The architecture of SMARTIVE platform includes an Ingest layer feeding a data lake, Data analytics layer and finally the user interface.
For the specific data analytic module, related with SMARTCAST product, SMARTIVE has a defined structure in Python shown as follows.
In the python-cast and packages modules SMARTIVE combines technologies of supervised and non-supervised Machine and Deep Learning specifically,
• Regressive Modelling (PLS, gradient boosting, CNN)
• Clustering (autoencoders, SOM)
• Classifiers (ANN, LSTM, RF, Naïve Bayes)
The data structure in those blocks is defined as follows:
The methodology used for supervised algorithms failure analysis is detailed shown as follows, it includes the labelling of fault condition on the historical data.