Identification of Degradation and Faults in PV Systems using a Time-Varying Efficiency Model

Anton Driesse PV Performance Labs, Freiburg, --, Germany

Reliable detection of abnormal operation and accurate quantification of gradual performance reductions are high on the wish-list of PV system owners and operators. Monitoring systems have been used for decades to take relevant measurements, then store and display the data so that human operators or computer algorithms could identify problems. The number of systems and amount of data has grown steadily so that most or all of this analysis now needs to be done by algorithms. In this work we address the challenge of fitting PV module efficiency models to field measurements. This class of module performance models represents the efficiency, or equivalently the power output, of a PV module as an explicit function of one or more external variables. The main inputs are plane-of-array (POA) irradiance and module temperature. The model parameters are mostly empirical, but some can be linked to physical characteristics. We use our method on several open data sets and demonstrate that composite degradation rates agree with those determinined by other methods; however, this approach allows us to go one step further to distinguish between degradation at low vs. high irradiance, which helps to distinguish between different causes of degradation.