An Accurate Model for Short-Term Load Forecasting

This paper presents a simple algorithm for precise short-term load forecasting. This forecast is essentially for monitoring and controlling power system operation. An adaptive model based on records of the hourly loads and weather informations is introduced. Detection for the bad data is made. The daily and weekly correlations are considered. An adaptive error estimation ( mean and variance ) is used to have the expected loads on a probabilistic base. The proposed model has the advantage of low storage requirements and quick calculations easily implemented in on-line computer system. Application of the proposed algorithm on real data banks is made and a reasonable accuracy is obtained.