Channel ALTITUDE_FT
Altitude
Channel AIRSPEED_KTS
Airspeed
Channel FUEL_PCT
Fuel level
Evidence-backed fault isolation
Diagnostics
Inspect every finding and validation issue without losing invalid source rows.
Rule-engine output
Findings 0
| Severity | Rule ID | Source and time | Observed | Expected | Evidence |
|---|---|---|---|---|---|
| Load telemetry to run diagnostics. | |||||
Validation containment
Quarantined rows 0
No quarantined rows.
Deterministic test stimulus
Fault injection
Create a reproducible candidate from the current run. The original source remains unchanged.
Before-and-after regression analysis
Verification
Compare stable finding identities and expose resolved, persisting, and new behavior.
Baseline
Not captured
- Hash
- --------
- Samples
- 0
- Findings
- 0
Candidate
Not selected
- Hash
- --------
- Samples
- 0
- Findings
- 0
Outcome
Comparison pending
Capture a baseline and candidate to compare.
Resolved
0- None
Persisting
0- None
Newly introduced
0- None
Release evidence
Requirement results
| Requirement | Result | Evidence |
|---|---|---|
| REQ-VER-001 | Pending | Comparison has not run. |
Temporal fault intelligence
Investigation
Reproduce a labeled synthetic mission, fuse redundant sensors, inspect temporal evidence, and compare authoritative rules with experimental model hypotheses.
Linked sample
0 / 0
No comparison baseline captured.
Current and captured baseline waveforms
Observed and predicted state
Innovation evidence
Normalized sensor residual
Declared lifecycle
Fault timeline
Run a synthetic scenario to inspect onset and recovery.
Advisory only
Ranked hypotheses
- The temporal model is disabled.
Authoritative diagnostics
Rule evidence
- No investigation has run.
Hysteresis state machine
Phase transitions
- No phase evidence is available.
Authoritative and advisory signals
Detector agreement
- No detector evidence is available.
Reproducible evaluation campaign
Scenario lab
Run declared scenarios across disjoint seeds without blocking the interface.
Not started.
- Status
- Not run
- Cases
- 0
- Episode F1
- N/A
- False alarms / run
- N/A
- Median detection delay
- N/A
- Abstention rate
- N/A
Campaign results will show expected, missing, and unexpected detections.
Versions, provenance, and experiment controls
Configuration
Every analysis is tied to explicit schemas, adapters, profiles, rules, and hashes.
Active analysis
- Application
- 2.2.0
- Schema
- telemetry-run/2.0.0
- Adapter
- None
- Profile
- None
- Profile version
- None
- Dataset hash
- Not loaded
Source data is excluded by default and only included by this explicit choice.
Version 1 protocol
Streaming simulator
- Sources
- 0
- Messages
- 0
- Dropped
- 0
- Heartbeat age
- N/A
- Queue depth
- 0
- Reconnects
- 0
Experimental comparison only
Learned baseline
Deterministic rules remain authoritative. The model is enabled only when its stored held-out evaluation meets both release gates.
- Model version
- Not loaded
- Held-out F1
- Not measured
- False-positive rate
- Not measured
- Current anomaly score
- N/A
Profile-specific model registry
Temporal model
Eligibility requires exact schema, profile, channel, unit, cadence, window, artifact, and configuration identities. The rule engine remains authoritative.
- Registry entry
- Not loaded
- Compatibility
- Not evaluated
- Artifact SHA-256
- Not loaded
- Configuration SHA-256
- Not loaded
- Window
- Not loaded
- Training evidence
- Not loaded
- Calibration evidence
- Not loaded
- Held-out evidence
- Not loaded
- Authority
- Deterministic rules
Run compatibility evaluation to see evidence.
Profile-driven logic
Enabled rules
| Rule ID | Severity | Condition |
|---|---|---|
| Select a profile. | ||