A NOVEL PROBABILISTIC SHORT-TERM LOAD FORECASTING METHOD FOR LARGE POWER GRID

Short-term load forecasting plays an important role in power system operation. If load and its deviation distribution could be accurately predicted, it's very helpful for power system reliability and economical efficiency. According predicting outcomes, percentage reserve would be optimized, and it's helpful for energy-saving and emission-reduction. In this paper, a novel method is proposed to predict load and its deviation for large power grid. Firstly, the large power grid is divided into some subnets. Then, the load of each subnet is predicted. Thirdly, the deviation distribution of each subnet is evaluated through historical data. Finally, the load of large power grid is predicted, and the deviation distribution is calculated. According to practical application in some provinces of China, it's proved that the new method is a simple method with high and stable accuracy. In China, the power consumption balance calculation is carried out by provinces, which are served by large power grids. Therefore, the research findings in this paper can improve the optimization level directly.