The Forecasting algorithms predict future values on the basis of the system’s historical data and sensor readings. Our advanced forecasting algorithms can cope with data seasonality and predict energy or market demand trends.
1.Control slowly-adapting systems
4.Optimization by Simulation
5.Find Business Opportunities
The Artificial Intelligence can find complex relations in data to automatically produce accurate mathematical models.
Here we see how our Deep Neural Network predict the future temperature of a large building. This prediction is then be used to manage the energy budget and drive the HVAC systems.
The graph above depicts the true value (blue line) against the predicted value (orange line); as can be seen the orange line follows the blue line closely and the error is quite low. However it is possible to improve the model, for example if the train error is too low compared to the test error, it could be helpful include a form of regularisation inside the model.
For a more technical information check the Jupyter Notebook here