Assessing the grid-connected capacity of county-level photovoltaic systems: A study to enhance flexibility and reliability

Abstract

As photovoltaic (PV) systems become more widely adopted, maximizing their grid-connected capacity in urban settings is crucial for improving energy flexibility and reliability. While progress is being made in integrating PV with renewable energy sources, further research is required to comprehensively understand how urban PV systems can support the grid, particularly regarding flexibility and reliability. Most prior studies have concentrated on centralized, grid-level evaluations, often neglecting urban-specific variances and the real-time challenges of grid management. Additionally, the dynamic nature of PV power production and consumption patterns in rapidly urbanizing areas poses difficulties for conventional approaches. This study aims to evaluate the grid-connected capacity of PV systems in urban environments by utilizing advanced deep learning (DL) algorithms to optimize and adjust grid connection flexibility and reliability while predicting electricity demands in these areas. The KO-DynSpikNet model is employed to analyze historical solar power data, consumption trends, and grid stability indicators. The model is trained to estimate optimal PV capacity integration for various urban locations, taking into account weather variability, changes in energy demand, and grid constraints. Sensors are utilized for data collection, while discrete wavelet transform (DWT) and min-max scaling are applied for preprocessing and feature extraction. The results indicate that the DL model offers improved forecasting accuracy for grid-connected capacity compared to conventional methods. Enhanced PV system designs contribute to grid stability and increase flexibility in real-time energy distribution. This showcases how local energy networks can become more flexible and reliable while facilitating the adoption of environmentally friendly energy solutions in decentralized systems.