Forecasting of Intelligent Thermal Performance in Two Types of Solar Air Heater Using Artificial Neural Networks

Abstract

Applying solar collectors is a popular tool to harness solar energy. In this research, a flat plate solar air collector with two types of glass cover, including slatted and flat, was investigated under direct solar radiation. The study was conducted to evaluate the capability of Perceptron Neural Network for modeling and predicting the efficiency of heat collectors by input parameters, input fluid mass flow, inlet and outlet air temperature from collector, temperature of the absorber, its thickness and porosity, and also solar energy. The tests were conducted in three replications on very clear sky days during 11 to 13 O′ clock (average solar energy was reported to be 1040 Wm-2 during the interval). Values obtained from tests were compared with the predicted values of the neural network. According to obtained coefficient of determination, for flat (0.98) and slatted (0.99) glass cover, it has been concluded that using ANN is an accurate method to predict the thermal performance of solar air collectors.