模型结构对负荷模型泛化能力的影响

Model Structure Effect on Generalization Capability of Load Model

分析了基于实测的考虑时变性的自适应(TVA)负荷模型和考虑配电网支路的配电网综合负荷模型(SLM),并利用现场的实测数据对2种模型进行了回响测试,以比较2种模型在仿真计算中的准确性。在分别考虑辨识配电网阻抗与定配电网阻抗2种情况下,利用实测数据对SLM的参数进行了辨识与测试,进而推出了使用基于全网数据辨识参数的SLM计算结果偏于保守的原因,并以吉林电网为算例进行了模型泛化能力的比较。最终得出结论:对于吉林这类通道型电网,这2种模型的泛化能力与其模型结构关系较小,与模型参数关系较大。

Two load models are analyzed, one is time variant adapting （TVA） load model based on measured data from spot, the other is synthetic load model （SLM） considering distribution spur track. Resound tests on the two load models are conducted to compare the accuracy of the two models. The parameters of SLM are identified in the cases of certain and uncertain distribution spur track impedance, and why the SLM based on the whole network identification being conservative is found. Simulations on Jilin Grid indicate that the generalization capability of SLM based on node data identification is equivalent to that of TVA model, and it is concluded that the generalization capabilities of the two types of load models are less dependent on the model structure than on their parameters for a power system like Jilin Grid.