SEED4AI

The proposed framework considers the development of appropriate distributed lightweight AI tools, intended to exploit the shared resources (data and processing) available from the Smart Energy Edge Boxes (SEEBs) which correspond to low-cost micro-boards.

The foreseen AI tools will have different use purposes, spanning from: i) data-driven soft sensing tools, based on deep learning architectures to complement the existing physical sensing capacity, reduce CAPEX costs and BAC intrusiveness, ii) feedback-based Building Optimization and Control (BOC) – implementing a distributed lightweight RL approach mutually optimized in a distributed manner by a group of synergetic training agents at the edge – to implement load-control policy optimization, based both on the available physical and virtual sensing capacity as well as the shared network processing resources.

The AIs considered for both purposes will be periodically retrained to continuously ensure their self-adaptation to the current situation of the building plant. To avoid delays during the online retraining phase the control deep-actors and effectively exploit the shared edge network resources, the distributed optimization will consider a novel multi-agent gradient-free optimization scheme. It will focus on improving the building energy performance without jeopardizing thermal comfort in a continuous and adaptive manner, periodically refining the control strategy in a model-free manner, according to strategically selected memorized real-life observations. To increase its societal impact, SEED4AI will align its real-life end-use cases with the Greek accommodation situation; focusing on domestic thermostatic heating and RFoperated cooling control applications, as the most energy-intensive end-use types in the European and the Greek building sector.

Project Consortium

The formed consortium has been shaped of sixteen (16) partners from eight (8) countries.The project combines different key players and organisation in order to ensure a multifarious cooperation.

Full title:The Smart Energy Edge box enabling Distributed AIs for low-cost energy management applications

This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No. 101004152