engleski

Photovoltaic panel and array static models for power production prediction: Integration of manufacturers' and on-line data

In this paper we develop and verify a static model for a photovoltaic array power production prediction by integrating manufacturers' and on-line data. The static model is a fundamental part of dynamic models that are used to predict the photovoltaic array power production along a prediction horizon, and it is important to assess its limit performance to (i) reach maximum accuracy in the power production prediction, and to (ii) enable monitoring of the photovoltaic plant for alerting the owner in the case of unexpectedly low performance. The static model is developed in two subsequent steps: (i) power production model parameters are identified on manufacturers' data, and (ii) a model corrector is identified on on-line solar irradiance and the photovoltaic array temperature data, which significantly improves accuracy of the static power production model compared to the model identified on manufacturers' data only. Verification is performed on measurements of solar irradiance components, the PV array temperature and the output power during a 17-month time period. The proposed combination of the model initialization with manufacturers' data and on- line data-based correction shows very high model accuracy, and enables adaptation to different system setups. It also incorporates robustness to systematic solar irradiance prediction errors.

PV array static power production model ; PV panel single-diode model ; Implicit equation ; Neural networks ; Statistical analysis ; Fast numerical algorithms

nije evidentirano