Parametric analysis and performance prediction of an ultra-low temperature cascade refrigeration freezer based on an artificial neural network

A thermodynamic analysis is proposed to improve the design and operating parameters in an ultra-low temperature cascade refrigeration system (CRS). The expressions for predicted parameters of the CRS are obtained based on an artificial neural network (ANN) method. In the first part, a parametric analysis is carried out to analyze the influences of seven variables on the COP, total compressor input power, discharge temperature of two compressors, total exergy destruction, and exergy efficiency. The results show that the condensation temperature of the low-temperature circuit has an optimal value, which maximizes COP and exergy efficiency but minimizes the total compressor input power and exergy destruction. In the second part, the performance of CRS is predicted by the ANN model. Eighty sets of input-output parameters are applied as both the training and testing data. It is found that the correlation coefficients for training-testing data in the range of 0.9886–0.9994 are estimated. The mean absolute errors obtained in the prediction for COP, total compressor power, total exergy destruction, exergy efficiency, and discharge temperature of high and low-temperature cycles with the testing set are 0.0027, 0.9090, 1.0314, 0.1691, 1.1438, and 1.0230, respectively. The outcome indicates that the ANN model has good advantages in predicting the cascade refrigeration system's performance.