A holiday short term load forecasting considering weather information

Load forecast for holidays is always hard to be processed, due to the dissimilar load behaviors compared with normal weekdays and to the insufficient samples for normal algorithm. The load during holiday is mainly influenced by the long time load increase and weather condition. The proposed method uses a hybrid method of similar days to obtain scaled load curve and fuzzy inference method to forecast the holiday load level. Weather information and load annual increase are considered in fuzzy inference to eliminate their influence. The proposed method is implemented in real-time EMS with actual load data. The test results show good accuracy especially in weather change days