Data-Driven Analysis of Solar Photovoltaic Systems: Correlation and Distribution Patterns

With the increasing number of photovoltaic (PV) systems, it has become increasingly important to understand the behavior of all variables that permeate a system and their relationships to detect anomalies and optimize system performance. This work proposes applying Data Science concepts to three years' worth of data from a 336.96 kWp system located in Campinas, Brazil. One of the challenges seen in the literature was defining what to apply to analyze the data. To this end, a new methodology was devised based on analyzes applied in other works, proposing a way of analyzing PV data. For the PV plant studied, the results showed that there was a reduction in energy generation over the years, possibly due to degradation or soiling in the modules, and the correlation of variables for the inverter model studied. The application of data science techniques can provide valuable information to optimize system performance, increase energy efficiency, and reduce maintenance costs, especially when combined with PV inverters power electronic data.