A Dynamic Feedforward Control Strategy for Energy-efficient Building System Operation. (arXiv:2302.10179v1 [cs.LG])

The development of current building energy system operation has benefited
from: 1. Informational support from the optimal design through simulation or
first-principles models; 2. System load and energy prediction through machine
learning (ML). Through the literature review, we note that in current control
strategies and optimization algorithms, most of them rely on receiving
information from real-time feedback or using only predictive signals based on
ML data fitting. They do not fully utilize dynamic building information. In
other words, embedding dynamic prior knowledge from building system
characteristics simultaneously for system control draws less attention. In this
context, we propose an engineer-friendly control strategy framework. The
framework is integrated with a feedforward loop that embedded a dynamic
building environment with leading and lagging system information involved: The
simulation combined with system characteristic information is imported to the
ML predictive algorithms. ML generates step-ahead information by rolling-window
feed-in of simulation output to minimize the errors of its forecasting
predecessor in a loop and achieve an overall optimal. We tested it in a case
for heating system control with typical control strategies, which shows our
framework owns a further energy-saving potential of 15%.