Optimizing Methodology for Estimating Global Horizontal Irradiance (GHI) Using Solar Photovoltaics’ Output AC Power Measurements

38th European Photovoltaic Solar Energy Conference and Exhibition; 923-928
In this paper, optimization of a model for estimation of global horizontal irradiance (GHI) using PV’s AC power output is proposed such that errors in estimated GHI are minimized. Modification of the developed model is carried out to adopt the methodology into different programming environment and to make it consistent with the available datasets. Results show that the developed model on Python language has Mean Absolute Percentage Error (MAPE) of 5.77% compared to GHI measurements using pyranometers for estimation of one month’s data. Modifications are done in detection of clear sky periods and minimization of loss functions for estimation of orientation in contrast with other methods. During model optimization it was also found that Ineichen/Perez’s Clear Sky GHI model, Direct Insolation Simulation Code (DISC) Model for Direct Normal Irradiance (DNI) estimation and Perez Model for sky diffuse irradiance calculations had minimum MAPE. Moreover, modelling parameters in the model were also adjusted to find best estimates of GHI. It was observed that timestep of 10 mins and past data of 120 days for estimation of orientation weights and error function were suitable. Number of iterations were also reduced to 10 from a starting number of 30 for estimation of GHI considering MAPE and computation times. Therefore, the main purpose of this research is to develop an optimized model that could be commercialized aiming to convert a PV system into virtual GHI sensor providing; information of solar potential and serve as an input data for solar forecasting or performance monitoring.