CBM Sensor Health Correlation Index (CSHCI)

GMI-RE-046Renewable EnergyIntegration-RequiredPattern O-8: Sensor Disambiguation

Definition

Correlation between CBM-recommended interval extensions and sensor health degradation indicators. Tests whether extensions cluster on assets with failing sensors, the case where the model cannot distinguish improving asset condition from a sensor losing sensitivity.

The Measurement Gap

The Measurement Gap

Condition-based maintenance models interpret sensor data to assess asset health and recommend maintenance intervals. When sensor readings indicate normal condition, the model recommends extending the interval. When readings indicate degradation, the model recommends advancing the maintenance. This logic works correctly when the sensors are healthy and accurately reporting asset condition. The failure mode is that sensor degradation itself can produce readings that look like healthy asset condition.

A vibration sensor losing sensitivity will report lower vibration amplitudes, which the CBM model interprets as healthy bearings. An oil analysis sensor with calibration drift may report cleaner oil than actually exists. A temperature sensor with increasing thermal resistance may underreport temperatures. In each case, the sensor degradation produces readings that mimic asset health improvement, and the CBM model responds by extending the maintenance interval. The asset is getting worse while the model says it is getting better, and the maintenance interval extends at exactly the moment it should be shortened.

Without CSHCI, this failure mode is invisible because the CBM model has no mechanism to question its own sensor inputs. It takes the data at face value. Sensor health diagnostics may exist in the condition monitoring system, but they are not typically integrated into the interval recommendation logic. The sensor health data sits in one system while the interval decision happens in another.

The Decision It Enables

CSHCI provides a statistical test for the most dangerous CBM failure mode: intervals extending because sensors are going deaf rather than because assets are genuinely healthy. When CSHCI is near zero or negative, interval extensions are not correlated with sensor degradation, indicating the CBM system is operating on trustworthy sensor data. When CSHCI is positive and statistically significant, the pattern indicates that longer extensions are associated with unhealthier sensors. This finding triggers a sensor health gate requirement: no interval extension should be applied unless the contributing sensors pass a health check confirming they are operating within calibration and sensitivity specifications.

Formula

Primary Formula:

CSHCI = Pearson_Correlation(Interval_Extension_Magnitude, Sensor_Health_Degradation)

Variable Definitions:

Variable Definition
Interval_Extension_Magnitude the ratio of the extended interval to the original interval per asset. A value of 2.0 means the interval was doubled. A value of 1.5 means the interval was extended by 50%
Sensor_Health_Degradation a composite sensor health score decline over the measurement period per asset, normalized on a scale from 0 (no degradation) to 1 (failed sensor). The composite should incorporate signal-to-noise r...

Calculation Example:

A fleet of 60 turbines with CBM-managed maintenance intervals is analyzed. For each turbine, the interval extension magnitude and the composite sensor health degradation score are computed over a 12-month period. The Pearson correlation between these two variables is 0.31. CSHCI = 0.31, indicating a moderate positive correlation: turbines with larger interval extensions tend to have more degraded sensors. This suggests the CBM model may be interpreting sensor degradation (reduced signal amplitude) as healthy asset condition (normal vibration levels).

Edge Cases and Measurement Boundaries:

Sensor health metrics may not exist for all sensor types. Prioritize vibration sensors and temperature sensors, which have the most established health diagnostic capabilities and the most significant impact on CBM decisions. Short measurement periods may produce random correlations; use at least 12 months of data for meaningful CSHCI computation. Asset age is a confounder because newer assets may have both healthier sensors and genuinely healthier conditions. Control for asset age through stratified analysis or partial correlation to isolate the sensor health effect from the age effect.

Proposed formula, pending validation pilots.

At a Glance

Unit
Correlation (-1 to 1)
Trend Direction
Lower; approaching 0
Reporting Frequency
Annually
Category
Condition-Based Maintenance Integrity
← CBM Interval Outcome Validation Rate (CIOVR)Interval Extension Coverage Gap (IECG) →

Assess Your Organization

The AI-GPS Assessment evaluates your AI governance maturity and recommends KPIs like this one for your implementation roadmap.

Take the Free AI-GPS Assessment