Output Prediction of Photovoltaic Power Generation System with Multiple Uncertainties in Windy/sandy Weather

Accurately predicting photovoltaic (PV) power generation is essential for effective power system scheduling. However, increased incidences of sandy weather in recent years pose significant challenges for PV power plants in desert regions. This study addresses the inaccuracies in PV power forecasting under windy-sandy conditions by conducting wind tunnel experiments to evaluate the impact of varying wind speeds and PV panel tilt angles on power generation. Polynomial equations are developed through polynomial surface fitting to establish a relationship between the power loss rate, sand-blowing wind speed, and panel tilt angle. Latin hypercube sampling is used to extend the fitted surface scenarios, simulating real-world variations in PV power generation to enhance the accuracy and applicability of the model. In addition, the model incorporates three uncertainties related to wind direction, module aging, and airflow effects on PV arrays. A case study utilizing data from a PV power plant employs the Pseudo-similar day selection method, enabling a comparative analysis of the predicted outcomes with traditional methodologies. This comparison demonstrates the accuracy and effectiveness of the proposed method for improving PV power prediction under challenging conditions.