Supervision of the Energy Performance of a Multi-Arrays Photovoltaic Plant by means of the Bollinger Bands on Seasonal Energy Datasets

The paper proposes a strategy for supervising the operation of multi-arrays photovoltaic plants, in absence of environmental data. This approach can be applied when the energy values of each array are collected and stored separately, and when the extension of the PV plant is limited because this constraint guarantees that all the arrays are affected by the same solar irradiation and air temperature. Under this limitation, that is usually valid for small-medium photovoltaic plants, the arrays should produce the same energy, if no anomaly/fault is present. The proposed methodology exploits the Bollinger bands which are a financial statistical tool to evaluate the volatility, i.e., the variability of a listed company or currency. In this paper, the methodology based on the Bollinger bands is applied to the energy dataset of every array of the PV plant to investigate the energy behavior and find anomalies, if any.