Frost prediction on evaporator coils of supermarket display cabinets using artificial neural networks

Defrosting in supermarket refrigeration systems is normally controlled by a preset time cycle with most display cabinets timed to defrost every 6 hours. It is widely acknowledged that timed defrost may cause a number of unnecessary defrost cycles and this reduces the energy efficiency of refrigeration systems as well as the accuracy of temperature control of the cases. This paper investigates the possibility of modelling the amount of frost on the coil by using neural networks and proposes a demand defrost method based on it which should overcome the disadvantages of other demand defrost approaches.