Sistema autônomo de sensoriamento remoto espaço-temporal do céu por imagens e sua aplicação em sistemas distribuídos de energia.

Abstract

Cloud cover instability is the main impact and uncertainty factor on the forecast of energy generated in photovoltaic systems. Establishing reliable metrics for large-scale photovoltaic (PV) power forecasting models while considering the uncertainty of climatic variables requires efficient and rapid response models to satisfy the requirements imposed by the complex energy market scenarios and the global economy. However, the massive inclusion of photovoltaic generation in the world electrical system requires accurate and short-term forecasts of the energy generated. This work describes an instrument capable of measuring the cloud base height (ABN) and defining areas of cloud cover and clear-sky over large areas of spatial resolution of the photovoltaic energy generation, the cloud cover sky imager (CCSI). This autonomous tracking system aligns the sun to the center of the camera, according to solar angles. As it scans the sun's movement on the horizon, it captures images of the sky and projects them over an area of interest. The developed method employs a set of engineering techniques of solar tracking, electronic systems, image processing, global positioning and information systems, and omnidirectional camera calibration enable the functionality of the device and validation of method employed. The imaging system uses stereo triangulation with two cameras and pixel matching for sky images reconstruction. The results of stereography validation compared to heights measured by ceilometer and used to measure ABN, showing levels of accuracy with correlation close to the unit, significance below 0.05 and linearity between the data. The system was able to project the shadow of clouds over the area of interest from a geographic information system, covering areas of 6 km on the ground. The technique contributes to the state-of-the-art with a standalone solar tracker that can monitor the sky in a 120° image cone and obtain the reflected surface coverage for any location in the world. This novel technique involves the ability to recognize cloud cover and its impact in large territorial areas. We conclude that this approach provides in practice a fundamental basis for understanding cloud cover and its impact on photovoltaic generation, in addition to valuables contributions to the works that will follow from this.