change to how it is calculated.

In depth: Defining renewable energy KPIs that survive real data

Binding

The link between a named variable on a modeled asset and the raw address that supplies it, such as a specific node on a controller. Binding is where most of the work in an industrial data project actually lives, because until a signal is bound it is a number with no owner.

In depth: Industrial digital twin

C

Capacity factor

Energy actually produced over a period divided by the energy the asset would have produced running at rated capacity for all of it. It measures how hard an asset worked and says nothing on its own about health, since a low-wind month and a broken turbine both push it down.

In depth: Renewable energy asset management

Channel

One measured signal from one asset, sampled over time: a single temperature, a single power reading. Channel counts are the usual way industrial platforms size a site, because they track the work involved far better than asset counts do.

In depth: Industrial data operations

Cold archive

Long-horizon storage of historical readings in a compressed columnar format, kept cheaply and read less often than recent data. It is where multi-year analysis is answered from, at the cost of slower reads than the hot store.

In depth: Industrial data fabric

Cold buffer

The on-disk half of an edge buffer, holding readings that must survive a restart of the site's software or a longer outage. It trades write speed for durability, which is the right trade for readings you cannot afford to lose.

In depth: Store and forward at the edge

Completeness grade

How much of the data a metric expected was actually present in the window it covers, published alongside the metric itself. A KPI computed from half its inputs is not so much wrong as unlabeled, and a completeness figure is what lets a reader decide whether to trust it.

In depth: Defining renewable energy KPIs that survive real data

Condition monitoring

Watching measured signals such as vibration, temperature, or pressure to judge the health of equipment while it runs, rather than during a scheduled inspection. It is the input to most predictive maintenance work, not a synonym for it.

In depth: Asset performance management for renewable fleets

Contextualization

Also: Contextualisation

Attaching meaning to a raw signal: which asset it came from, what it measures, in what unit, and how it relates to other signals. Without it a time series is a column of numbers only its original author can interpret.

In depth: Industrial data fabric

Curtailment

Deliberately reducing a plant's output below what conditions allow, usually because the grid cannot absorb the energy or prices have gone negative. It is an industry term for something a grid operator or plant controller does: Fleetera measures and reports curtailment, it does not perform it.

In depth: Renewable energy asset management

D

Data fabric

Also: Industrial data fabric

An architecture in which data from many sources stays where it is best stored but is modeled, governed, and queried as one connected layer. In industrial use the distinguishing feature is that the connective tissue is the asset model, not a second copy of the data.

In depth: Industrial data fabric

Data gap

A stretch of time with no readings for a signal that should have had them. Gaps come from lost connectivity, a stopped collector, or equipment that was genuinely switched off, and telling those cases apart is a large part of judging data quality.

In depth: Store and forward at the edge

Data lake

A store holding raw data in close to its original form, cheap to write to and light on schema, with structure applied when it is read. It solves storage economics rather than meaning, which is why lakes usually need a modeling layer above them before anyone can use the contents.

In depth: Industrial data fabric

Depth of discharge

Also: DoD

How much of a battery's usable capacity a discharge consumed, as a percentage. Deeper cycles extract more energy per cycle and generally shorten cell life, so it is a lever operators trade off against revenue.

In depth: Renewable energy asset management

Derating

Running equipment below its rated capability on purpose, to protect it or to respect a limit such as temperature, grid conditions, or a control setpoint. A derated asset looks like an underperforming one in a report unless the reason is recorded alongside the output.

In depth: Renewable energy asset management

Device shadow

A cloud-held copy of a device's intended configuration and its last reported state, so applications can read and set it even while the device is offline. A shadow is about state and configuration sync; a digital twin is about modeling the asset, and the two are routinely confused.

In depth: Industrial digital twin

Digital twin

A software model of a physical asset, kept current from that asset's own data. Usage varies widely: the term covers a 3D visual model of the equipment, a physics simulation of its behavior, and a structured operational data model of it. The three imply very different work and very different claims.

In depth: Industrial digital twin

E

Edge gateway

Hardware or software at the site that sits between equipment and the wider network, collecting readings locally and forwarding them onward. It is the general industry term, and what it covers ranges from simple protocol translation to running analysis on site.

In depth: Store and forward at the edge

Edge runtime

The software process running at a site that speaks to the equipment, buffers what it reads, and forwards it to the cloud. Fleetera uses the term for its own on-site component, which holds the protocol drivers, the buffer, and the forwarding path in a single process rather than a stack of services.

In depth: Store and forward at the edge

Equivalent full cycles

Also: EFC

Total energy cycled through a battery over a period, expressed as the number of complete charge and discharge cycles it adds up to. It is the standard way to compare batteries that never cycle in neat whole numbers, and it usually drives warranty terms.

In depth: Renewable energy asset management

F

Fleet

A group of assets of the same kind, operated together and compared against one another, such as every turbine across a set of wind sites. The point of the word is comparability, which assumes the assets are modeled the same way.

In depth: Renewable energy asset management

H

Historian

A time-series database built for process data, traditionally installed on site, storing readings with timestamps and quality flags. Historians are excellent at recording and weak at cross-site analysis, which is why fleet questions usually need a layer above them.

In depth: Industrial data fabric

Hot buffer

The in-memory half of an edge buffer, holding the most recent readings for fast forwarding while the link is healthy. It is quick and small, and it does not survive a restart, which is what the cold half exists for.

In depth: Store and forward at the edge

Human in the loop

A design in which an automated system proposes and a person decides, so no consequential action happens without someone approving it. It is a control rather than a disclaimer: the review has to be a step the system genuinely cannot skip.

In depth: Agentic AI in industrial operations

Hysteresis

A deliberate gap between the level that raises an alert and the level that clears it, so a value hovering on the threshold does not raise and clear repeatedly. It is the simplest effective defense against noisy alerting.

In depth: Asset performance management for renewable fleets

I

IIoT

Also: Industrial internet of things

Industrial internet of things: connected sensors and equipment in industrial settings, plus the networks and platforms that collect from them. The label covers an enormous range of maturity, from one connected sensor to a fully modeled fleet.

In depth: Industrial data operations

Industrial data operations

The practice of getting equipment data out of a plant and into a usable state: connecting controllers over their own protocols, buffering readings when the link drops, modeling raw signals as named variables on real assets, and keeping a queryable history.

In depth: Industrial data operations

K

KPI

Also: Key performance indicator

A key performance indicator: a defined, repeatable measure of how something is performing, such as availability or capacity factor. The value of a KPI comes from its definition being written down and agreed, not from the number on its own.

In depth: Defining renewable energy KPIs that survive real data

M

Mean time between failures

Also: MTBF

The average operating time between one failure and the next, over a population of assets and a stated period. It is meaningful only with its denominator stated, since a short observation window on new equipment produces a flattering figure.

In depth: Asset performance management for renewable fleets

Modbus

A simple, long-lived industrial protocol for reading and writing numbered registers over serial links or TCP. It carries no type or unit information, so what a register means has to come from documentation or from the model you build around it.

In depth: Industrial data operations

MQTT

A lightweight publish and subscribe messaging protocol widely used to move sensor data across unreliable networks. It defines how messages travel, not what they contain, so payload structure stays a separate agreement between publisher and consumer.

In depth: Industrial data operations

O

OPC UA

Also: OPC Unified Architecture

A modern industrial protocol with a typed, browsable address space, so a client can discover what a device exposes instead of being told in advance. It is the usual choice for new industrial integrations and the most information-rich of the common protocols.

In depth: Industrial data operations

Operational technology

Also: OT

The hardware and software that runs physical processes, such as controllers, drives, and control systems, as distinct from the IT systems that run the business. The two worlds have different priorities, and most industrial data work happens on the boundary between them.

In depth: Industrial data operations

P

Performance ratio

Also: PR

For solar plants, the energy actually produced divided by what the measured irradiance and the plant's rating say it should have produced. It normalizes the weather out, which is what makes it comparable across sites and across months.

In depth: Renewable energy asset management

PLC

Also: Programmable logic controller

A programmable logic controller: the ruggedized computer that runs a machine or a process, executing control logic on a repeating cycle. It is usually the source of the values an industrial data platform reads.

In depth: Industrial data operations

Polling

Also: REST polling

Reading a value by asking for it on a fixed interval, rather than waiting for the device to publish a change. It is how Modbus and most REST integrations work, and the interval sets both the resolution of your history and the load you put on the equipment.

In depth: Industrial data operations

Portfolio

The full set of assets an owner or operator is responsible for, usually across several technologies and locations. Where a fleet groups like with like for comparison, a portfolio is a commercial grouping and often mixes wind, solar, and storage.

In depth: Renewable energy asset management

Predictive maintenance

Using condition data to act before a failure, rather than on a fixed schedule or after a breakdown. Claims in this area vary enormously in strength, and the credible version is narrow: a named failure mode, on a named asset class, with evidence behind it.

In depth: Asset performance management for renewable fleets

R

Round-trip efficiency

Also: RTE

For storage, the energy a battery returns divided by the energy put into it over a full charge and discharge cycle. Where the measurement boundary sits matters: a figure taken at the cells and one taken at the grid connection differ, because conversion losses and auxiliary loads fall in between.

In depth: Renewable energy asset management

S

SCADA

Also: Supervisory control and data acquisition

Supervisory control and data acquisition: the system operators use to watch and control a plant in real time. It is built for live operation rather than for history or analysis, which is why its data is usually copied elsewhere before anyone analyzes it.

In depth: Industrial data operations

Site model

The structure of one location: which assets it contains, how they are grouped, and what identifies the site itself. It is the level at which most reporting and most access control are scoped.

In depth: Industrial digital twin

State of charge

Also: SoC

How much energy a battery currently holds, as a percentage of its usable capacity. It is estimated rather than measured directly, so two battery management systems can report noticeably different values for the same physical state.

In depth: Renewable energy asset management

Store and forward

Also: Store-and-forward

Holding readings locally while the connection to the cloud is unavailable and sending them once it returns, so an outage costs continuity rather than data. Every buffer has a limit, and past it the oldest readings are dropped: the honest question is how much and for how long, not whether loss is possible.

In depth: Store and forward at the edge

See also: Hot buffer, Cold buffer, Data gap

T

Tag

The industrial name for a single readable value on a device, inherited from control system practice. Tag names are typically written for whoever configured the controller, which is why platforms model them into named variables rather than exposing them as they are.

In depth: Industrial data operations

Telemetry

Measurements taken automatically at a distance and transmitted somewhere else to be recorded. In industrial platforms it is the general word for the stream of timestamped readings flowing from equipment into storage.

In depth: Industrial data operations

Time-series store

A database designed for timestamped readings, optimized for appending continuously and for reading ranges across many signals at once. It is the difference between a long-history range query returning promptly and a general-purpose database struggling with the same shape of question.

In depth: Industrial data fabric

Tool call

A step where a language model asks the surrounding system to do something concrete, such as run a query, and reads the result before continuing. Tool calls are where an agent's permissions actually live, so the set of tools defines what an agent can and cannot do.

In depth: Agentic AI in industrial operations

U

Unit normalization

Also: Unit of measure normalization, Unit of measure normalisation

Converting readings into agreed units so values from different equipment can be compared and added, and recording the unit alongside the value. Skipping it is how a fleet report ends up summing kilowatts and megawatts.

In depth: Industrial data fabric

V

Variable

A named, typed, unit-carrying measurement on a modeled asset, such as a turbine's active power. Fleetera says variable for the modeled thing and tag for the raw source it is bound to, a distinction the wider industry does not draw consistently.

In depth: Industrial digital twin