本文对月度电力消费数据进行全假日调整,并对气象数据按电力消费权重加权处理,弥补了国内此类模型中考虑季节和周期性因素为主,气象数据处理简单的不足,并应用了分段线性化与季节ARlMA模型相结合的机制,有效提升了我国全社会电力消费预测的精确性,并经过样本外预测检验证实了该模型结构的稳定性。最后,在设定不同气象务件的情景下,使用该模型预测了相应的电力消费。 In this paper, the monthly power consumption data are adjusted on a full-holiday basis and the meteorological data are weighted according to the power consumption weights, which make up for the lack of seasonal and periodic factors and simple meteorological data processing in the domestic models. The linearization combined with the seasonal ARlMA model effectively improves the accuracy of the power consumption forecast of the whole society in our country and confirms the stability of the model structure through the out-of-sample prediction. Finally, using this model to predict the corresponding electricity consumption in the setting of different meteorological services.