Effective model for next day load curve forecasting based upon combination of perceptron and kohonen ANNs applied to iran power network

A novel hybrid neural network model to the problem of short-term load forecasting (STLF) is proposed in this paper. The electric load is strongly related to metrological conditions and forecast models depend on climatic studies. The most used variable is the air temperature, because there is a close relation between thermal state of well being and the corresponding load (air-conditional apparatus for instance). The proposed general model in this paper is made up of two modules: module 1 is a self-organizing map (SOM) load model that is able to forecast normal and abnormal days load of year such as holidays, ceremonies, religious and etc without considering of climate conditions. Also module 2 is a multi-layer perceptron (MLP) thermal model that make load model sensitive to atmospheric factors such as temperature. General model is able to forecast load in each day of week, special holidays, the days before special holidays and the days after special holidays. Both the SOM and the MLP models are trained and assessed on Iran load and temperature data extracted from Iran National Dispatching Center. MAD for days at years of 2002, 2003 and 2004 is 1.1%, 1.35% and 1.20%. Final results prove that this model can be applied to the prediction of Iran load in real case with high accuracy.