Understand Anomaly Detection

Anomaly Detection depends on a detector on the pipeline that can register passage of the anomaly. Normally, this is a pig detector. An example of a pig detector is one that detects passage using the acoustic emissions of the passing pig. The Remote Terminal Unit (RTU) monitoring the detector must then have a method of signaling to the AVEVA Enterprise SCADA host that the pig has passed, normally using a point polled by a status or multistate record in the RealTime database (RTDB).

A script or process must be created that will be triggered by the state change at the status or multistate and have reference to an associated Adetect table record located in the Advanced Database Editor (ADE). The script or process timestamps the adetect record when a detection event has occurred.

All items in the above paragraph are external to the Anomaly Detection application, except for the reference to an adetect record.

The Anomaly Detection application monitors the Adetect table, checking for new detection event timestamps. When a new timestamp is received, arrival of an anomaly has occurred. If the calculated position is outside volume or time deadbands, an alarm results and a new anomaly is inserted marking the position where the anomaly was detected. Anomaly Detection will also check for anomalies that pass the deadband without detection. A missed alarm is issued if this happens.