Performance Analysis of Domestic Refrigerator Using Hydrocarbon Refrigerant Mixtures with ANN and Fuzzy Logic System

Abstract

This paper presents a new methodology for the performance prediction of domestic refrigeration system with hydrocarbon refrigerant mixture (R290/R600a), which is used as a working refrigerant at different weight combinations. Artificial neural network (ANN) and fuzzy logic system (FLS) techniques are used to predict the system performance of such as coefficient of performance (COP). This paper also describes the experimental test setup for collecting the required experimental test data the experimental values are calibrated at steady state conditions. While varying the input parameters like different masses of refrigerant charge, evaporator temperature and varying length of capillary tube. The ANN and FLS models are working under MATLAB toolbox. The back propagation algorithm with different variants and logistic sigmoid transfer function were used in the network. The outputs predicted from the ANN model agree with experimental values with help of coefficient of correlation (R2 > 0.9886), and the percentage of error is less than 5%. In the comparison of performance, results obtained by experimentally and same has compared with the developed fuzzy model with COP are investigated, at all input variants in the system. This result gives that the ANN model gives good accuracy and reliability than the fuzzy logic system for predicting the performance of the domestic refrigeration system.