A Power Load Forecasting Method Based on Matching Coefficient of Meteorological Factors and Similar Load Modification

It is of great significance to enhance the precision of power load forecasting for the security and stability of power system. According to the power load characteristics of both regularity and randomness, a novel load forecasting method based on the integration of similar load and recent historical load is proposed. In terms of similar load, this algorithm, firstly, utilizes matching coefficient of meteorological factors to quantify the similarity, then uses membership of day-type for the modification of similarity. After the process of filtering, the input sequence is made up of the combination of modified similar load and recent historical load. The results of simulation indicate that, compared with these forecasting algorithms merely based on similar load or on recent historical load, both mean relative error and mean square deviation of the algorithm presented in this paper are smaller and the precision of load prediction is higher.