摘要
针对传统考虑气温因素的短期负荷预测模型存在预测精度不高的问题,提出了一种考虑负荷和气温周期特征的短期负荷预测模型。首先对气温及负荷序列采用VMD分解得到两组特征互异的分量,取其中与原始负荷相关性最大的分量分别作为气温特征和负荷特征。然后,将历史负荷特征结合所取的气温特征和负荷特征一起输入GRU模型进行预测。算例分析表明,该模型的平均绝对百分误差为0.765%,验证了所提方法在电力负荷预测的有效性。
The traditional short-term load forecasting model that takes into account the temperature factor has the problem of low forecast accuracy.A short-term load forecasting model considering the periodic characteristics of load and temperature is proposed.First,the temperature series and the load series are decomposed by VMD to obtain two sets of components with different characteristics,and the components with the greatest correlation with the original load are used as the temperature feature and the load feature respectively.Then,the historical load features combined with the temperature feature and load feature taken are input into the GRU model for prediction.The analysis of calculation examples shows that the mean absolute percentage error of the model in this paper is 0.765%,which verifies the effectiveness of proposed method in power load forecasting.
作者
黄冬梅
唐振
胡安铎
孙锦中
HUANG Dongmei;TANG Zhen;HU Anduo;SUN Jinzhong(College of Electronics and Information Engineering,Shanghai University of Electric Power,Shanghai 200090,China)
出处
《电工技术》
2021年第8期146-149,共4页
Electric Engineering
基金
上海市科委地方院校能力建设项目(编号20020500700)。
关键词
短期负荷预测
气温分解
特征选择
门控循环单元
short-term load forecasting
temperature decomposition
feature select
gated recurrent unit
