Definition
Predictive analytics applies statistical or machine-learning models to historical and current condition, event, and operating data to estimate how an asset is likely to behave, for example, a probability or time estimate for future degradation. It is the analytical method predictive maintenance decisions can be built on, not the maintenance strategy itself.
Key facts
- Produces an estimate or probability about future behavior, not just a record of past or current condition.
- Relies on historical data volume and quality; sparse or inconsistent history limits model accuracy.
- Distinct from predictive maintenance: predictive analytics is the analytical method, predictive maintenance is the strategy that acts on its output.
- Outputs typically need a defined threshold or business rule to translate into a planned maintenance action.
How does Predictive analytics work in maintenance?
Predictive analytics takes historical failure records, condition-monitoring data, and operating context and fits statistical or machine-learning models to estimate future outcomes (such as remaining useful life or failure probability within a time window) rather than only describing the present state.
What does Predictive analytics require?
Predictive maintenance is the maintenance strategy that plans work close to actual need; predictive analytics is one of the analytical methods that can generate the forecast it relies on. Anomaly detection instead flags deviations from expected patterns without necessarily forecasting a specific future outcome: predictive analytics is forward-looking estimation, anomaly detection is present-state deviation flagging.
FAQ
Frequently asked questions
How are future developments estimated from data?
Predictive analytics fits statistical or machine-learning models to historical and current data to estimate a future outcome, such as failure probability or remaining useful life, which a predictive-maintenance decision can then act on.
Last reviewed: