The Accurex™ SE is ACOEM’s Smart Exception AI (Artificial Intelligence for Anomaly Detection). It’s a new feature integrated into the Sparrow wireless IoT sensor and accessible via the NEST Predictive Maintenance Software platform. Its core task is to provide reliable fault detection and precisely identify abnormal developments in machine behavior.
Technology and Functionality:
Accurex™ SE uses Unsupervised Learning. Unlike Supervised Learning, it doesn’t require pre-qualified diagnoses; instead, it identifies patterns in untagged machine data. This makes it ideal for monitoring complex systems or machines with rapidly changing operating conditions, such as wind turbines or stamping presses.
The system initially self-trains using individual machine data, which can include vibration and various process data. The reference behavior is established during an automatic learning phase. A crucial feature is that no manual alarm thresholds need to be set, which prevents false alarms caused by poor threshold configuration. The technology significantly reduces false alarms (by a factor of 4).
Accurex™ SE features Smart ON/OFF detection, ensuring analysis only occurs during the machine’s actual operation. Furthermore, the AI learns continuously: it can also process user feedback on false alarms to reinforce the reference model and increase accuracy.