Search

Physical model chain is a step-by-step modeling framework for the conversion of irradiance to photovoltaic (PV) power. When a model chain is fed with irradiance forecasts, it provides the corresponding PV power forecasts. Despite its advantages, forecasting with model chains has yet to receive the attention that it deserves. In several recent works, however, the idea of model-chain-based solar forecasting has been formally modernized, though the framework was restricted to deterministic forecasting. In this work, the model-chain-based forecasting framework is extended to the probability space, in that, a calibrated ensemble of model chains is used to generate probabilistic PV power forecasts. Using two-year data from eight PV plants in Hungary, alongside professional weather forecasts issued by the Hungarian Meteorological Services, it is empirically shown that the raw model-chain ensemble forecasts tend to be underdispered, but adequate post-processing is able to improve calibration and reduce the continuous ranked probability score of raw ensembles by 20%. Given the fact that uncertainty quantification has a cardinal importance to grid integration, this probabilistic extension of the model-chain-based solar forecasting framework is thought beneficial. • The physical photovoltaic model chains are extended for probabilistic forecasting. • The raw ensemble of model chains results in underdispersed power forecasts. • Linear quantile regression is suitable for calibration of the model-chain ensemble. • The proposed method is also beneficial for deterministic power forecasting. [ABSTRACT FROM AUTHOR]

Irradiance-to-power conversion is an essential step of state-of-the-art photovoltaic (PV) power forecasting, regardless of the source and post-processing of irradiance forecasts. The two distinct approaches for mapping the irradiance forecasts to PV power are physical and data-driven, which can also be hybridized. The contribution of this paper is twofold; first, it proposes a concept and identifies the best implementation of a hybrid physical and machine learning irradiance-to-power conversion method. Second, a head-to-head comparison of the physical, data-driven, and hybrid methods is performed for the operational day-ahead power forecasting of 14 PV plants in Hungary based on numerical weather prediction (NWP). To respect the rule of consistency but still obtain as complete picture as possible, two directives are set, namely minimizing the mean absolute error (MAE) and minimizing the root mean square error (RMSE), and separate sets of forecasts are optimized for both directives. The results reveal that for two years of training data, the hybrid method that involves the most physically-calculated predictors can reduce the MAE by 5.2% and 10.4% compared, respectively, to the optimized physical model chains and the machine learning without any physical considerations. The two most important physical modeling steps are separation and transposition modeling, and the rest of the physical PV simulation can be left to machine learning in hybrid models without a significant increase in the errors. The optimization of the physical model chains is found to be important even in the case of hybrid modeling; therefore, it should become a standard procedure in practical applications. Finally, the hybrid method is only beneficial for at least one year of training data, while in the initial period of the operation of a PV plant, it is advised to stay with optimized physical modeling. The guidelines and recommendations of this paper can help both researchers and practitioners design and optimize their power conversion model to increase the accuracy of the PV power forecasts. • New concept for hybrid physical and machine learning irradiance-to-power conversion. • Comparison of physical, data-driven, and hybrid approaches for two directives. • Physical global tilted irradiance calculation is sufficient in hybrid modeling. • Optimization of physical model chains is important even in hybrid models. • The recommended method depends on the length of the available training data. [ABSTRACT FROM AUTHOR]

[Display omitted] • Photovoltaic power curves are used to illustrate the forecast error propagation. • The power forecast accuracy of PV plants depend on their design parameters. • A tilt angle of 45° increases forecast errors by up to 49% compared to a horizontal plane. • Undersizing the inverters by a sizing factor of 1.5 reduces forecast errors by up to 25% • The expected forecast errors should be considered in the PV design optimization. The growth of the installed photovoltaic (PV) capacity calls for accurate power forecasts, which are commonly calculated from irradiance forecasts using physical model chains. This study analyses the impact of the PV plant design parameters on the additional error of the irradiance-to-power conversion using three approaches. First, PV power curves are plotted for different design parameters to illustrate the amplification of errors and to enhance the understanding of the presented phenomena. Second, six forecast accuracy metrics are calculated as a function of the design parameters for the irradiance observation and forecast data of ten stations from five climate zones and two forecast providers. Third, the Pareto-optimal tradeoff between the annual energy production and the absolute forecast errors are obtained by the NSGA-II multi-objective optimization algorithm. The results reveal that the design parameters highly influence the PV power forecast errors, which is important to consider in all studies presenting PV power forecast verification. A tilt angle of 45° increases the error metrics up to 49% compared to a horizontal plane, and undersizing the inverters by a 1.5 sizing factor lead to an average error decrease up to 25%. Including the expected forecast errors as an objective during their design optimization process is important to maximize the net revenues of PV plants, especially if the unit imbalance costs are more than 50% of the electricity prices. [ABSTRACT FROM AUTHOR]

The increase of the worldwide installed photovoltaic (PV) capacity and the intermittent nature of the solar resource highlights the importance of power forecasting for the grid integration of the technology. This study compares 24 machine learning models for deterministic day-ahead power forecasting based on numerical weather predictions (NWP), tested for two-year-long 15-min resolution datasets of 16 PV plants in Hungary. The effects of the predictor selection and the benefits of the hyperparameter tuning are also evaluated. The results show that the two most accurate models are kernel ridge regression and multilayer perceptron with an up to 44.6% forecast skill score over persistence. Supplementing the basic NWP data with Sun position angles and statistically processed irradiance values as the inputs of the learning models results in a 13.1% decrease of the root mean square error (RMSE), which underlines the importance of the predictor selection. The hyperparameter tuning is essential to exploit the full potential of the models, especially for the less robust models, which are prone to under or overfitting without proper tuning. The overall best forecasts have a 13.9% lower RMSE compared to the baseline scenario of using linear regression. Moreover, the power forecasts based on only daily average irradiance forecasts and the Sun position angles have only a 1.5% higher RMSE than the best scenario, which demonstrates the effectiveness of machine learning even for limited data availability. The results of this paper can support both researchers and practitioners in constructing the best data-driven techniques for NWP-based PV power forecasting. • 24 machine learning models tested for day-ahead photovoltaic power forecasting. • Kernel ridge and multilayer perceptron are the overall most accurate models. • Predictor selection is even more important than model selection. • Hyperparameter optimization is essential for the highest accuracy. • Up to 13.9% RMSE improvement over a baseline linear regression model. [ABSTRACT FROM AUTHOR]