Adaptiv temperaturreglering av bostadshus
The control of indoor temperature must be able to handle large time constants and both measureable disturbances like outdoor temperature, and non-measurable disturbances, like waste heat from appliances. Due to cheaper microcontrollers (small computers with designated tasks) with better performance, more advanced and computation-demanding methods for control can be used.In this thesis, Generalized Predictive Control (GPC) has been implemented and tested on a theoretically derived model of a house. GPC states that a model of the system is estimated in every sample, and the control is carried out simultaneously, based on predictions of the model. Disturbances like solar radiation can be estimated and used in the control as well. The control signals are determined by quadratic programming, which makes it possible to consider conditions, for example limitations on the control signals.GPC requires a number of parameters to be defined, like model order and control signal penalty, but is relatively robust for variations in them. Feed-forward by measuring or predicting the outdoor temperature did not give any better control due to the fact that the feedback is relatively fast, but better control can be achieved by predicting the solar radiation.