Imaging-Based Super-Resolution Reconstruction of Volatile Photovoltaic Generation Curves

High-resolution electrical data is a foundational element of data-driven analytics in smart grids. Currently, due to constraints related to measurement frequency, communication costs, and data storage, high-resolution data is often subjected to compression, leading to the loss of high-frequency details. This paper proposes a super-resolution scenario generation method for electrical data to enhance low-resolution photovoltaic (PV) data to high-resolution. First, the one-dimensional electrical data is converted into a two-dimensional electrical image using the Gramian Angular Field (GAF). Then, the low resolution images is enhanced to a high resolution images using a generative adversarial network (GAN). To capture the dependency between PV data and weather conditions, weather data is incorporated into the GAN-based model alongside the low-resolution image. The proposed method has been evaluated using public datasets and compared with traditional data reconstruction methods. The results show that the proposed method not only recovers the high-frequency details of the data with smaller errors, but also generalizes well to different datasets in the same region, yielding satisfactory results.