The relevance analysis between electrical day peak load and meteorological index based on wavelet denoising and SVM

The variability of electrical load is affected by many meteorological factors, especially in summer and winter, their relations are closer. The paper introduces four meteorological indexes to quantify the synthesis influences of temperature, humility and wind speed to load. First, the paper uses the wavelet denoising theory to eliminate the error data and noises included in the load data and the meteorological data. Then, some measures are taken to strip some other influence factors from load, such as holidays, load natural growth, and thermal incubation effect etc. Base on the above, grey correlation theory is used to analyze the relevance between load and each meteorological index, selecting the one which has the best relevance to load to carry on the sensitivity analysis and the electrical daily peak load forecast using SVM(support vector machine). The rules about how peak load is affected by meteorological factors can provide the related power departments some useful reference information.