Short term load forecasting using genetically optimized neural network cascaded with a Modified Kohonen clustering process

Erkmen, İsmet
Ozdogan, A
In this study, a new intelligent approach is developed for short term load forecasting (STLF). The technique consists of three basic modules. The first module employ the clustering of daily load curves using Modified Kohonen algorithm (MKA). Second module determine the most appropriate supervised neural network topology and associated initial weight values for each cluster, extracted from the historical data base, by using genetic algorithm (GA). At the third module, genetically optimized three layered back propagation (BP) network is trained and run to perform hourly load forecasting. Effects of each module on the forecasting accuracy are considered separately. The proposed system is tested extensively with the load curves of Turkish electrical power system of 1993 using different day types from different times of the year and promising results are obtained with approximately 1 % of mean error for the days distributed throughout the year.