针对短期负荷预测的特点，通过各因子与负荷数据的相关性分析来评价各因子对于负荷预测的影响，从中选取最合适的影响因素作为训练样本。并对气温的积累效应进行了深入的研究，引入了错位样本概念，将其加入到训练样本中提高了预测精度。(In view of the characteristics of short-term load forecasting, the influence of each factor on load forecasting is evaluated through the correlation analysis of various factors and load data, and the most suitable influencing factors are selected as training samples. In addition, the concept of misplaced samples was introduced into the training samples to improve the prediction accuracy.)