Key Objectives

Develop the interoperable plug-n-play Smart Energy Edge Box (SEEB)

Discover and abstract the available edge resources in a real-time manner
Design efficient lightweight AIs for the network edge
Develop federated AI-optimization mechanisms for continuous building energy performance for climating applications

Summary

Enhancing Smart Building Renovation with Distributed AI for Energy Optimization

The SEED4AI concept attempts to “edgify” traditional IoT-based building smart renovation strategies, that
exploit conventional low-cost commercial infrastructure. SEED4AI aims to automate building optimization
and control (BOC) for energy performance purposes. More specifically, the SEED4AI BOC system will
leverage lightweight distributed AI optimization schemes and topologies for sensing and load-control
purposes. The optimization will consider the overall performance (accumulated performance of all training
agents) of the optimized (in a distributed manner) models used for forecasting specific building attributes and
for emulating the applied control policy (actor) for the domestic energy-intensive loads (e.g., HVACs). The
overall/accumulated optimization criterion/reward will consider minimizing the corresponding energy
consumption without violating standardized thermal comfort bounds.

Work Packages

WP1

Overall Research and Development Framework Requirements

WP4

Integration, Pilot Demonstrations & Impact Assessment

WP2

Edge Resources Configuration, Abstraction and Management

WP5

Communication and Exploitation

WP3

Virtual Sensing, Control-Policy and Distributed Policy Optimization at the Edge

WP6

Project Management