Configuration & lifecyclenormative v1.4.0bundle v1.4.132 blocks

22. Telemetry, media, tests, dyno, simulation, condition, and evidence quality

Concept & directionLNK

Telemetry is not a collection of unlabeled numbers, and a photograph is not self-explanatory truth. Every dataset needs a manifest connecting channels, clocks, instruments, calibration, configuration, operation, rights and quality. Tests and derived metrics need procedures, acceptance criteria and reproducible processing. High-volume bytes live in immutable artifact storage; the lifecycle ledger stores their meaning and custody.

#22.1 Telemetry segment manifest

Manifest area	Required fields
Scope	asset/assembly, exact configuration(s), operation/segment, mode, control regime and location/privacy context
Source	device/controller/logger, hardware/software/firmware, sensor instances, calibration and capture authority
Time	start/end, phenomenon/result time, clock, timezone, synchronization method, drift, precision and producer sequence
Channels	stable signal ID, observed property, feature of interest, datatype, quantity kind, unit, frame, sign convention and null semantics
Sampling	rate/schedule, trigger, anti-alias/filter, resampling/interpolation and event versus periodic behavior
Quality	validity, range, saturation, dropout, packet loss, out-of-order data, duplicates, clock gaps and anomaly summary
Artifact	encoding/container, compression, schema, byte count, digest, chunk index, storage locator and encryption
Policy	privacy class, precise-location treatment, retention, export, license, redaction and permitted derived uses

#22.2 Channel descriptor and signal adapters

A channel refers to a phenomenon/quantity and feature of interest, not only a bus address or column label.
Preserve the native signal name, source namespace, datatype, unit, enum/version and scaling before mapping to a shared concept.
COVESA VSS can be a road-vehicle signal adapter; it is not the universal tree for vessels, aircraft, robots, blimps or spacecraft.
Map sensors, observation procedures, observed properties, result time and actuations through SOSA/SSN-compatible semantics where useful.
Do not publish a value when its native source marks it unavailable; preserve unavailable, not-measured and unsupported separately.
The same physical property from two sensors remains two observations until a declared fusion method derives a new estimate.

#22.3 EvidenceArtifact and media provenance

Area	Required semantics
Identity/integrity	artifact ID, media type, byte length, digest, encoding, storage, encryption and availability state
Capture	captured time/precision, device, actor, location/frame, operation/configuration and source clock
Custody	transfer, import, copy/derivative, verification, access, loss, destruction and retention events
Edit lineage	parent artifacts, editor/tool, transformations, crops/redactions, transcoding and output digest
Synthetic disclosure	none, assisted, partially synthetic or fully synthetic; generator/model/prompt/seed where policy permits
Rights	copyright, likeness, vehicle livery/franchise, license, purpose, territory, term and redistribution/model-training rights
Privacy	people, plates, exact location, private property, interiors, documents, audio/voice and redaction policy
Evidence role	supports which bounded proposition, direct/indirect/illustrative role and limitations

Table

Content credentials are provenance, not truth
A valid content credential can show binding, signer and edit lineage; it does not prove that the depicted event happened or that the caption is accurate. A synthetic image can still be valuable creative media when clearly disclosed and isolated from factual maintenance, incident and registry evidence.

#22.4 TestDefinition and TestRun

Definition layer	Run layer
Purpose, requirement and acceptance criteria	Actual test article(s), configuration, state and deviations
Procedure/revision and applicability	Start/end, operator, facility, permissions and incidents
Apparatus, sensors, channel schema and calibration	Exact equipment/sensor instances, calibration status and setup evidence
Environment, load and preconditions	Observed environment, load, fuel/energy batch, payload and resource state
Excitation/profile, sampling and abort criteria	Commands/excitation, sequence, anomalies, aborts and excluded intervals
Uncertainty budget and processing plan	Raw artifacts, pipeline version, transformations, corrections and uncertainty
Evidence and reviewer requirements	Results, pass/fail/partial/invalid, reviewer, approvals, limitations and retest links

#22.5 Dyno and performance comparability

Area	Required context
Machine + measurement location	engine/motor bench, crank/shaft, chassis roller, hub, wheel, propeller shaft, thrust stand, hydraulic/tool bench
Load device	inertia, eddy-current, water brake, electrical machine, aerodynamic/hydrodynamic load or other controlled sink
Configuration	engine/motor, gearing, tires/pressure/temperature, propeller/rotor, controller, calibration, exhaust/intake and restraints
Resource state	fuel/material batch, battery SOC/SOH/temperature, voltage, cooling, lubricants and ambient supply
Reference/correction	environment, correction method/standard, smoothing/filter, drivetrain-loss method and run-rejection rule
Raw + derived	time-series channels, curves, peak/average/integral metrics, formulas, uncertainty and processing version
Disposition	valid/invalid/partial/aborted, anomalies, acceptance, repeatability and comparability group

Never compare crank power with wheel power as equivalent.
Never compare corrected and uncorrected values without the exact correction basis.
Never hide smoothing, run rejection, gearing, tire, loss-estimation, battery or environment differences.
A headline peak retains the run, curve, measurement location, configuration and uncertainty that created it.

#22.6 Derived metrics, simulation, and digital-twin evidence

Artifact	Must retain
Derived metric	input records/artifacts, formula/model, implementation digest, parameters, exclusions, units, uncertainty and execution receipt
Simulation model	purpose, fidelity, assumptions, equations/solver, geometry/configuration, material/aero/hydro data, boundary/initial conditions and validation
Simulation run	model version, exact inputs, compute environment, random seeds, convergence/error, outputs, anomalies and reviewer
Digital twin projection	physical source asset, synchronization authority, data cutoff, state estimator/model versions, latency and unresolved divergence
AI inference	model/runtime, input schema and records, output schema, confidence calibration, operating domain, human review and prohibited uses
Comparison set	compatibility criteria, normalization, exclusions, sample size, selection policy and information-loss warning

#22.7 Condition, defect, damage, failure, cause, and repair

Object	Required distinction
Condition observation	Measured/observed property, target, method, result, uncertainty, time, configuration and evidence
Defect / nonconformance	Requirement violated or abnormal state, location, severity, status, operational effect and disposition
Damage	Type, geometry/location/frame, extent, initiating event, progression, allowable limit and scans/NDT
Failure event	Function lost/degraded/unintended, time, item/configuration, symptoms, effects, detection, shutdown and recovery
Causal assertion	Candidate cause/mechanism, contributing factors, method, evidence, confidence, alternatives and dispute state
Repair definition	Applicability, geometry/material/process, strength/performance basis, inspection, life limit and approval
Repair execution	Actual damage, preparation, removed material, repair feature/lots, process/cure, deviations, measurements, NDT and release

Table

Cause never overwrites observation
A crack, fault code or power loss is an observation/event. Root-cause analysis is a separate, revisable assertion with competing hypotheses. Repair can restore function but never erases the damage, failure or evidence history.

#22.8 Health, remaining life, and readiness projections

Health, wear, remaining useful life, readiness and predicted due date are derived projections with algorithm/model version, input meters/condition, uncertainty and as-of time.
Separate component, system, mission and fleet readiness. One degraded camera need not make a truck immobile; one failed brake may make it unavailable for road operation.
Separate detected fault, diagnosed cause, deferred defect, operational limitation, maintenance due and release-to-service status.
A projection can change when new evidence arrives without rewriting the original telemetry, inspection or work records.
Public profiles should expose friendly summaries while owner/technician views preserve the exact evidence and conflicts behind them.