A combination method for short term load forecasting used in Iran electricity market by NeuroFuzzy, Bayesian and finding similar days methods

Short term load forecasting (STLF) plays an important role for the power system operational planners and also most of the participants in the nowadays electricity markets. With the importance of the STLF in power system operation and electricity markets, many methods for arriving careful results, are represented. In this paper, a combination approach for STLF is proposed. The proposed approach is based on the weighted method for STLF results from Bayesian neural network, neurofuzzy and finding similar days methods. According to the obtained research, these 3 mentioned methods have the best results for the STLF of Iran national power system. Because Iran calendar is a combination of two solar and lunar calendars, so the special conditions, such as: solar and lunar holidays, days after or between holidays have the variable results with these 3 methods. For arriving STLF careful results, the least square method is used for combining these 3 methods. By using this technique, the effect of improper results is ignored. The results for Iran power system, shows that the idea can improve the performance of the STLF.