Health Scores, Baselines, and Model Readiness

Health scores and baselines turn raw equipment signals into an operating view. They are useful only when the team understands what data was used, which equipment the result describes, and whether the model or rule path is ready for review.

Health Review Flow

Prerequisites

Requirement	Why it matters
Equipment profile	Health output needs a stable asset boundary.
Signal history	Baselines and trends depend on chronological readings.
Operating-state context	Running, stopped, standby, and maintenance periods affect interpretation.
Template or rule path	Equipment classes need reviewed thresholds, health bands, or model assignment.
Review owner	A person or team must decide whether the output is ready for operations.

Source Data Inputs

Input	Use
Equipment profile	Defines the asset, class, location, and identity bridge.
Time-series readings	Feed baselines, health history, and anomaly review.
Operating schedule	Separates real operating trends from idle periods.
Maintenance history	Explains repaired, replaced, calibrated, or inspected assets.
Failure-mode catalog	Provides diagnosis context and review labels.
Rule or template configuration	Defines thresholds, bands, and advisory behavior.

What the Health View Should Answer

Question	Review surface
Which equipment is being scored?	Equipment profile and main equipment bridge.
What is the current state?	Latest health snapshot and dashboard summary.
Is the trend getting worse?	Health history and signal trend views.
Which signals contributed?	Feature contribution, fault signature, or diagnosis evidence where enabled.
Is the model ready?	Model assignment, baseline status, data gates, and lifecycle events.
What should the team do next?	Advisory, maintenance recommendation, or manual review note.

Health Score Inputs

Health scoring can combine multiple dimensions depending on equipment class and configured templates.

Input	Example interpretation
Vibration health	Mechanical stress, imbalance, looseness, bearing risk, or ISO grade.
Temperature health	Thermal load, overheating, or abnormal temperature trend.
Operating state	Running-state filtering so stopped equipment is handled separately from operating trend data.
Maintenance history	Recent maintenance may reset or explain signal changes.
Failure mode catalog	Helps connect measured symptoms to known failure patterns.
Rule or template configuration	Defines thresholds, health bands, and advisory behavior for the equipment type.

Baseline and Threshold Readiness

Before using a health score in planning, check whether the baseline or threshold is usable.

Readiness item	What to check
Data window	Enough chronological history exists for the selected equipment and signals.
Signal coverage	Required channels are present and mapped to the right units.
Operating state	Running, stopped, standby, and maintenance periods are handled consistently.
Template match	The equipment class has an applicable template or explicitly reviewed rule set.
Review state	The current baseline, threshold, or template is approved for operational use where approval is required.
Drift evidence	Recent changes in equipment, sensor, or operating mode are visible.

No-Data and Low-Confidence States

A no-data state is better than a fabricated score. Show the reason clearly so operators know what to fix.

State	Likely cause	Action
No latest health	No recent signal history or profile linkage issue.	Check equipment mapping and recent readings.
Sparse history	History window is too short for trend or baseline review.	Load more history or limit the decision to manual review.
Stale signals	Source connector has not synced recently.	Check DFS sync history or source system status.
Unassigned model	No model or rule path is active for the equipment.	Assign a template, rule set, or baseline workflow.
Blocked readiness	Data gate or approval gate is not satisfied.	Fix the blocker before using the output for dispatch.

Review Model Readiness

Where the model console is enabled, use it to check:

• active assignment for the equipment;
• model or rule state;
• recent lifecycle events;
• retraining or recalibration blockers;
• manual-review or held-for-review state;
• evidence behind the current assignment.

Model readiness is an operations control. It helps teams decide whether an advisory can proceed, needs human review, or should be held until data quality improves.

Use Health Output in Operations

Health scores should support review, not replace maintenance judgement. A typical decision path is:

1. Check the equipment profile and current health snapshot.
2. Review health history and recent signal changes.
3. Confirm data freshness and readiness.
4. Compare similar equipment in fleet or criticality views.
5. Open related anomalies or advisories.
6. Record the decision and outcome.

Expected Output

The review should produce:

• a current health state for each selected asset;
• a clear no-data or low-confidence reason where scoring is unavailable;
• readiness status for the active template, rule, or model path;
• evidence links for the signals and history behind the score;
• an action path for anomaly review, manual inspection, or data-quality repair.

Validation Checklist

• Health score and health history reference the intended equipment.
• Required signal channels are present for the equipment class.
• Baseline or threshold state is visible to the reviewer.
• No-data and low-confidence states explain what is missing.
• Recent maintenance events are considered before interpreting trend changes.
• Output is reviewed before being used for dispatch or work-order creation.

Troubleshooting

Symptom	Check
Latest health is unavailable	Recent readings, equipment profile, tenant scope, and required signals.
Health changes sharply after maintenance	Repair event, sensor calibration, replacement, and baseline reset needs.
Model readiness is blocked	Missing channels, stale data, low coverage, or approval gate.
Health score conflicts with field observation	Source signal quality, operating state, sensor position, and recent work orders.

Related Docs

• Equipment, Sensors, and Signal History
• Anomaly Review and Maintenance Advisories
• Fleet Triage and Criticality
• Predictive Maintenance AI Tools