Training of Multi-layered Neural Network for Data Enlargement Processing Using an Activity Function

WINNER, a German research project, integrates local photovoltaic systems, charging infrastructure for electric vehicles and tenant households, focusing devices with and without smart grid abilities. The project’s goal is to manage the local power grid operations in a way that allows locally produced energy to be consumed locally. This local optimised consumption is done by using currently available devices. Further, we want to analyse accruing data streams and optimise the usage of local devices to manage this time-base shifted consumption scenario by implementing a non-hard real-time processing system. In this paper, we outline the project’s primary objectives from a technical point of view. First, we present “Wohnungswirtschaftlich integrierte netzneutrale Elektromobilitat in Quartier und Region” (WINNER) and some related research projects. We describe the integration tasks, the data sources, and sinks. So, a project overview can be given. Afterwards, we compare our approach and already developed technologies. Requirements are derived from the system overview. According to them, we can outline an architectural view of the core component of data stream processing within this scenario. Finally, the results are discussed, and consequences are drawn.