Digital twin simulation of a Danish school building

An auto-tuning algorithm based on Model Predictive Control (MPC) is developed for radiator thermostats embedded in a Danish school that can generate and replace the experimental thermal data of buildings through a Functional Mockup Unit (FMU) exported form Modelica. To achieve this, the estimation of relevant parameters is conducted through an algorithm that compares the experimental data, the simulated FMU data, and the data obtained from the AutoRegressive eXogenous (ARX) statistical models implemented in Python. The models are used to implement a digital twin of a school building integrating model predictive controller with low-order plant. After validating the estimated parameters, different controllers are tested and tuned.