CWA 17492:2020

Predictive control and maintenance of data intensive industrial processes CWA 17492:2020

Scope

This document contains a methodology detailing the machine/deep learning techniques that should be employed, through the different steps to be followed, with the aim to predict industrial processes or equipment drifts and trigger alarms and potentially help to improve overall equipment effectiveness or the workshop performances.
NOTE The triggered alarms are related to the process in such a way a small deviation affecting the production can be detected in advance, but these alarms are not related to safety.
This document can be used as a guide by:
- Manufacturing plant managers: it contains two examples of real use cases that show the possibilities offered by machine/deep learning techniques applied to the control and optimization of manufacturing processes and to the predictive maintenance of plant machinery;
- Data Scientists: The actual use cases shown reflect the problems they will face when applying these techniques in an industrial environment, which has its own characteristics.