A New Daily Electrical Load Forecasting Model Based on Pattern Recognition and Forecasting Abilities of Self-Organizing-Map ANN A Case Study on Iran Power Network

In this study, a novel intelligent approach is developed for short term load forecasting (STLF). The proposed model consists of two basic Kohonen Artificial Neural Network (ANN) modules. Thefirst module called Load SOM Model is an initial load forecaster without considering temperature effect and the second module called Temperature SOM Model that modifies initial forecasted load by applying temperature effect. The proposed model can be extended sothat be sensitive to all of the other atmospheric factors such as wind speed, moisture and cloud covering. In addition it is able to forecastnormal and abnormal days of year such as holidays, ceremonies, religious and etc, with high accuracy. Ten sub-models are consideredfor forecasting each day of week, special holidays, the days before special holidays and the days after special holidays .The proposedmodel , is tested extensively with the temperature and load curves ofIran electrical power system of 1994 to 2004 using different day types from different times of the year and promising results are obtained with approximately 1.5% of mean error for the days distributed throughout the year.This case study that is performed on load historical data of Iran power system demonstrate the accuracy of the proposed method and show that the forecasted model is simple with highly accuracy.

Keywords:

load forecasting , artificial neural networks , pattern recognition , self organization map (SOM) , Kohonen neural network