The impact of external variables on the power production of over 1000 photovoltaic (PV) power plants under real-world operating conditions was studied. These PV systems are part of the SDLE Research Center's network, which includes 787 PV power plants in 13 climate zones.
The analysis of these 5 to 15 year-long PV power datasets was driven by a non-relational data warehouse of multiple heterogeneous energy data. We developed a stable, scalable, fast processing, and querying technology based on Hadoop and its NoSQL database tool, HBase.
The system's power output was predicted from a month-by-month fitted linear regression model. Furthermore, a piecewise linear regression model was fitted to the monthly predicted power output values, and the annual change rate was calculated. Finally, through a stepwise Akaike Information Criterion (AIC), the variables that had significant influence on the system change rate were selected and rank ordered.
The rank-ordered contributors from this analysis demonstrate that Koppen-Geiger Climate Zones is the largest factor determining the power change rate, a result not previously acknowledged in the PV research community.
|