基于统计学分类的综合负荷模型实测建模

Load Modeling with Composite Load Model Based on Statistical Classification

电力系统负荷模型参数的简化是负荷模型实用化的关键,基于所辨识的负荷模型参数进行负荷统计学特性分析对于建立具有强泛化能力的负荷模型具有重要意义。以综合负荷模型为研究对象,采用标准阶跃响应为特征向量,用统计学中的系统聚类法并结合方差分析指标对负荷数据加以筛选和分类,并对分类后的结果进行观察,得到了一些参数和时间上的特征,揭示了负荷数据内部所存在的部分统计学规律。进而应用支持向量机工具对分类后的负荷建立了具有良好泛化能力的负荷模型。归纳出了负荷数据由筛选、分类到建立模型的一条思路。

A load model with the simplest structure and the least parameters to be identified is very important for power system analysis and control. Since the load is always changing, it＇s necessary to find out the statistic laws from identified model parameters. The responses of the load model are chosen to the standard step function input as the characteristic vectors, and the aggregation theory of the statistic analysis is applied to classify the load models. Analysis results show the statistic law of the load behind the seemingly random variation of the data. The Support Vector Machine （SVM） tool is applied to build the load model with good generalization capability based on the classification. A systematic way is shown in building the measurement-based load model, which includes the selection of the data, model classification and building the load model based on the classification.