Estimation of medium flow through the fan coil system

Mjerni podaci prikupljeni su u okviru sustava za grijanje i hlađenje na 9. katu C zgrade Fakulteta elektrotehnike i računarstva u Zagrebu u periodu od mjesec dana. Provedena je identifikacija ventilokonvektorskog modela u stacionarnom stanju temperature izlaznog zraka u prostoriji za svaku brzinu ventilokonvektora posebno. Tražila se ovisnost temperature izlaznog zraka u stacionarnom stanju o temperaturi zraka u sobi, protoku na katnom razvodu te o ulaznoj temperaturi medija. Identifikacija stacionarnog stanja vršena je uz pomoć neuronskih mreža, a modeliranje dinamike korištenjem filtra prvog reda. Konačno dobiveni Hammerstein model može se koristiti kao osnova za upravljanje čitavim ventilokonvektorskim sustavom.This work was based on the identification of the steady state model of one fan coil unit that is present in the HVAC system of the 9th floor of the C building of the Faculty of electrical engineering and computing, University of Zagreb. The model tried to map the connection and dependence of the output fan coil air temperature to the room air temperature, medium flow and input medium temperature. The identification was performed using neural networks and first-order discrete filter on the basis of samples that were recorded in the HVAC system over the period of a month. Obtained Hammerstein class model can be used as a basis for the control of the entire fan coil system