Grey Box Modelling of Supermarket Refrigeration Room

Abstract

Aiming to enable robust large-scale fault diagnostics and optimized control for supermarket refrigeration systems, a data-driven grey box model for cooling rooms and cabinets was developed. The analysis scopes a single cold room in a supermarket in Otterup (Denmark) and was done using one-minute of sampling data. A resistance-capacitor diagram of the room was analyzed to derive three state-space equations for the model - the following were the states: the room temperature, the temperature of the goods and the refrigerant mass in the evaporator. The model parameters were then estimated using a Kalman filter and the maximum likelihood method. In the present paper, the resulting model is demonstrated through a five-hour simulation and the importance of ongoing re-estimation of parameters is highlighted, as the dynamics of the room constantly change, as goods are added and removed. Furthermore, the physical meaning of the parameters is discussed and a case where the parameter estimates became physically meaningless is highlighted - suggesting that robustness was an issue and further studies with simpler models and other solver algorithms are necessary for large-scale implementation.