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

A new intelligent approach is developed for short-term load forecasting (STLF). The technique consists of three basic modules. The first module employs the clustering of daily load curves using a modified Kohonen algorithm (MKA). The second module determines the most appropriate supervised neural network topology and associated initial weight values for each cluster extracted from a historical database, by using a genetic algorithm (GA). In the third module, a genetically optimized three-layered backpropagation (BP) network is trained and run to perform hourly load forecasting. The effects of each module on the forecasting accuracy are considered separately. The proposed system is tested extensively with the load curves of the Turkish electrical power system in 1993 using different day types from different times of the year, and promising results are obtained with approximately 1% mean error for days distributed throughout the year.