Control and optimization of air- conditioning systems - Intelligent Buildings and Building Automation - page 170

used to tune the model parameters to make them represent the actual system.
The primary role of the optimization technique is to seek the energy-efficient
and/or cost-efficient control settings (i.e. operation mode and set-points) to
minimize the system energy input or operating cost while still maintaining
satisfactory controlled variables. At a sampling instant, the optimization
technique is applied to these models to evaluate the control settings that
minimize power consumption and/or operating cost as characterized by
the models. The control strategies determined in this manner react quickly
to the rapid changes of internal and external conditions. According to the
knowledge of the system utilized to formulate the models, the model-based
optimal control can be further divided into physical model-based optimal
control, grey-box model-based optimal control and black-box model-based
optimal control.
8.5.3 Hybrid optimal control methods
In hybrid optimal control, different types of models and/or model-based
control method and model-free control method are combined together to
formulate the optimal control strategies. For instance, some hybrid optimal
control methods utilize a mix of physical/grey-box/black-box models to design
the control system, in which some component models are physical models
while others are grey-box or black-box models. Some hybrid optimal control
methods use both the model-based approach and model-free approach (e.g.
reinforcement learning approach) to construct the optimal control methods,
in which the features of model-based approach and model-free approach are
combined to achieve high control performance. The optimal control meth-
ods formulated in this manner may provide good control performance if the
controllers are designed reasonably well.
8.5.4 Performance map-based optimal control methods
Compared to the three methods presented above, performance map-based
optimal control is somewhat different. This method often uses results
generated from the detailed simulation (or experimental tests) of the tar-
geted system over the range of expected operating conditions to draw a
performance map, and then utilizes this map for optimal control of HVAC
systems. For instance, various combinations of cooling loads, ambient air
temperatures, the number of operating chillers, the number of operating
pumps as well as the number of operating cooling towers and their indi-
vidual fan speeds can be used as inputs to the simulation platform based on
detailed component models of an electrically driven chiller plant without
significant thermal energy storage. At each operating condition, the power
consumption or performance data for all combinations are computed, and
the control settings giving minimum energy value or best performance are
identified. A performance map can then be drawn using those combinations