Identifying a greenhouse climate model by using subspace methods

This paper presents the development of a climate dynamics model for a greenhouse located on the Bogota plateau. A black-box model was estimated from experimental data, considering a novel system structure derived from a first principles model and experimental tests. It considered two control volumes, one given by the air over the crop and a second one formed by the air trapped by crop foliage. The model was selected from a set of linear, discrete-time, state-variable systems using subspace identification methods. The estimated system was able to predict climate dynamics for both control volumes, having errors below 8%. Such performance was comparable to previous work reported in literature while the obtained model was a low-complexity linear system