Jesús Febres, Raymond Sterling, Marcus Keane
Year:
2015
Bibliographic info:
Building Simulation, 2015, Hyderabad, India

This paper presents a new approach to calibrate air handling unit models. This approach studies every heat exchanger component separately based on the inverse problem framework, the Preisach model of hysteresis and machine learning techniques. For each component model, the first step is to solve the inverse problem in order to calculate the optimal control signal that generates the output values expected from real data. Then, a modified Preisach model is calibrated using machine learning techniques where the input-output pair samples correspond to the actual control signal taken from the real data and the optimal values obtained from the previous step. The last step is coupling both the first principle based model of the heat exchanger and the calibrated Preisach model. A detailed case study is presented.