COMPARAÇÃO DE MODELOS PREDITIVOS PARA O NÍVEL DE RADIAÇÃO NO MUNICÍPIO DE GARANHUNS

Abstract

Looking for analyze to large form the radiation data the last 10 years to Garanhuns city, the time series models classified like more indicated was adjusted, not just the nature for presented information was consider in this study like too the particular interesting in predit the insolation levels for the nexts years in the local, being most loyal possible to presented data shape. To select the most adequate model the purpose the work, we realize tests and comparative analyses for to search which model presented better adjust e consequently the better capacity to realize prevision to insolation levels in the stipulated period in Garanhuns. In the research was utilize the daily data to radiation levels in the city for the analyze cycle and information to others variables like air humidity, temperature, wind characteristics (direction, gust and speed), among others, available by Instituto Nacional de Meteorologia (INMET). For comparative analyze between models, was utilize the tests Dicker-Fuller Increased,Shapiro-Wilk, Kruskal-Wallis, looking verify the stationary, normality and seasonality, respectively, for models in study. Too, the models types compared ARIMA and ARIMAX the similar origin, this reforce the importance to accurate application to tests and comparative analyses. Was constacted the better model applied is the ARIMAX, that identify the cycles of growth and decrease in the levels of radiation during the study period. For forecast, for the next 10 years, it is expected that even with the occurrence of periodic behavior, the variability should increase. Lastly, conclud that use of the ARIMAX model contributes to the prevention diseases and for more efficient and intelligent use of solar energy in Garanhuns.