The use of ANN and conventional solar-plant meteorological variables to estimate atmospheric horizontal extinction

Abstract

In the search to optimize electricity generation systems based on renewable energy sources, solar power plants are a clear alternative for reducing pollution on the planet. In particular, concentrating solar power plants with central tower technology supply energy to large population centers and they are generally located at desert sites. Production in these plants drops due to the presence of particles in the environment. These particles are complex to measure and doing so usually requires the use of dedicated, expensive instrumentation. In this work, we present a methodology called Extinction that estimates this attenuation phenomenon utilizing meteorological variables, along with the use of artificial neural networks (ANN). Direct normal irradiance, relative humidity, temperature and pressure have been the only meteorological variables needed to estimate the Extinction. The results from the estimations presented a correlation coefficient value (R) of 0.88 (between the measured and estimated atmospheric horizontal extinction with ANN), a normalized Mean Bias Error (nMBE) of 0% and a normalized Root-Mean Square Deviation (nRMSD) of 7%.