Comparación entre los métodos de Promedio Móvil Integrado Autorregresivo (ARIMA) y Tylor Kriging (TK) para la predicción de series temporales de velocidad del viento, caso de estudio: Estación Meteorológica Chimborazo

Abstract

The meteorological station located in the province of Chimborazo, Riobamba canton , parish of Loma de Rasutambo, Pinanquil area, is a place where the different variables that affect the state of the atmosphere can be measured , in addition the observation of atmospheric phenomena, due to the limited infrastructure this meteorological station does not have current data to be able to project the possible installation of a wind turbine for wind power generation, therefore, the objective of this research is to compare the a utoregressive integrated moving average (ARIMA) and Tylor Kriging(TK) methods for the prediction of wind speed time series. The methodology implemented had a non experimental design of epistemological type, with a quantitative approach , obtaining the data matrix from a secondary source, the study population is the information of the years (2014, 2015, 2016) of the wind speed measured in each hour, with the use of an anemometer. The per formance of the autoregressive integrated moving average (ARIMA) and Tylor kriging (TK) methods were compared with the measures of mean absolute error (MAE) and root mean square error (RMSE), resulting in the ARIMA model an MAE of 26.42 % and RMSE of 30.23 % 30.23 %. On the other hand, the Taylor Kriging model had an MAE of 50.09 % and an RMSE of 60.143 % based on these metrics, the closer these values are to zero the better the performance of the model concluding that the ARIMA model (1, 1, 3) is the best in terms of accuracy of the predictions of the wind speed variable, since it presents a lower MAE and RMSE value compared to the Taylor Kriging model.