Building forecasting Markov models with Self-Organizing Maps

This work presents a new methodology for short term loading forecasting whereas the influence of climatic variables (temperature, relative humidity of the air and wind speed) in the consumption behavior of an electrical power distribution system. The proposed methodology involves creating a discrete probability model (Markov chain) from the classification of historical data in a Self-Organizing Map (SOM). It is therefore possible to estimate the probability of a certain demand level to happen given a current climatic condition, as well as the number of time intervals (hours) until this happens. In addition, the Self-Organizing Maps allows the knowledge extraction, i.e. the search for relationships between the variables involved in the problem.