基于数据融合的中长期概率性负荷预测方法研究

Research on medium and long-term probabilistic load forecasting method based on data fusion

中文关键词: 中长期负荷预测  细粒度  数据融合  概率性预测

英文关键词: medium- and long-term load forecasting  fine-grained  data fusion  probabilistic forecasting

中文摘要:

      月度负荷预测是电力系统的中长期运行和营销工作开展的基础，考虑到不断变化的政策、环境、高比例可再生能源等因素，中长期负荷具有不确定性，月度概率性电力负荷预测对中长期不确定性的刻画将更利于新型电力系统的快速发展。传统的中长期负荷预测仅依赖月度负荷数据，并未使用更细粒度数据，观测数据有限，无法使用足够的解释变量来捕捉中长期电力负荷的所有突出特征。以系统负荷作为研究对象开展中长期概率性预测方法研究，结合多元线性回归模型和“自下而上”的时间层级协调策略，提出了基于细粒度数据融合的中长期概率性预测方法，并通过实际数据，验证了方法的有效性。

英文摘要:

      Monthly load forecasting is the basis for the medium- and long-term operation of the power system and the development of marketing efforts. Considering the changing policies, envi-ronment, high proportion of renewable energy and other factors, the medium- and long-term loads are uncertain, and the monthly probabilistic power load forecasting's portrayal of the medium- and long-term uncertainty will be more favorable for the rapid development of new power systems. Traditional medium- and long-term load forecasts rely only on monthly load data without using finer-grained data, with limited observations that do not allow the use of sufficient explanatory variables to capture all the salient features of medium- and long-term power loads. A medium- and long-term probabilistic forecasting method is proposed based on the fusion of fine-grained data by taking the system load as the research object and combining the multiple linear regression model and the "bottom-up" temporal hierar-chical coordination strategy, and the validity of the method is verified by the actual data.