Empirical Validation of Photo-voltaic Power Output Forecasting Methods

The electricity produced by renewable energy sources (RES) is constantly increasing worldwide thanks to government policies and technological advancements. Photo voltaic cells are 

actually a fast rising technology as the source of its power is a commodity that is abundant in 

this part of the hemisphere. Uncertainty also dominates in the area of the conditions at which 

the photovoltaic cells are operating, their optimal weather conditions and the effect of the 

conditions on their output. In this project the existing forecasting techniques for generation 

and consumption were studied so that they can be used in the multi-agent power grid control 

strategy, which will be developed in the upcoming tasks. The type of Forecasting method 

used in the project was the Numerical Weather Predictions model which uses meteorological 

data such as Irradiance, Ambient Temperature and with the aid of Computer models such as 

Artificial Neural Networks to model the power outputs of PV systems. These forecasts can be 

used to enhance knowledge in the area of power allocation in PV systems as a form of 

Renewable Power sources. The power outputs used in the project i.e Measured Power output 

and the Predicted Power Outputs were quite similar which means this model can be safely 

used as a mode of determining the amount of PV power required for a location and the power 

it is likely to produce at different times. The project concluded that this form of forecast 

actually produced values with high correlation values which means it can be used practically 

as a form of power sharing when PV systems are used as grided form of electricity