Simultaneous identification of linear building dynamic model and disturbance using sparsity-promoting optimization

We propose a method that simultaneously identifies control-oriented model of building’s temperature dynamics and transformed version the unmeasured disturbance affecting building. Our uses ? 1 -regularization to encourage identified be approximately sparse, which is motivated by slowly-varying nature occupancy determines disturbance. The proposed involves solving feasible convex optimization problem guarantees black-box model, linear time-invariant system, possesses known properties plant, especially input–output stability positive DC gains. These features enable one use as part self-learning control system in building updated periodically without requiring human intervention. Results from application on data simulated real are provided.