A robust modeling for predicting and optimizing the system production is critical to increase the interaction of PV systems with smart electricity grids and to optimize energy use, delivery and storage.
In this context, forecasting models for the PV production are developed and validated through Solar Tech laboratory measured data and in order to assess forecast applicability to perform and optimize energy injection/trading flows.
The state of the art forecasting technique include physical and stochastic method in the unique PHANN method (Physical Hybrid Artificial Neural Network), developed in Politecnico di Milano, which is providing top results in PV systems output power forecast (daily NMAE = 3.39%).
Forecasting the power production from renewable energy sources (RESs) has become fundamental in microgrid applications in order to optimize scheduling and dispatching of the available assets.
This hybrid forecasting technique has been successfully employed in the daily operation of micro-grid in the short-term time horizon.
A full dataset of PV power measurements from SolarTech Lab is available here
Measured power and the hourly average power computed according with the proposed validation process. The blue line represents the available power measurements. Power production hourly averages computed using all the available data (yellow line) and filtering the measurements according to the proposed methodology (orange line). The lower graph shows the acceptance criterion for the single power measurements.