基于模型激励响应的负荷分类及泛化能力

Classification and Generalization of the Load Model Based on Model Dynamic Responses

负荷时变性已经成为制约负荷模型推广应用的主要因素,而负荷分类则是解决这个问题的很好的办法。本文提出了一种新的负荷分类及模型泛化的方法:首先以模型阶跃响应为特征矢量,采用聚类分析的方法对负荷进行分类,然后用多曲线拟合的方法对每一类建立一个代表整个类的负荷模型,并对各类模型的参数以及各类扰动数据之间存在的规律进行分析,在此基础上按照一定的规律合并类,最终得到了具有较强泛化能力和实用性的负荷模型。

The time-variable characteristic of the load is a limiting factor in the using of the load model. Load classification is a good approach to solve this problem. A new load model classification and generalization method is proposed in this paper. First, load model responses are chosen as the characteristic vectors in the cluster analysis. Then, the multi-curve identification technique is applied in building the load models for each class. Finally, more robust and practical models are obtained after the combination of classes, which are also verified by their strong generalization capability and physical interpretation.