Simple model for load forecast weather sensitive

This work purpose is to establish an easy usage model and results exact enough for the daily peak electric load forecast. This model is developed for Edenor´s service area (Argentina) and became function of maximum temperature predicted for the day. A clustering algorithm is applied to maximum daily temperature readings to divide the historical database into seasonal classes. These classes of data are divided into subclasses depending on type of day, which are used to eliminate load growth into the different considered years, to standardize them to the last year value. A load-temperature curve is obtained doing a regression analysis over each seasonal subclass and integrated to the complete data spectrum. From this curve, for a particular day knowing the maximum temperature from meteorological forecast it can be possible to predict daily peak electric load. This can be done with enough accurate for the operation and planning purposes