NeuroCool ::field tests of an HVAC control algorithm

Innovative control algorithms for HVAC are necessary if the rising costs associated with air conditioning and more specifically cooling are to be remain reasonable. In the present article we briefly describe such a novel model-predictive based controller, and the practical aspects of its deployment. We describe an implementation based on a MATLAB server, that allows a rapid validation. Then the test sites and associated analysis methodology are presented. Finally, the preliminary results obtained on these sites are presented and analyzed. It is shown that the proposed control algorithm is stable with respect to human disturbances and that comfort is controlled as desired. The communication and deployment architecture has been validated and has proven to work as expected.