In my Bachelor's Thesis I inspect the possibilities of MPC control design and validation in Matlab Simulink. I am going step-by-step through the MPC design procedure with simulated plant model. The plant model consists of a room and its heating system with proportional valves to control the water mass flow, thus providing heat for the room. Losses at the facade was taken into consideration (outer walls and window), I assume no heat loss to the neighboring rooms.
The steady-state heat input for the room model is calculated by a formula and then fed to a Simscape thermal network. This network gives the transient response of the room by simulating the charging of the heat capacities, such as air and the mass of the room itself.
I design an MPC controller for the identified plant model. In the design procedure I study the impact of different control parameters to the control performance, such as the sampling time of the controller, the prediction and control horizon or the weights of the cost function.
Then, for a real physical system, an MPC controller is also designed. A paper box with some wood and some brick with reasonable heat capacity is studied. The heating mechanism is substituted by car headlights.
Designing an MPC controller for this system shows the challenges we facing in real applications, such as computing times, effect of measurement uncertainities and disturbances.
If the time spent for modeling and designing of the controller increases the comfort and/or reduces operating costs, the method can be feasible for real applications and integrated to an intelligent building system.