Power monitoring based photovoltaic installation fault detection

Photovoltaic installations are becoming increasingly popular, roof-top installations have never been as affordable and PV parks are also blooming. Unfortunately, as with all systems, problems can occur. These range from faulty cabling to theft, but also more subtle issues such as breakdown of a single cell and ageing of the installation. The objective of this paper is to present a simple method that can be used to detect such issues. The proposed method only requires two electric power sensors to be added to an existing PV installation. The algorithm relies on a self-adaptive PV model that does not need the user to provide any input regarding model parameters. The PV model is continuously trained so as to reflect the installation status. A continuous monitoring of the produced PV power is performed and a comparison with the expected (i.e. modeled power) is carried out in order to detect discrepancies. In this article, the complete algorithm for fault detection, including the underlying mode, is presented and validated on real data.