Load Matching, Grid Interaction, and Advanced Control - Modeling, Design, and Optimization of Net-Zero Energy Buildings - page 312

typically weather variables (temperature, solar radiation, humidity, wind
speed,etc.)andtheinternalloadsduetotheoccupants. Figure6.7 compares
MPC with a conventional control approach. Note that a TES device is
included in the MPC approach, which allows better planning of the use of
energy resources. In general, some kind of feedback action is still required
in an MPC scheme to add robustness to the control strategy.
Fig. 6.7 Model-based predictive control compared to conventional control
Numerous studies have been made on MPC for buildings (Gyalistras and
OptiControl Team, 2010; Henze and Krarti, 2005; Oldewurtel et al. , 2010a;
Oldewurtel et al. , 2010b). Most of these studies have focused on the control
of cooling systems. This is no coincidence; cooling equipment is
electricity-driven, and therefore it can have an impact on peak loads. Since
in most jurisdictions heating is delivered by burning a fuel, such as natural
gas, planning over a time horizon is less of an issue. 3)
In a wider sense, any strategy making use of forecast information and a
model of the building might be called an MPC strategy. With this
understanding, a simple rule-based strategy, such as “if sunny weather is
expected tomorrow, then delay the start of the heating system,” would
qualify as MPC. However, the term MPC is usually reserved for strategies
involving some kind of formal mathematical optimization.
In MPC, an objective function , J , (typically total cost over a control horizon
H ) is defined.