DB Energy engine primarily consists of a neural network model, enhanced through another ensemble method to get the maximum benefit.
- The neural network model used in energi.ai is an LSTM (Long Short-Term Memory) model that is one of the best machine learning model types to learn from multivariate time-series data.
- The other model in use is an Autoencoder which is essentially driven to detect anomalies from the KPI data received.
- While both the above models improve over time as they learn from the specific environment they are deployed at, the Autoencoder improves significantly over time.
- The models mentioned above deliver the primary feed for the actions and itS further strengthened by the Ensemble Method used to improve the accuracy of the output.