Abstract

Rakesh Kumar Agrawal, G.K. Agrawal

ABSTRACT

In this study, artificial neural networks (ANNs) with network type feed- forward back propagation used for performance analysis of single-stage vapor compression refrigeration system using refrigerant R134a, which does not damage the ozone layer. An experimental investigation was done to find the role of suction pressure and other variables like inlet temperature to the compressor, delivery pressure, outlet temperature to the compressor, to the heat absorbed at evaporator per kg of refrigerant. Experimentation has been performed under transient as well as steady condition as compressor speed changes due to the fluctuation of voltage and rate of cooling at condenser also varies due to day to day changes in environmental condition. Due to transient condition the conventional analytical approach involves more complicated analytical equation and theoretical assumptions , whereas experimental studies are more expensive and time-consuming , so in this paper an attempt has been made to train (ANNs) with network type feed- forward back prop with suction pressure, temperature inlet to compressor, delivery pressure, temperature outlet to compressor input parameter and heat absorbed at evaporator as output parameter, and network has been successfully trained to predict output, network resembles close to each other withR2=0.9999988, RMSE = 0.201kJ/kg, COV=0.1089%&ANNs with Network type -feedforward back propagation, training function- TRAINLM, adaptation learning function – LEARNGDM, can be successfully applied in the field of performance analysis of simple vapour compression refrigeration system.