LASSO based Building Thermal Model for Heating, Ventilation and Air-Conditioning Control

Sophisticated controllers using building models and effects of disturbances are required for saving energy in buildings. Occupancy and climate are the disturbances acting on the building thermal model. This investigation presents a procedure for constructing building thermal model using experiments, simple sensor and least-absolute shrinkage and selection operator (LASSO). The thermal model so obtained is simple and can be used to construct model based controllers. Further, the methods leads to a linear model as against the non-linear thermal building models making it easier to design model based controllers. The building model is illustrated on a heating, ventilation and air-conditioning (HVAC) test-bed in a laboratory. Our results show that the LASSO based captures the non-linear disturbances and improves the model accuracy. The obtained results are compared with the actual measurements and least-squares to illustrate the accuracy.