Asset performance management (APM) is software that learns what normal looks like for each machine from its readings, and raises an alert with the likely fault and a date when the readings drift away from normal. In industry, a plant or a large machine is called an asset. APM is the main software of predictive maintenance.
Consider a gas turbine at a power plant in Gujarat. For months, the temperature of its main bearing stays near 78 °C when the turbine runs at full load. Then it starts to rise by about 0.5 °C every week, although the load has not changed. No single reading is alarming. But APM software has learned from the turbine's own history that this slow rise is not normal. It raises an alert: "Bearing wear likely. Inspect within three weeks." The maintenance engineer plans the inspection for the next scheduled stop, instead of finding out when the turbine trips.
The software learns "normal" separately for each machine, because two identical machines can behave differently. It compares today's readings with the readings this machine gave in the same conditions, such as the same load and the same outside temperature. So a hot afternoon does not set off a false alarm. The date in the alert is an estimate. The software looks at how fast the reading is drifting and calculates when it would reach a level where the machine must stop.
The software gets its readings from three kinds of sources. A process historian stores the readings a plant already collects. An industrial IoT platform collects readings from machines that had no data link. An industrial data platform adds the repair history. The readings are a time series, and spotting a reading that has moved away from normal is anomaly detection.
APM covers three kinds of work. Predictive maintenance sensors add sensors to old machines. An AI maintenance assistant explains an alert and suggests the fix. OEM remote monitoring is sold by the machine's maker.
When APM raises an alert, the repair is booked in field service management software. The alert can also be shown on the plant digital twin. A simpler tool, OEE and machine monitoring, reports how much a machine produced. APM goes further and says whether the machine is about to fail.