Examining the Application of Neural Networks in Predicting Temperature and Solar Cell Power

Abstract

Cost-effective and efficient engineering of solar cells, in addition to creative design, necessitates a detailed examination of the physics involved in solar energy absorption. Solar energy does not operate at night without storage means such as batteries, and cloudy weather can render this technology unreliable during the day. Since solar cells are used to convert light into electricity, they must be composed of materials efficient in light absorption. Like any other power generation source, solar cells face challenges, including reduced output power with increasing cell surface temperature. With each degree rise in temperature, efficiency can decrease by up to 54.0%, highlighting the importance of addressing and implementing solutions to this issue. This study explores different specifications of solar cells after a general overview and modeling. It investigates methods for estimating solar cell temperature using ambient temperature and solar radiation, comparing them with a proposed neural network approach. If the neural network can achieve lower error rates than the desired thresholds, it would offer advantages over mathematical estimators mentioned in the literature. Additionally, this neural network demonstrates the capability to predict future solar cell temperatures, unlike instantaneous mathematical estimators for temperature and radiation. DOI: https://doi.org/10.52783/pst.558