Optimisation of Generation Models for Clusters of Photovoltaic Plants

The present work analyses the actual production profiles of a group of tens of thousands of PV plants. Actual PV generation profiles are represented as hourly average powers gathered from the energy meters measured at the point of common coupling with the grid. After filtering, data cleaning and statistical analysis, reference PV plants are selected for the improvements of the PV generation models. Production profiles are calculated by using literature generation models, with weather data as inputs. An optimisation is performed on the parameters of the literature models to minimise the differences between the cumulative distribution functions of calculated profiles and measured data. The study compares the performance of the different models and shows how the optimisation increases the quality of the calculated profiles.