What is unsupervised learning? Unsupervised Learning (see wikipedia) is used when no historical data is available that involves past faulty behaviour and/or ageing progression. Therefore,
DiagFit is impressively fast and easy to operate. Mostly operated in unsupervised failure prediction mode because companies have no historical failure data from their monitored equipment, the “Build” step takes usually minutes/hours to be completed.
It is possible to save one or several unsupervised failure prediction models generated in this “Build” step. The predictive model can then be moved to “Run” step. In unsupervised failure prediction mode, domain experts are alerted when deviations from the normality space occur, and can then label these deviations as future failures or aging signs. It is possible to import a supervised or unsupervised failure prediction model directly into the “Run” step.
A permanent iterative adjustment phase (as new data kicks-in) closes the process and loops back to the prognostics phase.
- Multiplatform: cloud, edge and embedded
- Possibility to improve the initial non supervised alarm system with expert feedback to create dictionary of incidents.
- Can be easily integrated with IIoT data collection platforms (upstream) and 3rd party applications (downstream)
State of the art architecture
- Based on the latest microservices architecture, the software will automatically scale to cover the diagnosis and failure prediction of your assets as you grow and as the amount of captured data grows
- Rest API for in-bound integration
- MQTT for out-bound messaging and alerts
- Java script web interface with professional UX-UI design