Optimizing Risk Control for Solar Farm Green Management Using DNI Estimation at the Edge Through a GLE-SVR Learning Model

The integration of solar farms into power networks necessitates a comprehensive analysis of various parameters and conditions to optimize performance and mitigate risks. This paper aims to enhance the deployment and efficiency of solar energy systems by addressing several key aspects. Initially, critical parameters related to Direct Normal Irradiance (DNI), essential for solar energy harvesting, are identified. The impact of natural hazards on solar farms is assessed using Genetic Landscape Evolution (GLE). Additionally, the Topographic Position Index (TPI) is employed to identify low-risk areas for installing solar panels, ensuring both safety and optimal performance. Edge-assisted local processing is considered to support data handling and preliminary analysis at the site, facilitating more efficient information management. Machine learning techniques, including Support Vector Regression (SVR) and Convolutional Neural Networks (CNN), are implemented to forecast DNI. The performance of solar panels is analyzed, considering various environmental and operational factors. The results indicate that Principal Component Analysis (PCA) reveals elevation as a significant topographical factor influencing DNI production in semi-arid areas. The GLE method shows favorable stability in areas prone to erosion, supporting the feasibility of solar panel installations in the southeastern part of the study area. Moreover, SVR proves to be an accurate method for forecasting DNI (<em>R</em> = 0.98). The performance assessment indicates a final yield of 179.3kWh/kWp in August and the highest reference yield of 1195.4kWh/kWp, demonstrating the effectiveness of this approach. The findings of this paper provide valuable insights for the development of resilient and efficient solar power networks, contributing to the advancement of renewable energy technologies.