Fault Diagnosis and Classification in Photovoltaic Systems Using SCADA Data

Abstract

Reliability is a key challenge faced by the fastgrowing photovoltaic (PV) power plants. This paper presents fault diagnosis and classification techniques using the PV plant operational data collected from the supervisory control and data acquisition (SCADA) system. Specifically, the proposed solution consists of three techniques: (1) a new statistical fault detection method; (2) a corrective performance ratio (CPR) based fault isolation method; and (3) an anomaly degree index (ADI) based fault recoverability analysis and classification method. The proposed fault diagnosis solution has been deployed in a real-world 40 MW PV plant. One-year operation demonstrates that the proposed fault diagnosis and classification solution delivers 97% accuracy on fault detection, and 90% accuracy on unrecoverable fault classification.