Search documents

Browse topics

Document details

Performance Evaluation of Distributed Solar PV Installations
D.-E. Archer, G. Almondo, N. Fatima, M. Ammar Khan
PV Systems and Storage – Modelling, Design, Operation and Performance
Subtopic: Operation, Performance and Maintenance of PV Systems
Event: 37th European Photovoltaic Solar Energy Conference and Exhibition
Session: 5CV.3.52
ISBN: 3-936338-73-6
0,00 EUR
Document(s): poster


The aim of this work is to develop a cost-efficient monitoring system for automatically detecting faults in PV using only measured production data from multiple PV sites. The method is suitable for small, distributed PV systems, since alternative monitoring solutions relying on data from physical sensors or inspection in the field are expensive relative to the profit margin of individual PV systems. The monitoring system works by nowcasting the expected energy production for each site based on the measured production of neighboring sites, with a technique called half-sibling regression. Significant deviation between the observed and expected energy production is taken as a sign of a fault. Half sibling regression accounts for environmental factors that influence energy production at a scale of about 10 kilometers, such as the sun's position in the sky, air temperature, mesoscale and synoptic scale weather systems. The advantage of this technique is that these environmental circumstances do not need to be explicitly measured or modeled. For detecting faults on a minute time scale we apply Matrix Profile, an efficient and recently developed technique for anomaly detection that to our knowledge is novel within the PV community. The system will be used to monitor thousands of PV systems installed in Sweden, with the aim of early fault detection and maximization of production.