A NEURAL-NETWORK-BASED TECHNIQUE FOR SHORT-TERM FORECASTING OF ANOMALOUS LOAD PERIODS

The paper illustrates a part of the research activity conducted by aut hors in the field of electric Short Term Load Forecasting (STLF) based on Artificial Neural Network (ANN) architectures. Previous experience s with basic ANN architectures have shown that, even though these arch itectures provide results comparable with those obtained by human oper ators for most normal days, they evidence some accuracy deficiencies w hen applied to ''anomalous'' load conditions occurring during holidays and long weekends. For these periods a specific procedure based upon a combined (unsupervised/supervised) approach has been proposed. The u nsupervised stage provides a preventive classification of the historic al load data by means of a Kohonen's Self Organizing Map (SOM) The sup ervised stage, performing the proper forecasting activity, is obtained by using a multi-layer perceptron with a backpropagation learning alg orithm similar to the ones above mentioned. The unconventional use of information deriving from the classification stage permits the propose d procedure to obtain a relevant enhancement of the forecast accuracy for anomalous load situations.