Research And Realization Of Power System Short-Term Load Forecasting Based On Artifical Intelligence

After introducing the importance and the methods of the load forecasting in electric power markets, this present thesis analyzes the factors which effect the precision of the load forecasting, studies the characteristics of the load of the holidays, and proposes short-term load forecasting model for predicting eveiy quarter minus electric power loads of 24 hours ahead for ShenYang. The factors are divided into computerizable values and non-computerizable values. The computerizable values are predicted by mathematics methods, which are divided into Spring model, Summer model, Autumn model and Winter model in Quarter. In date type load forecasting model is divided into two types: i) the workday model that is described as the sum of a basic load and a modi1~ing load, ii) the holiday model that is described as the weighted average of the history holiday load. At Last we will modify the forecasted result with Expert System .The effectiveness of the predicting is demonstrated by the data from the Shen Yang, A load forecasting software has been developed for Shen Yang.