Smart optimization and investigation of a PCMs-filled helical finned-tubes double-pass solar air heater: An experimental data-driven deep learning approach

초록

Solar Air Heaters (SAHs), widely utilized for space heating and drying in residential and commercial settings, exhibit performance dependent on several operational parameters, including solar radiation, ambient air temperature, exposure time, and temperature absorbance. Addressing these variables, a Deep Neural Network (DNN) was developed to predict the heat output of a solar air heater system. The research included a double-pass SAH equipped with circular and semi-circular fin-based aluminum tubes, filled with RT44HC and RT18HC as energy storage materials. Parametric data from three different heater configurations were collected to train and test the model, which yielded accurate heat output predictions (R2 = 0.995, RMSE = 0.018). The developed DNN model was then combined with a Genetic Algorithm (GA) to determine the optimal parameter combinations that achieved an efficiency of 76.328 %, surpassing the average efficiency of 64.51 % obtained without optimization. Furthermore, an Explainable AI (XAI) approach was utilized to analyze the influence of each input parameter on the system's heat output. The results demonstrated that temperature rise was the most influential factor impacting heat output, followed by absorber plate temperature, heat input, exposure time, ambient air temperature, and solar radiation. The dependence analysis further explains the relationship between input variables and how the interaction between the input parameters can affect the output, providing valuable insights for optimizing solar air heater systems.

키워드