Otimização e integração de um sistema de armazenamento de energia em baterias e câmera de céu em uma usina fotovoltaica

Abstract

The integration of new equipment in photovoltaic (PV) power plants, such as the battery energy storage system (BESS) and the all-sky camera (ASC), brings improvements and challenges, which are addressed in this work. The BESS transforms the PV plant into a dispatchable energy source, and the ASC enables the forecasting of shading events. In this study, the integration of these devices in PV plants was analyzed on three fronts. First, two scalable and open-source communication systems were developed to collect data from the PV/BESS system with ASC. The data collection requirements were established based on the systems that control the plant and the IEC-61724 standard. Subsequently, the power projection of the BESS was optimized to maximize the profit from energy sales. An exact method (nonlinear programming) and a heuristic method (genetic algorithm) were applied and compared, with the exact method providing an average gain of 6.94% in the profitability of the PV plant over 363 days. Finally, a very short-term irradiance forecasting method was developed, based on photos from the ASC and utilizing a hybrid neural network called Convolutional Neural Network - Long Short Term Memory (CNNLSTM). The model presents a mean absolute error of 46.751 W/m2, and an example of the application of this forecast in smoothing the power supplied to the electrical grid is provided. Thus, the integration of the BESS and ASC in PV plants has demonstrated a reduction in abrupt variations in energy generation, contributing to the stability of the electrical grid and allowing for a greater participation of solar energy in the energy matrix.