Redes neurais para previsão de radiação solar em Manta – Equador

In Ecuador, solar radiation and wind speed data are not always available for places of interest due to the absence of meteorological stations. Within the scope of this work, a low-cost automatic weather station prototype based on Raspberry technology was developed to measure the variables mentioned above. The objective of this work is twofold: a) to present a project proposal for a low-cost automatic meteorological station using the Raspberry Pi microcomputer, showing the viability of this technology as an alternative for the construction of an automatic meteorological station, and; b) to use neural network forecasting to predict solar radiation in Manta, Ecuador, based on historical data collected: solar radiation, wind speed, and direction. We prove that technology feasibility and machine learning have a high potential as a tool to be used in this field of study.