Methods for selecting predictive maintenance alarm thresholds

In predictive maintenance, the definition of alarm thresholds is crucial for detecting anomalies before they lead to critical failures. Well-defined thresholds make it possible to identify abnormal signals while minimizing false alarms, which can lead to costly unnecessary interventions or failure to detect critical faults. A balance must be struck in determining the threshold between sensitivity and accuracy, to optimize the various risks (human, financial, environmental, resilience and sustainability). The use of traditional Industry 3.0 methods, and advanced Industry 4.0 and 5.0 methods combining statistics and artificial intelligence, can be used to adjust these thresholds and facilitate decision-making.