基于数据驱动理念的电力日负荷曲线预测方法研究

针对短期日负荷预测的精度问题,本文提出一种基于数据驱动理念的电力负荷预测方法。在建立预测模型前对所给数据采取一定的预处理:首先提取所收集的海量数据的负荷特征,对负荷特征进行分析,然后进行负荷数据与影响负荷值的因素之间的相关性分析,以此确定对负荷影响较密切的因素,随后建立分类器得到各主要影响因素与各负荷类别之间的关系为后续预测模型奠定基础。对预处理后得到的不同类型的负荷数据采用最小二乘支持向量机方法建立不同的负荷预测模型。以南方某发达城市2008年的负荷数据作为算例验证数据,将本文所提负荷预测方法所得结果与未经数据预处理的负荷预测方法所得结果进行比较,结果表明本文提出的方法得到的预测结果精度较传统方法提高约6%。

Study on Daily Power Load Curve Forecasting Method Based on Data-driven Concept

To improve the accuracy of short-term daily load forecasting,a power load forecasting method based on the data-driven concept is proposed in this paper.Before a forecasting model is established,some pre-processing proce⁃dure is applied to the given data.First,the load characteristics of collected massive data are extracted and analyzed.Then,the correlation analysis method is used to analyze the relationship between load data and the corresponding influ⁃encing factors,thus determining the factors that have a close relationship with the load data.Afterwards,the coupling rules between the main influencing factors and load types are obtained by constructing a classifier,providing a basis for the subsequent forecasting model.At last,the least squares support vector machine method is used to establish different load forecasting models for different types of load data that have been pre-processed.The load data of one developed city in south China in 2008 are taken as an example.From the comparison of the result obtained using the proposed load forecasting method with that using the traditional method(without data pre-processing),it is indicated that the novel method can improve the accuracy of forecasting result by about 6%.