计及模型鲁棒性的有源配电网动态等值建模方法研究

A dynamic equivalent modeling method for active distribution network considering model robustness

在分析有源配电网与大电网相互作用时,准确的有源配电网模型至关重要。由于有源配电网内部可再生能源具有随机性和时变性,基于有源配电网某些特定运行点处系统动态特征所建立的传统等值模型,存在鲁棒能力差的问题。针对这个问题,文章提出了一种计及模型鲁棒性的有源配电网动态等值建模方法。首先建立能够表征有源配电网不同运行状态的系统特征数据库,并采用two-step聚类法和Fisher判别方法对有源配电网运行状态进行分类;然后,基于关键参数辨识方法消除参数辨识过程中的多解问题;最后,通过Elman神经网络,获得能够适应于系统不同运行状态的等值模型鲁棒性参数解。通过仿真算例验证了该方法的有效性。

Accurate model of Active Distribution Network(ADN) is crucial to analyze the interactions between the ADN and large-scale power system. However, due to the stochastic and time-varying nature of renewable resources, traditional equivalent model based on the dynamic responses of ADN under the specific operation condition might not robust enough. To overcome the limitations, this paper developed a robust-improved method for dynamic equivalent modeling of ADN. Firstly, the system feature database which can well represent the difference of ADN operation conditions is established. Besides, the operation conditions of ADN are grouped by twostep clustering method and Fisher discriminant analysis. Secondly, with the key parameter based identification technique applied, the multiple solution issue in parameter identification process could be avoided. Finally, Elman neural network is used to derive the robust sets of parameters are able to help the developed equivalent model adaptable for the wide range of distinct ADN operation conditions. The effectiveness of the proposed method is verified by the case study in the paper.