摘要
运用多种预测方法对中长期电力负荷预测所得结果会相差甚远,而综合各方法的组合预测能够避免其偏颇。由于在小样本和非线性拟合能力方面的优势,支持向量机方法被用于组合预测:多种传统方法预测值作为输入,拟合输入与输出之间的非线性关系,求得预测结果。针对SVM在处理回归问题时算法编程及参数寻优较为复杂的问题,提出了一种基于SVM图形用户界面(Graphical User Interface,GUI)工具箱的组合预测方法。算例分析表明,运用该方法,在预测过程中可直观、方便地应用通用软件工具包,且预测精度较高,便于推广和工程应用。
The long-term power load forecasting results obtailled by a variety of forecasting methods are a far cry. The combined forecasting approach synthesizes various methods to avoid bias. Due to the advantages of the small sample and non-linear fitting ability, support vector machine （SVM） method is used for combining forecasts. The predictive values of a variety of methods are employed as input to fit non-linear relationship between input and out- put and obtain the predicted results. To reduce the complexity of algorithm programming and parameter optimization by using SVM for regression problem, a combination forecasting method based on SVM graphical user interface tool- kit is proposed. The example shows that the application of general-purpose software toolkit in the forecasting process is intuitive and convenient, predicted result is accuracy, and the proposed method is convenient for engineering ap- plications.
出处
《电力科学与工程》
2013年第12期18-22,共5页
Electric Power Science and Engineering
基金
国家自然科学基金资助项目(51277016)
湖南省高校创新平台开放基金项目(12K074)
关键词
电力负荷
组合预测
支持向量机
图形用户界面
electric load
combination forecasting
support vector machine
graphical user interface
