A Method for the Estimation of Missing Strings in Very-Large-Scale Photovoltaic Power Plants

This article presents a novel methodology to detect missing strings in very-large-scale photovoltaic (VLSPV) systems, utilizing only data acquired at the stringbox level. Leveraging data analysis and unsupervised machine learning techniques, the proposed method estimates the quantity of missing strings per stringbox by comparing the direct current from each stringbox with neighboring stringboxes within the same region. The approach addresses a gap in the existing literature by providing a solution tailored to the typical instrumentation level of VLSPV plants. The work encompasses an analysis of the energy losses caused by missing strings, quantifying the impact on the overall system performance. Evaluation against real-world data showed a precision of around 90% of the proposed method in detecting missing strings. The findings offer valuable insights for operations and maintenance teams, enabling identification and mitigation of problematic strings in VLSPV plants.