DiagFit
Diagfit is a blind mode industrial equipment failure prediction software. Blind mode means that the creation of models is done by learning on the sound data of equipment operation. The principle: • For a given piece of equipment, a model is built on the basis of operating data in normal mode or "healthy data". The model defines the normality space of the equipment. • The data comes from physical sensors in the form of "time series". • The model is validated on the detection of a few occurrences of failures. The number of failure occurrences required to validate the model is much lower than if the model had to learn from all possible failures. Operation in two modes: 1) "BUILD" mode: • Consists of creating prediction models from sound learning data and validating it on a few occurrences of failures 2) ''RUN'' mode under control of the user: • It is in this step that the equipment is monitored by the predictive model. In this phase alerts are triggered if an anomaly is detected. An indication of the type of anomaly is provided if it is listed. • Feedback from the maintenance operator on the alerts raised allows the enrichment of the directory of anomalies. • Prediction results are displayed, can be saved and/or exported.