Review of electric load forecasting methods

Abstract

The different available load forecasting techniques can be classified according to the type of load data pattern, into three classes: (1) stationary; (2) nonstationary; and (3) nonstationary, seasonal and cyclical techniques. The criterion is divided into three stages: (1) identification; (2) estimation; and (3) diagnostic checking. The purpose of the identification stage is to ascertain the techniques(s) that appear to hold more promise for adequately describing a given data set. The paper presents an algorithmic procedure for selecting and constructing the most appropriate electric load forecasting model. From this review, it is clear that models for the selection and construction criteria are essential for a proper forecasting model.