Deploying an artificial neural network model for solar air heating modelling

In this work, we utilize Artificial Neural Networks (ANN) to model the efficiency of solar air heaters. The underlying thermodynamic principles and factors affecting the performance of a solar air heater were considered and used for training and testing to determine the efficiency of the solar air heater. Using an input - output mapping approach, a back propagation learning algorithm based neural network was used to train and test measurable, controlled and conditional factors as input for the modeling architecture. The output was set to be the overall efficiency and required energy by the fan used for forcing the air mass passing through the solar air heater. The network result has an efficiency of 56.17% at the highest air mass flow rate of 0.038kg/s. The variation in efficiency of the neural network was 2.032% and the neural network has reduction of 5.79% in thermal efficiency when compared with physical experimental results.