Physics-Informed Autoencoders with Intrinsic Differential Equations for Anomaly Detection in Industrial Processes
摘要
Anomaly detection plays a central role in monitoring industrial processes. The common understanding is that the time series of a variable contains visible traces of anomalies - which can be selected and validated by suited algorithms