Detecting Permanently Activated Bypass Diodes in Utility-Scale PV Plant Monitoring Data

Abstract

Data-driven operation and maintenance (O&M) schemes are needed to secure efficient operation through the full lifetime of a photovoltaic (PV) power plant. In this study, we expand the capabilities of an analytics platform for data-driven O&M by proposing a method for the detection of permanently activated bypass diodes from PV plant monitoring data. The new method is based on a step detection algorithm developed and employed on historical power monitoring data from a 75-MWp dc PV plant. The results are correlated with active bypass diodes detected by an infrared thermography (IRT) scan performed after 5 years of operation. A stepwise drop in normalized yield is detected for 54.6% of string sets with one active bypass diode. The same is only detected in 1% of string sets with no detected IRT signatures. By replacing modules affected by IRT signatures, the detected power loss is recovered, a strong validation of the method. The novelty of the approach is fault detection of small losses with no extra sensor requirements and that through precise detection of the onset of active diode faults, the historical fault rates can be determined with high time resolution.