Solar Power Generation Prediction Using a Lightweight Machine Learning Model for Edge Computing

Solar energy has become a crucial renewable energy source, requiring dependable prediction techniques for managing the power grid and utilizing energy. Therefore, accurately forecasting solar power generation is crucial for optimizing the grid, reducing inefficiencies, and improving the dependability of electricity transmission. Nevertheless, the inherent random and uncertain characteristics of solar irradiance, which are affected by weather and environmental factors, present substantial obstacles to achieving precise forecasting. This study introduces an efficient machine-learning approach that uses a Temporal Convolutional Network to predict the amount of electricity solar panels produce. The purpose of the model is to be used on network edge devices with limited resources. It takes advantage of edge computing capabilities and outperforms existing state-of-the-art performance research.