Expektra Predict is a new method to forecast short-term power consumption and production based on artificial intelligence with machine learning.
Evaluation with reference customers have shown that Expektra Predict has 10-20% better accuracy than conventional methods regarding prediction of electricity demand, corresponding to equal cost reductions over time.
Expektras method is based on a unique implementation of artificial neural networks which are represented by various mathematical models inspired by the network of biological neurons and synapses in the human brain.
One of the key strengths with this mathematical model is the ability to recognize non-linear relations between input- and output data. This property allow Expektra to assess the value of a range of new input variables not applicable with conventional methods.
Another feature is the ability to detect complex patterns in input data indicating the current system ‘state’, which in turn may dictate the current data relations. This feature is particularly valuable as the power system is characterized as a complex dynamic system affected by multiple exogenous variables and seasonally shifting between different states.
Expektra Predict is provided as SaaS (Software as a Service), which means minimal overhead for management, administration and maintenance.
Contact us to get more information.