Title: "Fault Detection, Classification and Protection in Solar Photovoltaic Arrays"
Abstract: Fault analysis and fault detection are important to the efficiency, safety and reliability of solar photovoltaic (PV) systems. Despite the fact that PV systems have no moving parts and usually require low maintenance, they are still subject to various fault conditions. Especially for PV arrays (dc side), it is difficult to shut down PV modules completely during faults, since they are always energized by sunlight in daytime. Furthermore, conventional series-parallel PV configurations increase voltage and current ratings, leading to higher risk of large fault currents or dc arcs.
This dissertation reviews the challenges and limitations of existing fault detection and protection solutions in solar PV arrays. For the first time, a 35kW commercial-scale PV laboratory is designed to study faults under real-working conditions and to discover the "blind spots" in conventional fault protection schemes. It is shown that the line-line fault may not be detectable by traditional overcurrent protection devices (OCPD) under certain conditions. Therefore, the fault may remain in the PV system as a safety concern.
To eliminate the detection "blind spot," outlier rules, such as statistical outlier detection rules and local outlier factors (LOFs) are proposed in PV-string monitoring systems. To further identify the fault types (or so-called fault classification), machine learning algorithms are studied in solar PV arrays. To overcome the drawbacks of supervised learning algorithms, a semi-supervised learning algorithm is proposed. The dissertation demonstrates the effectiveness in fault detection and classification in both simulation and experimental results.
Advisor: Professor Brad Lehman
Professor Jennifer G. Dy
Dr. Peng Li