Welcome to this Deep Learning project page !

Here, we develop an RNN model in order to detect early the forewarning signs of forthcoming machinery hard failure.

The model we train is ready for deployment. Taking vibration sensor signal as input, it is able to raise an alert, would the working conditions deteriorate to an extend that the material is very likely to fail in a foreseeable future, thus indicating that operation shall be stopped and replacement of the faulty part operated before larger damage could be incurred.

The below notebook contains an end-to-end ETL data pipeline plus a whole Bayesian Optimization cycle for the LSTM Autoencoder implemented to fit the bill :

KEYWORDS : Time Series, Anomaly Detection, Tensorflow, Keras, RNN, LSTM, Autoencoder, Bayesian Optimization, MongoDB, PySpark, ETL, Data Pipeline