The Misattribution Problem
When your predictive maintenance model flags an anomaly, it’s telling you something is off. What it can’t tell you is whether the problem is the machine or the sensor watching the machine. A drifting thermocouple looks identical to a bearing running hot. An aging vibration sensor mimics early-stage gear mesh defects. And your reliability engineers end up spending investigation hours chasing instrument problems that have nothing to do with asset health.
This is the single biggest source of false alarms in PdM programs — and no standard KPI measures it.
Sensor-vs-Asset Degradation Disambiguation
Sensor-vs-Asset Degradation Disambiguation (SADD) is a proposed KPI that quantifies how cleanly your analytics stack distinguishes instrument faults from actual asset degradation.
The formula evaluates each anomaly alert using cross-residual analysis — comparing what one sensor reads against what its physically correlated sensors predict it should read. When a single sensor deviates while its peers stay stable, that’s a sensor fault. When all correlated sensors trend together, that’s real degradation.
The Purity Score
The formula takes correctly attributed alerts — sensor alerts confirmed as sensor faults plus asset alerts confirmed as asset degradation — and divides by total investigated alerts. The result is a purity score between 0 and 1 that tells you whether your PdM system is solving the right problems.
The Data Already Exists in Maximo
If you’re running Maximo 9 and MAS, every data point in this formula maps to objects you already have: Condition Monitoring measurement points, ASSETMETER readings, MAS Monitor anomaly scores, calibration work orders (WORKTYPE=’CAL’), and FAILUREREPORT closure codes.
Full Deep Dive
The full article covers the complete formula with worked examples, the full Maximo field mapping, the supporting research from Isermann and Carvalho et al., and a concrete pilot plan for testing SADD in your environment.
Read the complete deep dive on Sensor-vs-Asset Degradation Disambiguation — formula, worked examples, Maximo field mapping, research foundations, and pilot plan.
Read Full Article on LinkedInAbout the Advanced KPI Series
This is part of the Advanced KPI Series — 45 proposed metrics for AI-era maintenance. Each article introduces one KPI that doesn’t exist yet in the industry, explains why it’s needed, shows where the data lives in Maximo, and invites the community to test it.