collecting data and their influencing factors, generating forecasts from the data and creating a self-optimising control system:
the WICKIE M system from mrm² automatisierungstechnik gmbh uses artificial intelligence for demand-oriented and self-learning control of automation processes in building management.
conventional systems are based on static parameters that are entered once. these can be changed at any time, but only manually. WICKIE M, on the other hand, is based on self-learning control. for this purpose, relevant data is recorded via the sensors. ai algorithms then calculate a forecast based on the recorded data and control the actuators according to this forecast. the decisive factor here is that all relevant parameters communicate and interact with each other through WICKIE M. in this way, up to 20% energy can be saved in building automation.
the intelligence of WICKIE M is based on time series predictions with a neural network. the lstm technology (long short-term memory) makes this neural network very powerful. machine learning algorithms collect the recorded data in a database, recognise patterns in the data, constantly update the calculation models and generate the predictions.
the methodology of WICKIE M is of course transferable to other tasks. time series forecasts can also be used profitably in the production process, for example to predict output quantities and material flow or for predictive machine maintenance. however, to really benefit from our self-learning optimisation, the target definition of the use cases must be very precise.
WICKIE M is our small, clever and intelligent solution for demand-driven, self-learning control, which shows strong performance especially in building automation and energy management.
