Modelación y pronóstico del número de abonados y el consumo de energía eléctrica en MWH de la Empresa Eléctrica Riobamba S.A.

Abstract

Modeling and prediction of the number of customers, energy consumption was carried out in Megawatt - hour (MWh) depending on voltage level categories and prepared by providing the database of the Planning Department in Empresa Eléctrica Riobamba S.A (EERSA), with the aim of obtaining forecast significant models for future knowledge of energy consumption and number of customers. It started with a monthly database with five variables of energy consumption and three variables of the number of customers with 48 data respectively, using the Exploratory Data Analysis of the information, it was identified the presence of outliers that were released, normality test was applied to obtain a variable that did not come from a normal distribution, being transformed by natural logarithm. Graphs of time series and autocorrelations identified the presence of stationarily, trend and cyclicality using the measurements of MAPE, MAD and MSE position obtaining the best prediction model. The obtained result was: a Holt model for the variables, number of customers in general category in low voltage level, number of customers in general category in medium voltage level, number of customers in residential category and energy consumption, MWh residential category; one quadratic model for energy consumption variables category MWh, lighting category in MWh energy consumption, general category at low voltage level and MWh energy, general category at médium voltage level; and a Brown model for the variable of MWh energy consumption in the general category of high voltage level. Forecasting models established were varied with the actual data in 2013 by the autocorrelation function of waste, indicating randomness which verifies that forecast models are right. The EERSA should institute an appropriate format for data storage, so that consumption records and the number of customers are presented in an orderly and understandable way to facilitate a future statistical and economic analysis.